entry_point
stringlengths 1
65
| original_triton_code
stringlengths 4.5k
619k
| python_code
stringlengths 208
60.9k
| triton_code
stringlengths 1.15k
275k
| repo_name
stringlengths 7
115
| module_name
stringlengths 1
65
| synthetic
bool 1
class | uuid
int64 0
18.5k
| licenses
listlengths 1
6
| stars
int64 0
19.8k
| sha
stringlengths 40
40
| repo_link
stringlengths 72
180
|
|---|---|---|---|---|---|---|---|---|---|---|---|
SingleHead
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/td/ctdybbibnws4d7ukbk3fpn35zkgapxylowdhzwx7vgsllncbdrxa.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# x => convolution
# x_1 => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_0 = async_compile.triton('triton_poi_fused_convolution_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/32/c32v7egt4mupqssam3gmac2qgv3ujprjybthsgweflmot256qqw7.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_2 => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4, 4), (64, 16, 4, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_0.run(buf1, primals_2, 256, grid=grid(256), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4, 4), (64, 16, 4, 1))
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf3, primals_5, 256, grid=grid(256), stream=stream0)
del primals_5
return (buf3, primals_1, primals_3, primals_4, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.utils.data
from itertools import product as product
from math import sqrt as sqrt
class SingleHead(nn.Module):
"""
Single head used in CenterNet Head.
"""
def __init__(self, in_channel, out_channel, bias_fill=False, bias_value=0):
super(SingleHead, self).__init__()
self.feat_conv = nn.Conv2d(in_channel, in_channel, kernel_size=3,
padding=1)
self.relu = nn.ReLU()
self.out_conv = nn.Conv2d(in_channel, out_channel, kernel_size=1)
if bias_fill:
self.out_conv.bias.data.fill_(bias_value)
def forward(self, x):
x = self.feat_conv(x)
x = self.relu(x)
x = self.out_conv(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channel': 4, 'out_channel': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.utils.data
from itertools import product as product
from math import sqrt as sqrt
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4, 4), (64, 16, 4, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_relu_0[grid(256)](buf1, primals_2, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4, 4), (64, 16, 4, 1))
buf3 = buf2
del buf2
triton_poi_fused_convolution_1[grid(256)](buf3, primals_5, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
return buf3, primals_1, primals_3, primals_4, buf1
class SingleHeadNew(nn.Module):
"""
Single head used in CenterNet Head.
"""
def __init__(self, in_channel, out_channel, bias_fill=False, bias_value=0):
super(SingleHeadNew, self).__init__()
self.feat_conv = nn.Conv2d(in_channel, in_channel, kernel_size=3,
padding=1)
self.relu = nn.ReLU()
self.out_conv = nn.Conv2d(in_channel, out_channel, kernel_size=1)
if bias_fill:
self.out_conv.bias.data.fill_(bias_value)
def forward(self, input_0):
primals_1 = self.feat_conv.weight
primals_2 = self.feat_conv.bias
primals_4 = self.out_conv.weight
primals_5 = self.out_conv.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
StevenGrove/DynamicHead
|
SingleHead
| false | 14,441 |
[
"Apache-2.0"
] | 69 |
d62aa84e1d1c6a0c74d46258ad77b11413c10bef
|
https://github.com/StevenGrove/DynamicHead/tree/d62aa84e1d1c6a0c74d46258ad77b11413c10bef
|
ConvLSTMCell
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ie/ciettq2a3562jfpgfe75iig4ki2hbm6pmbwujlvp6mw26i2odufm.py
# Topologically Sorted Source Nodes: [combined], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# combined => cat
# Graph fragment:
# %cat : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%primals_1, %primals_2], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 16) % 8
x0 = xindex % 16
x2 = (xindex // 128)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (16*x1) + (64*x2)), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr1 + (x0 + (16*((-4) + x1)) + (64*x2)), tmp6 & xmask, other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + (x3), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/oz/coz7e2n7woknuf3o2i4jra3wwj77ytmyhidqvpzncrl4dxljsxz6.py
# Topologically Sorted Source Nodes: [i, f, o, g, mul, mul_1, new_c, tanh_1, new_h], Original ATen: [aten.sigmoid, aten.tanh, aten.mul, aten.add, aten.sigmoid_backward]
# Source node to ATen node mapping:
# f => sigmoid_1
# g => tanh
# i => sigmoid
# mul => mul
# mul_1 => mul_1
# new_c => add
# new_h => mul_2
# o => sigmoid_2
# tanh_1 => tanh_1
# Graph fragment:
# %sigmoid : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%getitem,), kwargs = {})
# %sigmoid_1 : [num_users=3] = call_function[target=torch.ops.aten.sigmoid.default](args = (%getitem_1,), kwargs = {})
# %sigmoid_2 : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%getitem_2,), kwargs = {})
# %tanh : [num_users=2] = call_function[target=torch.ops.aten.tanh.default](args = (%getitem_3,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid_1, %primals_5), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid, %tanh), kwargs = {})
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
# %tanh_1 : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%add,), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid_2, %tanh_1), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %sigmoid_1), kwargs = {})
# %mul_14 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid_1, %sub_3), kwargs = {})
triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_1 = async_compile.triton('triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: '*fp32', 8: '*fp32', 9: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 9, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0, out_ptr1, out_ptr2, out_ptr3, out_ptr4, out_ptr5, xnumel, XBLOCK : tl.constexpr):
xnumel = 144
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = (xindex // 36)
x4 = xindex % 36
x1 = (xindex // 9) % 4
x3 = xindex
tmp0 = tl.load(in_ptr0 + (72 + x4 + (144*x2)), xmask)
tmp1 = tl.load(in_ptr1 + (8 + x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (x4 + (144*x2)), xmask)
tmp5 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (108 + x4 + (144*x2)), xmask)
tmp9 = tl.load(in_ptr1 + (12 + x1), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (36 + x4 + (144*x2)), xmask)
tmp13 = tl.load(in_ptr1 + (4 + x1), xmask, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr2 + (x3), xmask)
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tmp6 = tmp4 + tmp5
tmp7 = tl.sigmoid(tmp6)
tmp10 = tmp8 + tmp9
tmp11 = libdevice.tanh(tmp10)
tmp14 = tmp12 + tmp13
tmp15 = tl.sigmoid(tmp14)
tmp17 = tmp15 * tmp16
tmp18 = tmp7 * tmp11
tmp19 = tmp17 + tmp18
tmp20 = 1.0
tmp21 = tmp20 - tmp15
tmp22 = tmp15 * tmp21
tmp23 = libdevice.tanh(tmp19)
tmp24 = tmp3 * tmp23
tl.store(out_ptr0 + (x3), tmp3, xmask)
tl.store(out_ptr1 + (x3), tmp7, xmask)
tl.store(out_ptr2 + (x3), tmp11, xmask)
tl.store(out_ptr3 + (x3), tmp19, xmask)
tl.store(out_ptr4 + (x3), tmp22, xmask)
tl.store(out_ptr5 + (x3), tmp24, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (16, 8, 4, 4), (128, 16, 4, 1))
assert_size_stride(primals_4, (16, ), (1, ))
assert_size_stride(primals_5, (4, 4, 3, 3), (36, 9, 3, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8, 4, 4), (128, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [combined], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(primals_1, primals_2, buf0, 512, grid=grid(512), stream=stream0)
del primals_1
del primals_2
# Topologically Sorted Source Nodes: [A], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_3, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 16, 3, 3), (144, 9, 3, 1))
buf3 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf2 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf4 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf5 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf7 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf6 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
# Topologically Sorted Source Nodes: [i, f, o, g, mul, mul_1, new_c, tanh_1, new_h], Original ATen: [aten.sigmoid, aten.tanh, aten.mul, aten.add, aten.sigmoid_backward]
triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_1.run(buf1, primals_4, primals_5, buf3, buf2, buf4, buf5, buf7, buf6, 144, grid=grid(144), stream=stream0)
del buf1
del primals_4
return (buf6, buf5, buf3, primals_3, primals_5, buf0, buf2, buf3, buf4, buf5, buf7, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((16, 8, 4, 4), (128, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((16, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch.autograd import Variable
import torch.nn as nn
class ConvLSTMCell(nn.Module):
def __init__(self, input_channels, hidden_channels, kernel_size, bias=True
):
super(ConvLSTMCell, self).__init__()
assert hidden_channels % 2 == 0
self.input_channels = input_channels
self.hidden_channels = hidden_channels
self.bias = bias
self.kernel_size = kernel_size
self.num_features = 4
self.padding = (kernel_size - 1) // 2
self.conv = nn.Conv2d(self.input_channels + self.hidden_channels, 4 *
self.hidden_channels, self.kernel_size, 1, self.padding)
self._initialize_weights()
def _initialize_weights(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.normal(m.weight.data, std=0.01)
if m.bias is not None:
m.bias.data.zero_()
def forward(self, input, h, c):
combined = torch.cat((input, h), dim=1)
A = self.conv(combined)
ai, af, ao, ag = torch.split(A, A.size()[1] // self.num_features, dim=1
)
i = torch.sigmoid(ai)
f = torch.sigmoid(af)
o = torch.sigmoid(ao)
g = torch.tanh(ag)
new_c = f * c + i * g
new_h = o * torch.tanh(new_c)
return new_h, new_c, o
@staticmethod
def init_hidden(batch_size, hidden_c, shape):
return Variable(torch.zeros(batch_size, hidden_c, shape[0], shape[1])
), Variable(torch.zeros(batch_size, hidden_c, shape[0], shape[1]))
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 3, 3])]
def get_init_inputs():
return [[], {'input_channels': 4, 'hidden_channels': 4, 'kernel_size': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch.autograd import Variable
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 16 % 8
x0 = xindex % 16
x2 = xindex // 128
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 16 * x1 + 64 * x2), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp9 = tl.load(in_ptr1 + (x0 + 16 * (-4 + x1) + 64 * x2), tmp6 & xmask,
other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + x3, tmp10, xmask)
@triton.jit
def triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_1(in_ptr0,
in_ptr1, in_ptr2, out_ptr0, out_ptr1, out_ptr2, out_ptr3, out_ptr4,
out_ptr5, xnumel, XBLOCK: tl.constexpr):
xnumel = 144
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex // 36
x4 = xindex % 36
x1 = xindex // 9 % 4
x3 = xindex
tmp0 = tl.load(in_ptr0 + (72 + x4 + 144 * x2), xmask)
tmp1 = tl.load(in_ptr1 + (8 + x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (x4 + 144 * x2), xmask)
tmp5 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (108 + x4 + 144 * x2), xmask)
tmp9 = tl.load(in_ptr1 + (12 + x1), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (36 + x4 + 144 * x2), xmask)
tmp13 = tl.load(in_ptr1 + (4 + x1), xmask, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr2 + x3, xmask)
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tmp6 = tmp4 + tmp5
tmp7 = tl.sigmoid(tmp6)
tmp10 = tmp8 + tmp9
tmp11 = libdevice.tanh(tmp10)
tmp14 = tmp12 + tmp13
tmp15 = tl.sigmoid(tmp14)
tmp17 = tmp15 * tmp16
tmp18 = tmp7 * tmp11
tmp19 = tmp17 + tmp18
tmp20 = 1.0
tmp21 = tmp20 - tmp15
tmp22 = tmp15 * tmp21
tmp23 = libdevice.tanh(tmp19)
tmp24 = tmp3 * tmp23
tl.store(out_ptr0 + x3, tmp3, xmask)
tl.store(out_ptr1 + x3, tmp7, xmask)
tl.store(out_ptr2 + x3, tmp11, xmask)
tl.store(out_ptr3 + x3, tmp19, xmask)
tl.store(out_ptr4 + x3, tmp22, xmask)
tl.store(out_ptr5 + x3, tmp24, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (16, 8, 4, 4), (128, 16, 4, 1))
assert_size_stride(primals_4, (16,), (1,))
assert_size_stride(primals_5, (4, 4, 3, 3), (36, 9, 3, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8, 4, 4), (128, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(512)](primals_1, primals_2, buf0, 512,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_1
del primals_2
buf1 = extern_kernels.convolution(buf0, primals_3, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 16, 3, 3), (144, 9, 3, 1))
buf3 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf2 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf4 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf5 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf7 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
buf6 = empty_strided_cuda((4, 4, 3, 3), (36, 9, 3, 1), torch.float32)
triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_1[grid(144)](
buf1, primals_4, primals_5, buf3, buf2, buf4, buf5, buf7, buf6,
144, XBLOCK=256, num_warps=4, num_stages=1)
del buf1
del primals_4
return (buf6, buf5, buf3, primals_3, primals_5, buf0, buf2, buf3, buf4,
buf5, buf7)
class ConvLSTMCellNew(nn.Module):
def __init__(self, input_channels, hidden_channels, kernel_size, bias=True
):
super(ConvLSTMCellNew, self).__init__()
assert hidden_channels % 2 == 0
self.input_channels = input_channels
self.hidden_channels = hidden_channels
self.bias = bias
self.kernel_size = kernel_size
self.num_features = 4
self.padding = (kernel_size - 1) // 2
self.conv = nn.Conv2d(self.input_channels + self.hidden_channels, 4 *
self.hidden_channels, self.kernel_size, 1, self.padding)
self._initialize_weights()
def _initialize_weights(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.normal(m.weight.data, std=0.01)
if m.bias is not None:
m.bias.data.zero_()
@staticmethod
def init_hidden(batch_size, hidden_c, shape):
return Variable(torch.zeros(batch_size, hidden_c, shape[0], shape[1])
), Variable(torch.zeros(batch_size, hidden_c, shape[0], shape[1]))
def forward(self, input_0, input_1, input_2):
primals_3 = self.conv.weight
primals_4 = self.conv.bias
primals_1 = input_0
primals_2 = input_1
primals_5 = input_2
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0], output[1], output[2]
|
Starboy-at-earth/DMRA
|
ConvLSTMCell
| false | 14,442 |
[
"MIT"
] | 84 |
596cc6106ab5f1f03deb60a7f4bb0c2ad1029a83
|
https://github.com/Starboy-at-earth/DMRA/tree/596cc6106ab5f1f03deb60a7f4bb0c2ad1029a83
|
ILN
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/um/cumfxxnuff67646tii7qigo5xu4wp3dwgamqagptzcoe6havnpgp.py
# Topologically Sorted Source Nodes: [ln_mean, ln_var, add_1, sqrt_1], Original ATen: [aten.mean, aten.var, aten.add, aten.sqrt]
# Source node to ATen node mapping:
# add_1 => add_1
# ln_mean => mean_1
# ln_var => var_1
# sqrt_1 => sqrt_1
# Graph fragment:
# %mean_1 : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [1, 2, 3], True), kwargs = {})
# %var_1 : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%primals_1, [1, 2, 3]), kwargs = {correction: 1, keepdim: True})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%var_1, 1e-05), kwargs = {})
# %sqrt_1 : [num_users=2] = call_function[target=torch.ops.aten.sqrt.default](args = (%add_1,), kwargs = {})
triton_per_fused_add_mean_sqrt_var_0 = async_compile.triton('triton_per_fused_add_mean_sqrt_var_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_mean_sqrt_var_0', 'mutated_arg_names': ['in_out_ptr0', 'in_out_ptr1'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 4, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_mean_sqrt_var_0(in_out_ptr0, in_out_ptr1, in_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 64, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 64.0
tmp20 = tmp4 / tmp19
tmp21 = 63.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + (x0), tmp25, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/5i/c5in4qum3v6uczba2ipqijwx2a55yoloa7pndazn53hbcswn6ktk.py
# Topologically Sorted Source Nodes: [in_mean, in_var, sub, add, sqrt, out_in, sub_1, out_ln, mul, sub_2, mul_1, out, mul_2, out_1], Original ATen: [aten.mean, aten.var, aten.sub, aten.add, aten.sqrt, aten.div, aten.mul, aten.rsub]
# Source node to ATen node mapping:
# add => add
# in_mean => mean
# in_var => var
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# out => add_2
# out_1 => add_3
# out_in => div
# out_ln => div_1
# sqrt => sqrt
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# Graph fragment:
# %mean : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [2, 3], True), kwargs = {})
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%primals_1, [2, 3]), kwargs = {correction: 1, keepdim: True})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%var, 1e-05), kwargs = {})
# %sqrt : [num_users=2] = call_function[target=torch.ops.aten.sqrt.default](args = (%add,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %sqrt), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean_1), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_1, %sqrt_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%expand, %div), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %expand), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_2, %div_1), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_2, %expand_2), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_2, %expand_3), kwargs = {})
triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1 = async_compile.triton('triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[16, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: '*fp32', 8: '*fp32', 9: 'i32', 10: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1', 'mutated_arg_names': ['in_out_ptr0', 'in_out_ptr1'], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 4, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1(in_out_ptr0, in_out_ptr1, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 16
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
x2 = xindex % 4
x3 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (r1 + (16*x0)), xmask, other=0.0)
tmp26 = tl.load(in_ptr1 + (x2), xmask, eviction_policy='evict_last')
tmp32 = tl.load(in_ptr2 + (x3), xmask, eviction_policy='evict_last')
tmp34 = tl.load(in_ptr3 + (x3), xmask, eviction_policy='evict_last')
tmp38 = tl.load(in_ptr4 + (x2), xmask, eviction_policy='evict_last')
tmp40 = tl.load(in_ptr5 + (x2), xmask, eviction_policy='evict_last')
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 16, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 16.0
tmp20 = tmp4 / tmp19
tmp21 = 15.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tmp27 = tmp0 - tmp20
tmp28 = tmp27 / tmp25
tmp29 = tmp26 * tmp28
tmp30 = 1.0
tmp31 = tmp30 - tmp26
tmp33 = tmp0 - tmp32
tmp35 = tmp33 / tmp34
tmp36 = tmp31 * tmp35
tmp37 = tmp29 + tmp36
tmp39 = tmp37 * tmp38
tmp41 = tmp39 + tmp40
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + (x0), tmp25, xmask)
tl.store(out_ptr0 + (r1 + (16*x0)), tmp41, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_3, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_4, (1, 4, 1, 1), (4, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf6 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf9 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf7 = reinterpret_tensor(buf6, (4, 1, 1, 1), (1, 1, 1, 1), 0); del buf6 # reuse
buf11 = reinterpret_tensor(buf9, (4, 1, 1, 1), (1, 1, 1, 1), 0); del buf9 # reuse
# Topologically Sorted Source Nodes: [ln_mean, ln_var, add_1, sqrt_1], Original ATen: [aten.mean, aten.var, aten.add, aten.sqrt]
stream0 = get_raw_stream(0)
triton_per_fused_add_mean_sqrt_var_0.run(buf7, buf11, primals_1, 4, 64, grid=grid(4), stream=stream0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf3 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0); del buf0 # reuse
buf5 = reinterpret_tensor(buf3, (4, 4, 1, 1), (4, 1, 1, 1), 0); del buf3 # reuse
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [in_mean, in_var, sub, add, sqrt, out_in, sub_1, out_ln, mul, sub_2, mul_1, out, mul_2, out_1], Original ATen: [aten.mean, aten.var, aten.sub, aten.add, aten.sqrt, aten.div, aten.mul, aten.rsub]
triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1.run(buf1, buf5, primals_1, primals_2, buf7, buf11, primals_3, primals_4, buf12, 16, 16, grid=grid(16), stream=stream0)
del primals_4
return (buf12, primals_1, primals_2, primals_3, buf1, buf5, buf7, buf11, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
from torch.nn.parameter import Parameter
class ILN(nn.Module):
def __init__(self, num_features, eps=1e-05):
super(ILN, self).__init__()
self.eps = eps
self.rho = Parameter(torch.Tensor(1, num_features, 1, 1))
self.gamma = Parameter(torch.Tensor(1, num_features, 1, 1))
self.beta = Parameter(torch.Tensor(1, num_features, 1, 1))
self.rho.data.fill_(0.0)
self.gamma.data.fill_(1.0)
self.beta.data.fill_(0.0)
def forward(self, input):
in_mean, in_var = torch.mean(input, dim=[2, 3], keepdim=True
), torch.var(input, dim=[2, 3], keepdim=True)
out_in = (input - in_mean) / torch.sqrt(in_var + self.eps)
ln_mean, ln_var = torch.mean(input, dim=[1, 2, 3], keepdim=True
), torch.var(input, dim=[1, 2, 3], keepdim=True)
out_ln = (input - ln_mean) / torch.sqrt(ln_var + self.eps)
out = self.rho.expand(input.shape[0], -1, -1, -1) * out_in + (1 -
self.rho.expand(input.shape[0], -1, -1, -1)) * out_ln
out = out * self.gamma.expand(input.shape[0], -1, -1, -1
) + self.beta.expand(input.shape[0], -1, -1, -1)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'num_features': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_add_mean_sqrt_var_0(in_out_ptr0, in_out_ptr1, in_ptr0,
xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 64, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 64.0
tmp20 = tmp4 / tmp19
tmp21 = 63.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + x0, tmp25, xmask)
@triton.jit
def triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1(in_out_ptr0,
in_out_ptr1, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5,
out_ptr0, xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
x2 = xindex % 4
x3 = xindex // 4
tmp0 = tl.load(in_ptr0 + (r1 + 16 * x0), xmask, other=0.0)
tmp26 = tl.load(in_ptr1 + x2, xmask, eviction_policy='evict_last')
tmp32 = tl.load(in_ptr2 + x3, xmask, eviction_policy='evict_last')
tmp34 = tl.load(in_ptr3 + x3, xmask, eviction_policy='evict_last')
tmp38 = tl.load(in_ptr4 + x2, xmask, eviction_policy='evict_last')
tmp40 = tl.load(in_ptr5 + x2, xmask, eviction_policy='evict_last')
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 16, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 16.0
tmp20 = tmp4 / tmp19
tmp21 = 15.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tmp27 = tmp0 - tmp20
tmp28 = tmp27 / tmp25
tmp29 = tmp26 * tmp28
tmp30 = 1.0
tmp31 = tmp30 - tmp26
tmp33 = tmp0 - tmp32
tmp35 = tmp33 / tmp34
tmp36 = tmp31 * tmp35
tmp37 = tmp29 + tmp36
tmp39 = tmp37 * tmp38
tmp41 = tmp39 + tmp40
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + x0, tmp25, xmask)
tl.store(out_ptr0 + (r1 + 16 * x0), tmp41, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_3, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_4, (1, 4, 1, 1), (4, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf6 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf9 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf7 = reinterpret_tensor(buf6, (4, 1, 1, 1), (1, 1, 1, 1), 0)
del buf6
buf11 = reinterpret_tensor(buf9, (4, 1, 1, 1), (1, 1, 1, 1), 0)
del buf9
get_raw_stream(0)
triton_per_fused_add_mean_sqrt_var_0[grid(4)](buf7, buf11,
primals_1, 4, 64, XBLOCK=1, num_warps=2, num_stages=1)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf3 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0)
del buf0
buf5 = reinterpret_tensor(buf3, (4, 4, 1, 1), (4, 1, 1, 1), 0)
del buf3
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1[grid(16)](buf1,
buf5, primals_1, primals_2, buf7, buf11, primals_3, primals_4,
buf12, 16, 16, XBLOCK=1, num_warps=2, num_stages=1)
del primals_4
return buf12, primals_1, primals_2, primals_3, buf1, buf5, buf7, buf11
class ILNNew(nn.Module):
def __init__(self, num_features, eps=1e-05):
super(ILNNew, self).__init__()
self.eps = eps
self.rho = Parameter(torch.Tensor(1, num_features, 1, 1))
self.gamma = Parameter(torch.Tensor(1, num_features, 1, 1))
self.beta = Parameter(torch.Tensor(1, num_features, 1, 1))
self.rho.data.fill_(0.0)
self.gamma.data.fill_(1.0)
self.beta.data.fill_(0.0)
def forward(self, input_0):
primals_2 = self.rho
primals_3 = self.gamma
primals_4 = self.beta
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4])
return output[0]
|
SubZero12556/Cats2dogs_ONNX
|
ILN
| false | 14,443 |
[
"MIT"
] | 2,519 |
52a6a60d519e23b02f0847f0fa9f9ead89ca5f4e
|
https://github.com/SubZero12556/Cats2dogs_ONNX/tree/52a6a60d519e23b02f0847f0fa9f9ead89ca5f4e
|
ResidualBlock
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/fk/cfkcunh3plyysuvib63zgkougyqv2ia22pa4qcifvxy3tij7w7nx.py
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# relu => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%squeeze,), kwargs = {})
# %le_1 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/gi/cgiwoddpur5tomnu7k3swnvze3mtzxnbwbutjxviojoe347yddjj.py
# Topologically Sorted Source Nodes: [add, x_2], Original ATen: [aten.add, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# add => add
# x_2 => relu_1
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%squeeze_1, %primals_1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_1, 0), kwargs = {})
triton_poi_fused_add_relu_threshold_backward_1 = async_compile.triton('triton_poi_fused_add_relu_threshold_backward_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i1', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_relu_threshold_backward_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_relu_threshold_backward_1(in_out_ptr0, in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (x2), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = tl.full([1], 0, tl.int32)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = 0.0
tmp8 = tmp6 <= tmp7
tl.store(in_out_ptr0 + (x2), tmp6, xmask)
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1, 4, 4), (16, 4, 1), 0), primals_2, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf0, (1, 4, 4), (16, 4, 1))
buf1 = reinterpret_tensor(buf0, (4, 4), (4, 1), 0); del buf0 # reuse
buf5 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_3, buf5, 16, grid=grid(16), stream=stream0)
del primals_3
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(buf1, (1, 4, 4), (0, 4, 1), 0), primals_4, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf2, (1, 4, 4), (16, 4, 1))
buf3 = reinterpret_tensor(buf2, (4, 4), (4, 1), 0); del buf2 # reuse
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [add, x_2], Original ATen: [aten.add, aten.relu, aten.threshold_backward]
triton_poi_fused_add_relu_threshold_backward_1.run(buf3, primals_5, primals_1, buf4, 16, grid=grid(16), stream=stream0)
del primals_5
return (buf3, primals_2, primals_4, reinterpret_tensor(primals_1, (1, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf1, (1, 4, 4), (16, 4, 1), 0), buf4, buf5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch.nn as nn
class ResidualBlock(nn.Module):
def __init__(self, in_channels, out_channels, hidden_dim):
super().__init__()
self.fc_0 = nn.Conv1d(in_channels, hidden_dim, 1)
self.fc_1 = nn.Conv1d(hidden_dim, out_channels, 1)
self.activation = nn.ReLU()
if in_channels != out_channels:
self.shortcut = nn.Conv1d(in_channels, out_channels, 1)
else:
self.shortcut = nn.Identity()
nn.init.zeros_(self.fc_1.weight)
def forward(self, x):
x_short = self.shortcut(x)
x = self.fc_0(x)
x = self.fc_1(self.activation(x))
x = self.activation(x + x_short)
return x
def get_inputs():
return [torch.rand([4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'hidden_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.utils.data
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_add_relu_threshold_backward_1(in_out_ptr0, in_ptr0,
in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + x2, xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = tl.full([1], 0, tl.int32)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = 0.0
tmp8 = tmp6 <= tmp7
tl.store(in_out_ptr0 + x2, tmp6, xmask)
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1,
4, 4), (16, 4, 1), 0), primals_2, stride=(1,), padding=(0,),
dilation=(1,), transposed=False, output_padding=(0,), groups=1,
bias=None)
assert_size_stride(buf0, (1, 4, 4), (16, 4, 1))
buf1 = reinterpret_tensor(buf0, (4, 4), (4, 1), 0)
del buf0
buf5 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(16)](buf1,
primals_3, buf5, 16, XBLOCK=16, num_warps=1, num_stages=1)
del primals_3
buf2 = extern_kernels.convolution(reinterpret_tensor(buf1, (1, 4, 4
), (0, 4, 1), 0), primals_4, stride=(1,), padding=(0,),
dilation=(1,), transposed=False, output_padding=(0,), groups=1,
bias=None)
assert_size_stride(buf2, (1, 4, 4), (16, 4, 1))
buf3 = reinterpret_tensor(buf2, (4, 4), (4, 1), 0)
del buf2
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_add_relu_threshold_backward_1[grid(16)](buf3,
primals_5, primals_1, buf4, 16, XBLOCK=16, num_warps=1,
num_stages=1)
del primals_5
return buf3, primals_2, primals_4, reinterpret_tensor(primals_1, (1, 4,
4), (16, 4, 1), 0), reinterpret_tensor(buf1, (1, 4, 4), (16, 4, 1), 0
), buf4, buf5
class ResidualBlockNew(nn.Module):
def __init__(self, in_channels, out_channels, hidden_dim):
super().__init__()
self.fc_0 = nn.Conv1d(in_channels, hidden_dim, 1)
self.fc_1 = nn.Conv1d(hidden_dim, out_channels, 1)
self.activation = nn.ReLU()
if in_channels != out_channels:
self.shortcut = nn.Conv1d(in_channels, out_channels, 1)
else:
self.shortcut = nn.Identity()
nn.init.zeros_(self.fc_1.weight)
def forward(self, input_0):
primals_2 = self.fc_0.weight
primals_3 = self.fc_0.bias
primals_4 = self.fc_1.weight
primals_5 = self.fc_1.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
StructuralNeurobiologyLab/LightConvPoint
|
ResidualBlock
| false | 14,444 |
[
"Apache-2.0"
] | 58 |
3f353f45e9e910fa390a74520dfd478e3e88f104
|
https://github.com/StructuralNeurobiologyLab/LightConvPoint/tree/3f353f45e9e910fa390a74520dfd478e3e88f104
|
SplAtConv2d
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/cq/ccqcunoz44ytyqzy34r7cs2t74rk42s65mtajwcrygug6wcgzq24.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x => convolution
# x_1 => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_1, %primals_2, %primals_3, [1, 1], [0, 0], [1, 1], False, [0, 0], 2), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
# %le_1 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 8
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ve/cvelarc6diodfr33zpvmvgzuuzkvprbz6m6y4ohg2zlecikuyoyi.py
# Topologically Sorted Source Nodes: [add, gap, gap_1], Original ATen: [aten.add, aten.mean]
# Source node to ATen node mapping:
# add => add
# gap => add_1
# gap_1 => mean
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_2, 0), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add, %getitem_5), kwargs = {})
# %mean : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%add_1, [-1, -2], True), kwargs = {})
triton_poi_fused_add_mean_1 = async_compile.triton('triton_poi_fused_add_mean_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mean_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mean_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (8*x1)), xmask)
tmp3 = tl.load(in_ptr0 + (4 + x0 + (8*x1)), xmask)
tmp1 = 0.0
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = 1.0
tmp6 = tmp4 / tmp5
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/t4/ct4hhjqt2vge2xiycaomw3jiwzw326vnuf5jpebeysc4mpxrpciw.py
# Topologically Sorted Source Nodes: [x_2, gap_2], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# gap_2 => relu_1
# x_2 => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%mean, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_1,), kwargs = {})
triton_poi_fused_convolution_relu_2 = async_compile.triton('triton_poi_fused_convolution_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ym/cymnwinsgdg6rzof735q2mbji4z4uuuzffbkib2mmjzjggqvt5ti.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_3 => convolution_2
# Graph fragment:
# %convolution_2 : [num_users=2] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_1, %primals_6, %primals_7, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_3 = async_compile.triton('triton_poi_fused_convolution_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 8
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/al/calq4luacvjpbhaq6oadffm56qzlddnaiqtfsg2rnpgdoin4doyd.py
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# x_5 => amax, div, exp, sub, sum_1
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%permute, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%permute, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_4 = async_compile.triton('triton_poi_fused__softmax_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 4
x2 = (xindex // 8)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (8*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (4 + x0 + (8*x2)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp4 = tmp0 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 - tmp3
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp2 - tmp3
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp7 + tmp9
tmp11 = tmp5 / tmp10
tl.store(out_ptr0 + (x3), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/np/cnpcapaif5ggvdkeg53tnm4dojax33drj6dlnhtwxrttlaoiy23c.py
# Topologically Sorted Source Nodes: [mul, mul_1, add_2, out], Original ATen: [aten.mul, aten.add]
# Source node to ATen node mapping:
# add_2 => add_2
# mul => mul
# mul_1 => mul_1
# out => add_3
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%getitem_6, %getitem_2), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%getitem_7, %getitem_5), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, 0), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_2, %mul_1), kwargs = {})
triton_poi_fused_add_mul_5 = async_compile.triton('triton_poi_fused_add_mul_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_5(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (8*x1)), xmask)
tmp1 = tl.load(in_ptr1 + (x0 + (8*x1)), xmask)
tmp5 = tl.load(in_ptr0 + (4 + x0 + (8*x1)), xmask)
tmp6 = tl.load(in_ptr1 + (4 + x0 + (8*x1)), xmask)
tmp2 = tmp0 * tmp1
tmp3 = 0.0
tmp4 = tmp2 + tmp3
tmp7 = tmp5 * tmp6
tmp8 = tmp4 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (8, 2, 4, 4), (32, 16, 4, 1))
assert_size_stride(primals_3, (8, ), (1, ))
assert_size_stride(primals_4, (32, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_5, (32, ), (1, ))
assert_size_stride(primals_6, (8, 32, 1, 1), (32, 1, 1, 1))
assert_size_stride(primals_7, (8, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=2, bias=None)
assert_size_stride(buf0, (4, 8, 1, 1), (8, 1, 1, 1))
buf1 = reinterpret_tensor(buf0, (4, 8, 1, 1), (8, 1, 32, 32), 0); del buf0 # reuse
buf9 = empty_strided_cuda((4, 8, 1, 1), (8, 1, 1, 1), torch.bool)
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_threshold_backward_0.run(buf1, primals_3, buf9, 32, grid=grid(32), stream=stream0)
del primals_3
buf2 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [add, gap, gap_1], Original ATen: [aten.add, aten.mean]
triton_poi_fused_add_mean_1.run(buf1, buf2, 16, grid=grid(16), stream=stream0)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(buf2, primals_4, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 32, 1, 1), (32, 1, 1, 1))
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [x_2, gap_2], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_2.run(buf4, primals_5, 128, grid=grid(128), stream=stream0)
del primals_5
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
buf5 = extern_kernels.convolution(buf4, primals_6, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf5, (4, 8, 1, 1), (8, 1, 1, 1))
buf6 = buf5; del buf5 # reuse
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
triton_poi_fused_convolution_3.run(buf6, primals_7, 32, grid=grid(32), stream=stream0)
del primals_7
buf7 = empty_strided_cuda((4, 2, 1, 4), (8, 4, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten._softmax]
triton_poi_fused__softmax_4.run(buf6, buf7, 32, grid=grid(32), stream=stream0)
buf8 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, mul_1, add_2, out], Original ATen: [aten.mul, aten.add]
triton_poi_fused_add_mul_5.run(buf7, buf1, buf8, 16, grid=grid(16), stream=stream0)
return (buf8, primals_1, primals_2, primals_4, primals_6, reinterpret_tensor(buf1, (4, 4, 1, 1), (8, 1, 1, 1), 0), reinterpret_tensor(buf1, (4, 4, 1, 1), (8, 1, 1, 1), 4), buf2, buf4, buf6, reinterpret_tensor(buf7, (4, 4, 1, 1), (8, 1, 1, 1), 0), reinterpret_tensor(buf7, (4, 4, 1, 1), (8, 1, 1, 1), 4), buf9, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((8, 2, 4, 4), (32, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((32, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((8, 32, 1, 1), (32, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
from torch.autograd import Function
from torch.nn import Module
import logging
import torch
import torch.utils.data
import torch.distributed as dist
from torch import nn
import torch.nn.functional as F
from torch.autograd.function import Function
from torch.autograd import Function
from torch.nn.modules.utils import _pair
from torch.nn import BatchNorm2d
from torch.nn import ReLU
def get_norm(norm, out_channels):
"""
Args:
norm (str or callable): either one of BN, SyncBN, FrozenBN, GN;
or a callable that takes a channel number and returns
the normalization layer as a nn.Module.
Returns:
nn.Module or None: the normalization layer
"""
if isinstance(norm, str):
if len(norm) == 0:
return None
norm = {'BN': BatchNorm2d, 'SyncBN': NaiveSyncBatchNorm, 'FrozenBN':
FrozenBatchNorm2d, 'GN': lambda channels: nn.GroupNorm(32,
channels), 'nnSyncBN': nn.SyncBatchNorm}[norm]
return norm(out_channels)
class FrozenBatchNorm2d(nn.Module):
"""
BatchNorm2d where the batch statistics and the affine parameters are fixed.
It contains non-trainable buffers called
"weight" and "bias", "running_mean", "running_var",
initialized to perform identity transformation.
The pre-trained backbone models from Caffe2 only contain "weight" and "bias",
which are computed from the original four parameters of BN.
The affine transform `x * weight + bias` will perform the equivalent
computation of `(x - running_mean) / sqrt(running_var) * weight + bias`.
When loading a backbone model from Caffe2, "running_mean" and "running_var"
will be left unchanged as identity transformation.
Other pre-trained backbone models may contain all 4 parameters.
The forward is implemented by `F.batch_norm(..., training=False)`.
"""
_version = 3
def __init__(self, num_features, eps=1e-05):
super().__init__()
self.num_features = num_features
self.eps = eps
self.register_buffer('weight', torch.ones(num_features))
self.register_buffer('bias', torch.zeros(num_features))
self.register_buffer('running_mean', torch.zeros(num_features))
self.register_buffer('running_var', torch.ones(num_features) - eps)
def forward(self, x):
if x.requires_grad:
scale = self.weight * (self.running_var + self.eps).rsqrt()
bias = self.bias - self.running_mean * scale
scale = scale.reshape(1, -1, 1, 1)
bias = bias.reshape(1, -1, 1, 1)
return x * scale + bias
else:
return F.batch_norm(x, self.running_mean, self.running_var,
self.weight, self.bias, training=False, eps=self.eps)
def _load_from_state_dict(self, state_dict, prefix, local_metadata,
strict, missing_keys, unexpected_keys, error_msgs):
version = local_metadata.get('version', None)
if version is None or version < 2:
if prefix + 'running_mean' not in state_dict:
state_dict[prefix + 'running_mean'] = torch.zeros_like(self
.running_mean)
if prefix + 'running_var' not in state_dict:
state_dict[prefix + 'running_var'] = torch.ones_like(self.
running_var)
if version is not None and version < 3:
logger = logging.getLogger(__name__)
logger.info('FrozenBatchNorm {} is upgraded to version 3.'.
format(prefix.rstrip('.')))
state_dict[prefix + 'running_var'] -= self.eps
super()._load_from_state_dict(state_dict, prefix, local_metadata,
strict, missing_keys, unexpected_keys, error_msgs)
def __repr__(self):
return 'FrozenBatchNorm2d(num_features={}, eps={})'.format(self.
num_features, self.eps)
@classmethod
def convert_frozen_batchnorm(cls, module):
"""
Convert BatchNorm/SyncBatchNorm in module into FrozenBatchNorm.
Args:
module (torch.nn.Module):
Returns:
If module is BatchNorm/SyncBatchNorm, returns a new module.
Otherwise, in-place convert module and return it.
Similar to convert_sync_batchnorm in
https://github.com/pytorch/pytorch/blob/master/torch/nn/modules/batchnorm.py
"""
bn_module = nn.modules.batchnorm
bn_module = bn_module.BatchNorm2d, bn_module.SyncBatchNorm
res = module
if isinstance(module, bn_module):
res = cls(module.num_features)
if module.affine:
res.weight.data = module.weight.data.clone().detach()
res.bias.data = module.bias.data.clone().detach()
res.running_mean.data = module.running_mean.data
res.running_var.data = module.running_var.data
res.eps = module.eps
else:
for name, child in module.named_children():
new_child = cls.convert_frozen_batchnorm(child)
if new_child is not child:
res.add_module(name, new_child)
return res
class _NewEmptyTensorOp(torch.autograd.Function):
@staticmethod
def forward(ctx, x, new_shape):
ctx.shape = x.shape
return x.new_empty(new_shape)
@staticmethod
def backward(ctx, grad):
shape = ctx.shape
return _NewEmptyTensorOp.apply(grad, shape), None
class Conv2d(torch.nn.Conv2d):
"""
A wrapper around :class:`torch.nn.Conv2d` to support empty inputs and more features.
"""
def __init__(self, *args, **kwargs):
"""
Extra keyword arguments supported in addition to those in `torch.nn.Conv2d`:
Args:
norm (nn.Module, optional): a normalization layer
activation (callable(Tensor) -> Tensor): a callable activation function
It assumes that norm layer is used before activation.
"""
norm = kwargs.pop('norm', None)
activation = kwargs.pop('activation', None)
super().__init__(*args, **kwargs)
self.norm = norm
self.activation = activation
def forward(self, x):
if x.numel() == 0 and self.training:
assert not isinstance(self.norm, torch.nn.SyncBatchNorm
), 'SyncBatchNorm does not support empty inputs!'
if x.numel() == 0 and TORCH_VERSION <= (1, 4):
assert not isinstance(self.norm, torch.nn.GroupNorm
), 'GroupNorm does not support empty inputs in PyTorch <=1.4!'
output_shape = [((i + 2 * p - (di * (k - 1) + 1)) // s + 1) for
i, p, di, k, s in zip(x.shape[-2:], self.padding, self.
dilation, self.kernel_size, self.stride)]
output_shape = [x.shape[0], self.weight.shape[0]] + output_shape
empty = _NewEmptyTensorOp.apply(x, output_shape)
if self.training:
_dummy = sum(x.view(-1)[0] for x in self.parameters()) * 0.0
return empty + _dummy
else:
return empty
x = super().forward(x)
if self.norm is not None:
x = self.norm(x)
if self.activation is not None:
x = self.activation(x)
return x
class AllReduce(Function):
@staticmethod
def forward(ctx, input):
input_list = [torch.zeros_like(input) for k in range(dist.
get_world_size())]
dist.all_gather(input_list, input, async_op=False)
inputs = torch.stack(input_list, dim=0)
return torch.sum(inputs, dim=0)
@staticmethod
def backward(ctx, grad_output):
dist.all_reduce(grad_output, async_op=False)
return grad_output
class NaiveSyncBatchNorm(BatchNorm2d):
"""
`torch.nn.SyncBatchNorm` has known unknown bugs.
It produces significantly worse AP (and sometimes goes NaN)
when the batch size on each worker is quite different
(e.g., when scale augmentation is used, or when it is applied to mask head).
Use this implementation before `nn.SyncBatchNorm` is fixed.
It is slower than `nn.SyncBatchNorm`.
Note:
There isn't a single definition of Sync BatchNorm.
When ``stats_mode==""``, this module computes overall statistics by using
statistics of each worker with equal weight. The result is true statistics
of all samples (as if they are all on one worker) only when all workers
have the same (N, H, W). This mode does not support inputs with zero batch size.
When ``stats_mode=="N"``, this module computes overall statistics by weighting
the statistics of each worker by their ``N``. The result is true statistics
of all samples (as if they are all on one worker) only when all workers
have the same (H, W). It is slower than ``stats_mode==""``.
Even though the result of this module may not be the true statistics of all samples,
it may still be reasonable because it might be preferrable to assign equal weights
to all workers, regardless of their (H, W) dimension, instead of putting larger weight
on larger images. From preliminary experiments, little difference is found between such
a simplified implementation and an accurate computation of overall mean & variance.
"""
def __init__(self, *args, stats_mode='', **kwargs):
super().__init__(*args, **kwargs)
assert stats_mode in ['', 'N']
self._stats_mode = stats_mode
def forward(self, input):
if comm.get_world_size() == 1 or not self.training:
return super().forward(input)
B, C = input.shape[0], input.shape[1]
mean = torch.mean(input, dim=[0, 2, 3])
meansqr = torch.mean(input * input, dim=[0, 2, 3])
if self._stats_mode == '':
assert B > 0, 'SyncBatchNorm(stats_mode="") does not support zero batch size.'
vec = torch.cat([mean, meansqr], dim=0)
vec = AllReduce.apply(vec) * (1.0 / dist.get_world_size())
mean, meansqr = torch.split(vec, C)
momentum = self.momentum
else:
if B == 0:
vec = torch.zeros([2 * C + 1], device=mean.device, dtype=
mean.dtype)
vec = vec + input.sum()
else:
vec = torch.cat([mean, meansqr, torch.ones([1], device=mean
.device, dtype=mean.dtype)], dim=0)
vec = AllReduce.apply(vec * B)
total_batch = vec[-1].detach()
momentum = total_batch.clamp(max=1) * self.momentum
total_batch = torch.max(total_batch, torch.ones_like(total_batch))
mean, meansqr, _ = torch.split(vec / total_batch, C)
var = meansqr - mean * mean
invstd = torch.rsqrt(var + self.eps)
scale = self.weight * invstd
bias = self.bias - mean * scale
scale = scale.reshape(1, -1, 1, 1)
bias = bias.reshape(1, -1, 1, 1)
self.running_mean += momentum * (mean.detach() - self.running_mean)
self.running_var += momentum * (var.detach() - self.running_var)
return input * scale + bias
class rSoftMax(nn.Module):
def __init__(self, radix, cardinality):
super().__init__()
self.radix = radix
self.cardinality = cardinality
def forward(self, x):
batch = x.size(0)
if self.radix > 1:
x = x.view(batch, self.cardinality, self.radix, -1).transpose(1, 2)
x = F.softmax(x, dim=1)
x = x.reshape(batch, -1)
else:
x = torch.sigmoid(x)
return x
class SplAtConv2d(Module):
"""Split-Attention Conv2d
"""
def __init__(self, in_channels, channels, kernel_size, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), groups=1, bias=True, radix=2,
reduction_factor=4, rectify=False, rectify_avg=False, norm=None,
dropblock_prob=0.0, **kwargs):
super(SplAtConv2d, self).__init__()
padding = _pair(padding)
self.rectify = rectify and (padding[0] > 0 or padding[1] > 0)
self.rectify_avg = rectify_avg
inter_channels = max(in_channels * radix // reduction_factor, 32)
self.radix = radix
self.cardinality = groups
self.channels = channels
self.dropblock_prob = dropblock_prob
if self.rectify:
self.conv = RFConv2d(in_channels, channels * radix, kernel_size,
stride, padding, dilation, groups=groups * radix, bias=bias,
average_mode=rectify_avg, **kwargs)
else:
self.conv = Conv2d(in_channels, channels * radix, kernel_size,
stride, padding, dilation, groups=groups * radix, bias=bias,
**kwargs)
self.use_bn = norm is not None
if self.use_bn:
self.bn0 = get_norm(norm, channels * radix)
self.relu = ReLU(inplace=True)
self.fc1 = Conv2d(channels, inter_channels, 1, groups=self.cardinality)
if self.use_bn:
self.bn1 = get_norm(norm, inter_channels)
self.fc2 = Conv2d(inter_channels, channels * radix, 1, groups=self.
cardinality)
if dropblock_prob > 0.0:
self.dropblock = DropBlock2D(dropblock_prob, 3)
self.rsoftmax = rSoftMax(radix, groups)
def forward(self, x):
x = self.conv(x)
if self.use_bn:
x = self.bn0(x)
if self.dropblock_prob > 0.0:
x = self.dropblock(x)
x = self.relu(x)
batch, rchannel = x.shape[:2]
if self.radix > 1:
splited = torch.split(x, rchannel // self.radix, dim=1)
gap = sum(splited)
else:
gap = x
gap = F.adaptive_avg_pool2d(gap, 1)
gap = self.fc1(gap)
if self.use_bn:
gap = self.bn1(gap)
gap = self.relu(gap)
atten = self.fc2(gap)
atten = self.rsoftmax(atten).view(batch, -1, 1, 1)
if self.radix > 1:
attens = torch.split(atten, rchannel // self.radix, dim=1)
out = sum([(att * split) for att, split in zip(attens, splited)])
else:
out = atten * x
return out.contiguous()
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'channels': 4, 'kernel_size': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch.autograd import Function
from torch.nn import Module
import logging
import torch.utils.data
import torch.distributed as dist
from torch import nn
import torch.nn.functional as F
from torch.autograd.function import Function
from torch.autograd import Function
from torch.nn.modules.utils import _pair
from torch.nn import BatchNorm2d
from torch.nn import ReLU
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_0(in_out_ptr0,
in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 8
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_add_mean_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 8 * x1), xmask)
tmp3 = tl.load(in_ptr0 + (4 + x0 + 8 * x1), xmask)
tmp1 = 0.0
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = 1.0
tmp6 = tmp4 / tmp5
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_2(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 8
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 4
x2 = xindex // 8
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 8 * x2), xmask, eviction_policy='evict_last'
)
tmp2 = tl.load(in_ptr0 + (4 + x0 + 8 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp4 = tmp0 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 - tmp3
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp2 - tmp3
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp7 + tmp9
tmp11 = tmp5 / tmp10
tl.store(out_ptr0 + x3, tmp11, xmask)
@triton.jit
def triton_poi_fused_add_mul_5(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 8 * x1), xmask)
tmp1 = tl.load(in_ptr1 + (x0 + 8 * x1), xmask)
tmp5 = tl.load(in_ptr0 + (4 + x0 + 8 * x1), xmask)
tmp6 = tl.load(in_ptr1 + (4 + x0 + 8 * x1), xmask)
tmp2 = tmp0 * tmp1
tmp3 = 0.0
tmp4 = tmp2 + tmp3
tmp7 = tmp5 * tmp6
tmp8 = tmp4 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (8, 2, 4, 4), (32, 16, 4, 1))
assert_size_stride(primals_3, (8,), (1,))
assert_size_stride(primals_4, (32, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_5, (32,), (1,))
assert_size_stride(primals_6, (8, 32, 1, 1), (32, 1, 1, 1))
assert_size_stride(primals_7, (8,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1,
1), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=2, bias=None)
assert_size_stride(buf0, (4, 8, 1, 1), (8, 1, 1, 1))
buf1 = reinterpret_tensor(buf0, (4, 8, 1, 1), (8, 1, 32, 32), 0)
del buf0
buf9 = empty_strided_cuda((4, 8, 1, 1), (8, 1, 1, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_convolution_relu_threshold_backward_0[grid(32)](buf1,
primals_3, buf9, 32, XBLOCK=32, num_warps=1, num_stages=1)
del primals_3
buf2 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
triton_poi_fused_add_mean_1[grid(16)](buf1, buf2, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf3 = extern_kernels.convolution(buf2, primals_4, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 32, 1, 1), (32, 1, 1, 1))
buf4 = buf3
del buf3
triton_poi_fused_convolution_relu_2[grid(128)](buf4, primals_5, 128,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
buf5 = extern_kernels.convolution(buf4, primals_6, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf5, (4, 8, 1, 1), (8, 1, 1, 1))
buf6 = buf5
del buf5
triton_poi_fused_convolution_3[grid(32)](buf6, primals_7, 32,
XBLOCK=32, num_warps=1, num_stages=1)
del primals_7
buf7 = empty_strided_cuda((4, 2, 1, 4), (8, 4, 4, 1), torch.float32)
triton_poi_fused__softmax_4[grid(32)](buf6, buf7, 32, XBLOCK=32,
num_warps=1, num_stages=1)
buf8 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
triton_poi_fused_add_mul_5[grid(16)](buf7, buf1, buf8, 16, XBLOCK=
16, num_warps=1, num_stages=1)
return (buf8, primals_1, primals_2, primals_4, primals_6,
reinterpret_tensor(buf1, (4, 4, 1, 1), (8, 1, 1, 1), 0),
reinterpret_tensor(buf1, (4, 4, 1, 1), (8, 1, 1, 1), 4), buf2, buf4,
buf6, reinterpret_tensor(buf7, (4, 4, 1, 1), (8, 1, 1, 1), 0),
reinterpret_tensor(buf7, (4, 4, 1, 1), (8, 1, 1, 1), 4), buf9)
def get_norm(norm, out_channels):
"""
Args:
norm (str or callable): either one of BN, SyncBN, FrozenBN, GN;
or a callable that takes a channel number and returns
the normalization layer as a nn.Module.
Returns:
nn.Module or None: the normalization layer
"""
if isinstance(norm, str):
if len(norm) == 0:
return None
norm = {'BN': BatchNorm2d, 'SyncBN': NaiveSyncBatchNorm, 'FrozenBN':
FrozenBatchNorm2d, 'GN': lambda channels: nn.GroupNorm(32,
channels), 'nnSyncBN': nn.SyncBatchNorm}[norm]
return norm(out_channels)
class FrozenBatchNorm2d(nn.Module):
"""
BatchNorm2d where the batch statistics and the affine parameters are fixed.
It contains non-trainable buffers called
"weight" and "bias", "running_mean", "running_var",
initialized to perform identity transformation.
The pre-trained backbone models from Caffe2 only contain "weight" and "bias",
which are computed from the original four parameters of BN.
The affine transform `x * weight + bias` will perform the equivalent
computation of `(x - running_mean) / sqrt(running_var) * weight + bias`.
When loading a backbone model from Caffe2, "running_mean" and "running_var"
will be left unchanged as identity transformation.
Other pre-trained backbone models may contain all 4 parameters.
The forward is implemented by `F.batch_norm(..., training=False)`.
"""
_version = 3
def __init__(self, num_features, eps=1e-05):
super().__init__()
self.num_features = num_features
self.eps = eps
self.register_buffer('weight', torch.ones(num_features))
self.register_buffer('bias', torch.zeros(num_features))
self.register_buffer('running_mean', torch.zeros(num_features))
self.register_buffer('running_var', torch.ones(num_features) - eps)
def forward(self, x):
if x.requires_grad:
scale = self.weight * (self.running_var + self.eps).rsqrt()
bias = self.bias - self.running_mean * scale
scale = scale.reshape(1, -1, 1, 1)
bias = bias.reshape(1, -1, 1, 1)
return x * scale + bias
else:
return F.batch_norm(x, self.running_mean, self.running_var,
self.weight, self.bias, training=False, eps=self.eps)
def _load_from_state_dict(self, state_dict, prefix, local_metadata,
strict, missing_keys, unexpected_keys, error_msgs):
version = local_metadata.get('version', None)
if version is None or version < 2:
if prefix + 'running_mean' not in state_dict:
state_dict[prefix + 'running_mean'] = torch.zeros_like(self
.running_mean)
if prefix + 'running_var' not in state_dict:
state_dict[prefix + 'running_var'] = torch.ones_like(self.
running_var)
if version is not None and version < 3:
logger = logging.getLogger(__name__)
logger.info('FrozenBatchNorm {} is upgraded to version 3.'.
format(prefix.rstrip('.')))
state_dict[prefix + 'running_var'] -= self.eps
super()._load_from_state_dict(state_dict, prefix, local_metadata,
strict, missing_keys, unexpected_keys, error_msgs)
def __repr__(self):
return 'FrozenBatchNorm2d(num_features={}, eps={})'.format(self.
num_features, self.eps)
@classmethod
def convert_frozen_batchnorm(cls, module):
"""
Convert BatchNorm/SyncBatchNorm in module into FrozenBatchNorm.
Args:
module (torch.nn.Module):
Returns:
If module is BatchNorm/SyncBatchNorm, returns a new module.
Otherwise, in-place convert module and return it.
Similar to convert_sync_batchnorm in
https://github.com/pytorch/pytorch/blob/master/torch/nn/modules/batchnorm.py
"""
bn_module = nn.modules.batchnorm
bn_module = bn_module.BatchNorm2d, bn_module.SyncBatchNorm
res = module
if isinstance(module, bn_module):
res = cls(module.num_features)
if module.affine:
res.weight.data = module.weight.data.clone().detach()
res.bias.data = module.bias.data.clone().detach()
res.running_mean.data = module.running_mean.data
res.running_var.data = module.running_var.data
res.eps = module.eps
else:
for name, child in module.named_children():
new_child = cls.convert_frozen_batchnorm(child)
if new_child is not child:
res.add_module(name, new_child)
return res
class _NewEmptyTensorOp(torch.autograd.Function):
@staticmethod
def forward(ctx, x, new_shape):
ctx.shape = x.shape
return x.new_empty(new_shape)
@staticmethod
def backward(ctx, grad):
shape = ctx.shape
return _NewEmptyTensorOp.apply(grad, shape), None
class Conv2d(torch.nn.Conv2d):
"""
A wrapper around :class:`torch.nn.Conv2d` to support empty inputs and more features.
"""
def __init__(self, *args, **kwargs):
"""
Extra keyword arguments supported in addition to those in `torch.nn.Conv2d`:
Args:
norm (nn.Module, optional): a normalization layer
activation (callable(Tensor) -> Tensor): a callable activation function
It assumes that norm layer is used before activation.
"""
norm = kwargs.pop('norm', None)
activation = kwargs.pop('activation', None)
super().__init__(*args, **kwargs)
self.norm = norm
self.activation = activation
def forward(self, x):
if x.numel() == 0 and self.training:
assert not isinstance(self.norm, torch.nn.SyncBatchNorm
), 'SyncBatchNorm does not support empty inputs!'
if x.numel() == 0 and TORCH_VERSION <= (1, 4):
assert not isinstance(self.norm, torch.nn.GroupNorm
), 'GroupNorm does not support empty inputs in PyTorch <=1.4!'
output_shape = [((i + 2 * p - (di * (k - 1) + 1)) // s + 1) for
i, p, di, k, s in zip(x.shape[-2:], self.padding, self.
dilation, self.kernel_size, self.stride)]
output_shape = [x.shape[0], self.weight.shape[0]] + output_shape
empty = _NewEmptyTensorOp.apply(x, output_shape)
if self.training:
_dummy = sum(x.view(-1)[0] for x in self.parameters()) * 0.0
return empty + _dummy
else:
return empty
x = super().forward(x)
if self.norm is not None:
x = self.norm(x)
if self.activation is not None:
x = self.activation(x)
return x
class AllReduce(Function):
@staticmethod
def forward(ctx, input):
input_list = [torch.zeros_like(input) for k in range(dist.
get_world_size())]
dist.all_gather(input_list, input, async_op=False)
inputs = torch.stack(input_list, dim=0)
return torch.sum(inputs, dim=0)
@staticmethod
def backward(ctx, grad_output):
dist.all_reduce(grad_output, async_op=False)
return grad_output
class NaiveSyncBatchNorm(BatchNorm2d):
"""
`torch.nn.SyncBatchNorm` has known unknown bugs.
It produces significantly worse AP (and sometimes goes NaN)
when the batch size on each worker is quite different
(e.g., when scale augmentation is used, or when it is applied to mask head).
Use this implementation before `nn.SyncBatchNorm` is fixed.
It is slower than `nn.SyncBatchNorm`.
Note:
There isn't a single definition of Sync BatchNorm.
When ``stats_mode==""``, this module computes overall statistics by using
statistics of each worker with equal weight. The result is true statistics
of all samples (as if they are all on one worker) only when all workers
have the same (N, H, W). This mode does not support inputs with zero batch size.
When ``stats_mode=="N"``, this module computes overall statistics by weighting
the statistics of each worker by their ``N``. The result is true statistics
of all samples (as if they are all on one worker) only when all workers
have the same (H, W). It is slower than ``stats_mode==""``.
Even though the result of this module may not be the true statistics of all samples,
it may still be reasonable because it might be preferrable to assign equal weights
to all workers, regardless of their (H, W) dimension, instead of putting larger weight
on larger images. From preliminary experiments, little difference is found between such
a simplified implementation and an accurate computation of overall mean & variance.
"""
def __init__(self, *args, stats_mode='', **kwargs):
super().__init__(*args, **kwargs)
assert stats_mode in ['', 'N']
self._stats_mode = stats_mode
def forward(self, input):
if comm.get_world_size() == 1 or not self.training:
return super().forward(input)
B, C = input.shape[0], input.shape[1]
mean = torch.mean(input, dim=[0, 2, 3])
meansqr = torch.mean(input * input, dim=[0, 2, 3])
if self._stats_mode == '':
assert B > 0, 'SyncBatchNorm(stats_mode="") does not support zero batch size.'
vec = torch.cat([mean, meansqr], dim=0)
vec = AllReduce.apply(vec) * (1.0 / dist.get_world_size())
mean, meansqr = torch.split(vec, C)
momentum = self.momentum
else:
if B == 0:
vec = torch.zeros([2 * C + 1], device=mean.device, dtype=
mean.dtype)
vec = vec + input.sum()
else:
vec = torch.cat([mean, meansqr, torch.ones([1], device=mean
.device, dtype=mean.dtype)], dim=0)
vec = AllReduce.apply(vec * B)
total_batch = vec[-1].detach()
momentum = total_batch.clamp(max=1) * self.momentum
total_batch = torch.max(total_batch, torch.ones_like(total_batch))
mean, meansqr, _ = torch.split(vec / total_batch, C)
var = meansqr - mean * mean
invstd = torch.rsqrt(var + self.eps)
scale = self.weight * invstd
bias = self.bias - mean * scale
scale = scale.reshape(1, -1, 1, 1)
bias = bias.reshape(1, -1, 1, 1)
self.running_mean += momentum * (mean.detach() - self.running_mean)
self.running_var += momentum * (var.detach() - self.running_var)
return input * scale + bias
class rSoftMax(nn.Module):
def __init__(self, radix, cardinality):
super().__init__()
self.radix = radix
self.cardinality = cardinality
def forward(self, x):
batch = x.size(0)
if self.radix > 1:
x = x.view(batch, self.cardinality, self.radix, -1).transpose(1, 2)
x = F.softmax(x, dim=1)
x = x.reshape(batch, -1)
else:
x = torch.sigmoid(x)
return x
class SplAtConv2dNew(Module):
"""Split-Attention Conv2d
"""
def __init__(self, in_channels, channels, kernel_size, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), groups=1, bias=True, radix=2,
reduction_factor=4, rectify=False, rectify_avg=False, norm=None,
dropblock_prob=0.0, **kwargs):
super(SplAtConv2dNew, self).__init__()
padding = _pair(padding)
self.rectify = rectify and (padding[0] > 0 or padding[1] > 0)
self.rectify_avg = rectify_avg
inter_channels = max(in_channels * radix // reduction_factor, 32)
self.radix = radix
self.cardinality = groups
self.channels = channels
self.dropblock_prob = dropblock_prob
if self.rectify:
self.conv = RFConv2d(in_channels, channels * radix, kernel_size,
stride, padding, dilation, groups=groups * radix, bias=bias,
average_mode=rectify_avg, **kwargs)
else:
self.conv = Conv2d(in_channels, channels * radix, kernel_size,
stride, padding, dilation, groups=groups * radix, bias=bias,
**kwargs)
self.use_bn = norm is not None
if self.use_bn:
self.bn0 = get_norm(norm, channels * radix)
self.relu = ReLU(inplace=True)
self.fc1 = Conv2d(channels, inter_channels, 1, groups=self.cardinality)
if self.use_bn:
self.bn1 = get_norm(norm, inter_channels)
self.fc2 = Conv2d(inter_channels, channels * radix, 1, groups=self.
cardinality)
if dropblock_prob > 0.0:
self.dropblock = DropBlock2D(dropblock_prob, 3)
self.rsoftmax = rSoftMax(radix, groups)
def forward(self, input_0):
primals_2 = self.conv.weight
primals_3 = self.conv.bias
primals_4 = self.fc1.weight
primals_5 = self.fc1.bias
primals_6 = self.fc2.weight
primals_7 = self.fc2.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
|
Shun14/detectron2-ResNeSt
|
SplAtConv2d
| false | 14,445 |
[
"Apache-2.0"
] | 344 |
cda53a237199da3bbe7526d41c41b9d8df4c4814
|
https://github.com/Shun14/detectron2-ResNeSt/tree/cda53a237199da3bbe7526d41c41b9d8df4c4814
|
Net
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/r3/cr3febcwm3t44fuoitsx3ou2p6xg4sk4f7unagmmrvffasxf47te.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le_2 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, 4), (4, 1))
assert_size_stride(primals_9, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf0 # reuse
buf9 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf9, 256, grid=grid(256), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf1, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
buf3 = reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf2 # reuse
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_0.run(buf3, primals_5, buf8, 256, grid=grid(256), stream=stream0)
del primals_5
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf4 # reuse
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_0.run(buf5, primals_7, buf7, 256, grid=grid(256), stream=stream0)
del primals_7
buf6 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_9, reinterpret_tensor(buf5, (64, 4), (4, 1), 0), reinterpret_tensor(primals_8, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf6)
del primals_9
return (reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf1, (64, 4), (4, 1), 0), reinterpret_tensor(buf3, (64, 4), (4, 1), 0), reinterpret_tensor(buf5, (64, 4), (4, 1), 0), primals_8, buf7, primals_6, buf8, primals_4, buf9, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch.nn import functional as F
class Net(torch.nn.Module):
def __init__(self, n_feature, n_hidden, n_output):
super(Net, self).__init__()
self.hidden = torch.nn.Linear(n_feature, n_hidden)
self.hidden_two = torch.nn.Linear(n_hidden, n_hidden)
self.hidden_3 = torch.nn.Linear(n_hidden, n_hidden)
self.predict = torch.nn.Linear(n_hidden, n_output)
def forward(self, x):
x = F.relu(self.hidden(x))
x = F.relu(self.hidden_two(x))
x = F.relu(self.hidden_3(x))
x = self.predict(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'n_feature': 4, 'n_hidden': 4, 'n_output': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4, 4), (4, 1))
assert_size_stride(primals_9, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf0
buf9 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(256)](buf1,
primals_2, buf9, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf1, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
buf3 = reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf2
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_relu_threshold_backward_0[grid(256)](buf3,
primals_5, buf8, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_5
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf4
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_relu_threshold_backward_0[grid(256)](buf5,
primals_7, buf7, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_7
buf6 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_9, reinterpret_tensor(buf5, (64, 4), (
4, 1), 0), reinterpret_tensor(primals_8, (4, 4), (1, 4), 0),
alpha=1, beta=1, out=buf6)
del primals_9
return reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), reinterpret_tensor(buf1, (64, 4), (4, 1), 0), reinterpret_tensor(
buf3, (64, 4), (4, 1), 0), reinterpret_tensor(buf5, (64, 4), (4, 1), 0
), primals_8, buf7, primals_6, buf8, primals_4, buf9
class NetNew(torch.nn.Module):
def __init__(self, n_feature, n_hidden, n_output):
super(NetNew, self).__init__()
self.hidden = torch.nn.Linear(n_feature, n_hidden)
self.hidden_two = torch.nn.Linear(n_hidden, n_hidden)
self.hidden_3 = torch.nn.Linear(n_hidden, n_hidden)
self.predict = torch.nn.Linear(n_hidden, n_output)
def forward(self, input_0):
primals_1 = self.hidden.weight
primals_2 = self.hidden.bias
primals_4 = self.hidden_two.weight
primals_5 = self.hidden_two.bias
primals_6 = self.hidden_3.weight
primals_7 = self.hidden_3.bias
primals_8 = self.predict.weight
primals_9 = self.predict.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
|
SunHaozhe/modular-metalearning
|
Net
| false | 14,446 |
[
"MIT"
] | 70 |
c94dd18c6d105f18667d4de7bb4c81fa538a541c
|
https://github.com/SunHaozhe/modular-metalearning/tree/c94dd18c6d105f18667d4de7bb4c81fa538a541c
|
VarifocalLoss
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ax/caxbx6i4bwtsbtfhik5fb25fzuvlc7qeuptufx227j5xebtapxg3.py
# Topologically Sorted Source Nodes: [binary_cross_entropy_with_logits, gt, float_1, mul, pred_sigmoid, sub, abs_1, pow_1, mul_1, le, float_2, mul_2, focal_weight, loss, loss_1, loss_cls], Original ATen: [aten.binary_cross_entropy_with_logits, aten.gt, aten._to_copy, aten.mul, aten.sigmoid, aten.sub, aten.abs, aten.pow, aten.le, aten.add, aten.mean]
# Source node to ATen node mapping:
# abs_1 => abs_1
# binary_cross_entropy_with_logits => abs_2, exp, full_default, log1p, minimum, mul_3, neg, sub_1, sub_2, sub_3
# float_1 => convert_element_type
# float_2 => convert_element_type_1
# focal_weight => add
# gt => gt
# le => le
# loss => mul_4
# loss_1 => mean
# loss_cls => mul_5
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# pow_1 => pow_1
# pred_sigmoid => sigmoid
# sub => sub
# Graph fragment:
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg1_1), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %arg0_1), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %minimum : [num_users=1] = call_function[target=torch.ops.aten.minimum.default](args = (%full_default, %arg0_1), kwargs = {})
# %abs_2 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%arg0_1,), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%abs_2,), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%neg,), kwargs = {})
# %log1p : [num_users=1] = call_function[target=torch.ops.aten.log1p.default](args = (%exp,), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%minimum, %log1p), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_3, %sub_2), kwargs = {})
# %gt : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%arg1_1, 0.0), kwargs = {})
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%gt, torch.float32), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, %convert_element_type), kwargs = {})
# %sigmoid : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%arg0_1,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sigmoid, %arg1_1), kwargs = {})
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub,), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%abs_1, 2.0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%pow_1, 0.75), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%arg1_1, 0.0), kwargs = {})
# %convert_element_type_1 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%le, torch.float32), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_1, %convert_element_type_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_2), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_3, %add), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%mul_4,), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mean, 1.0), kwargs = {})
triton_per_fused__to_copy_abs_add_binary_cross_entropy_with_logits_gt_le_mean_mul_pow_sigmoid_sub_0 = async_compile.triton('triton_per_fused__to_copy_abs_add_binary_cross_entropy_with_logits_gt_le_mean_mul_pow_sigmoid_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__to_copy_abs_add_binary_cross_entropy_with_logits_gt_le_mean_mul_pow_sigmoid_sub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__to_copy_abs_add_binary_cross_entropy_with_logits_gt_le_mean_mul_pow_sigmoid_sub_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp3 = tl.load(in_ptr1 + (r0), None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp2 * tmp3
tmp5 = 0.0
tmp6 = triton_helpers.minimum(tmp5, tmp3)
tmp7 = tl_math.abs(tmp3)
tmp8 = -tmp7
tmp9 = tl_math.exp(tmp8)
tmp10 = libdevice.log1p(tmp9)
tmp11 = tmp6 - tmp10
tmp12 = tmp4 - tmp11
tmp13 = tmp0 > tmp5
tmp14 = tmp13.to(tl.float32)
tmp15 = tmp0 * tmp14
tmp16 = tl.sigmoid(tmp3)
tmp17 = tmp16 - tmp0
tmp18 = tl_math.abs(tmp17)
tmp19 = tmp18 * tmp18
tmp20 = 0.75
tmp21 = tmp19 * tmp20
tmp22 = tmp0 <= tmp5
tmp23 = tmp22.to(tl.float32)
tmp24 = tmp21 * tmp23
tmp25 = tmp15 + tmp24
tmp26 = tmp12 * tmp25
tmp27 = tl.broadcast_to(tmp26, [RBLOCK])
tmp29 = triton_helpers.promote_to_tensor(tl.sum(tmp27, 0))
tmp30 = 256.0
tmp31 = tmp29 / tmp30
tmp32 = tmp31 * tmp1
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp32, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [binary_cross_entropy_with_logits, gt, float_1, mul, pred_sigmoid, sub, abs_1, pow_1, mul_1, le, float_2, mul_2, focal_weight, loss, loss_1, loss_cls], Original ATen: [aten.binary_cross_entropy_with_logits, aten.gt, aten._to_copy, aten.mul, aten.sigmoid, aten.sub, aten.abs, aten.pow, aten.le, aten.add, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused__to_copy_abs_add_binary_cross_entropy_with_logits_gt_le_mean_mul_pow_sigmoid_sub_0.run(buf1, arg1_1, arg0_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn.functional as F
import torch.nn as nn
def reduce_loss(loss, reduction):
"""Reduce loss as specified.
Args:
loss (Tensor): Elementwise loss tensor.
reduction (str): Options are "none", "mean" and "sum".
Return:
Tensor: Reduced loss tensor.
"""
reduction_enum = F._Reduction.get_enum(reduction)
if reduction_enum == 0:
return loss
elif reduction_enum == 1:
return loss.mean()
elif reduction_enum == 2:
return loss.sum()
def weight_reduce_loss(loss, weight=None, reduction='mean', avg_factor=None):
"""Apply element-wise weight and reduce loss.
Args:
loss (Tensor): Element-wise loss.
weight (Tensor): Element-wise weights.
reduction (str): Same as built-in losses of PyTorch.
avg_factor (float): Avarage factor when computing the mean of losses.
Returns:
Tensor: Processed loss values.
"""
if weight is not None:
loss = loss * weight
if avg_factor is None:
loss = reduce_loss(loss, reduction)
elif reduction == 'mean':
loss = loss.sum() / avg_factor
elif reduction != 'none':
raise ValueError('avg_factor can not be used with reduction="sum"')
return loss
def varifocal_loss(pred, target, weight=None, alpha=0.75, gamma=2.0,
iou_weighted=True, reduction='mean', avg_factor=None):
"""`Varifocal Loss <https://arxiv.org/abs/2008.13367>`_
Args:
pred (torch.Tensor): The prediction with shape (N, C), C is the
number of classes
target (torch.Tensor): The learning target of the iou-aware
classification score with shape (N, C), C is the number of classes.
weight (torch.Tensor, optional): The weight of loss for each
prediction. Defaults to None.
alpha (float, optional): A balance factor for the negative part of
Varifocal Loss, which is different from the alpha of Focal Loss.
Defaults to 0.75.
gamma (float, optional): The gamma for calculating the modulating
factor. Defaults to 2.0.
iou_weighted (bool, optional): Whether to weight the loss of the
positive example with the iou target. Defaults to True.
reduction (str, optional): The method used to reduce the loss into
a scalar. Defaults to 'mean'. Options are "none", "mean" and
"sum".
avg_factor (int, optional): Average factor that is used to average
the loss. Defaults to None.
"""
assert pred.size() == target.size()
pred_sigmoid = pred.sigmoid()
target = target.type_as(pred)
if iou_weighted:
focal_weight = target * (target > 0.0).float() + alpha * (pred_sigmoid
- target).abs().pow(gamma) * (target <= 0.0).float()
else:
focal_weight = (target > 0.0).float() + alpha * (pred_sigmoid - target
).abs().pow(gamma) * (target <= 0.0).float()
loss = F.binary_cross_entropy_with_logits(pred, target, reduction='none'
) * focal_weight
loss = weight_reduce_loss(loss, weight, reduction, avg_factor)
return loss
class VarifocalLoss(nn.Module):
def __init__(self, use_sigmoid=True, alpha=0.75, gamma=2.0,
iou_weighted=True, reduction='mean', loss_weight=1.0):
"""`Varifocal Loss <https://arxiv.org/abs/2008.13367>`_
Args:
use_sigmoid (bool, optional): Whether the prediction is
used for sigmoid or softmax. Defaults to True.
alpha (float, optional): A balance factor for the negative part of
Varifocal Loss, which is different from the alpha of Focal
Loss. Defaults to 0.75.
gamma (float, optional): The gamma for calculating the modulating
factor. Defaults to 2.0.
iou_weighted (bool, optional): Whether to weight the loss of the
positive examples with the iou target. Defaults to True.
reduction (str, optional): The method used to reduce the loss into
a scalar. Defaults to 'mean'. Options are "none", "mean" and
"sum".
loss_weight (float, optional): Weight of loss. Defaults to 1.0.
"""
super(VarifocalLoss, self).__init__()
assert use_sigmoid is True, 'Only sigmoid varifocal loss supported now.'
assert alpha >= 0.0
self.use_sigmoid = use_sigmoid
self.alpha = alpha
self.gamma = gamma
self.iou_weighted = iou_weighted
self.reduction = reduction
self.loss_weight = loss_weight
def forward(self, pred, target, weight=None, avg_factor=None,
reduction_override=None):
"""Forward function.
Args:
pred (torch.Tensor): The prediction.
target (torch.Tensor): The learning target of the prediction.
weight (torch.Tensor, optional): The weight of loss for each
prediction. Defaults to None.
avg_factor (int, optional): Average factor that is used to average
the loss. Defaults to None.
reduction_override (str, optional): The reduction method used to
override the original reduction method of the loss.
Options are "none", "mean" and "sum".
Returns:
torch.Tensor: The calculated loss
"""
assert reduction_override in (None, 'none', 'mean', 'sum')
reduction = (reduction_override if reduction_override else self.
reduction)
if self.use_sigmoid:
loss_cls = self.loss_weight * varifocal_loss(pred, target,
weight, alpha=self.alpha, gamma=self.gamma, iou_weighted=
self.iou_weighted, reduction=reduction, avg_factor=avg_factor)
else:
raise NotImplementedError
return loss_cls
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn.functional as F
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused__to_copy_abs_add_binary_cross_entropy_with_logits_gt_le_mean_mul_pow_sigmoid_sub_0(
in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp3 = tl.load(in_ptr1 + r0, None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp2 * tmp3
tmp5 = 0.0
tmp6 = triton_helpers.minimum(tmp5, tmp3)
tmp7 = tl_math.abs(tmp3)
tmp8 = -tmp7
tmp9 = tl_math.exp(tmp8)
tmp10 = libdevice.log1p(tmp9)
tmp11 = tmp6 - tmp10
tmp12 = tmp4 - tmp11
tmp13 = tmp0 > tmp5
tmp14 = tmp13.to(tl.float32)
tmp15 = tmp0 * tmp14
tmp16 = tl.sigmoid(tmp3)
tmp17 = tmp16 - tmp0
tmp18 = tl_math.abs(tmp17)
tmp19 = tmp18 * tmp18
tmp20 = 0.75
tmp21 = tmp19 * tmp20
tmp22 = tmp0 <= tmp5
tmp23 = tmp22.to(tl.float32)
tmp24 = tmp21 * tmp23
tmp25 = tmp15 + tmp24
tmp26 = tmp12 * tmp25
tmp27 = tl.broadcast_to(tmp26, [RBLOCK])
tmp29 = triton_helpers.promote_to_tensor(tl.sum(tmp27, 0))
tmp30 = 256.0
tmp31 = tmp29 / tmp30
tmp32 = tmp31 * tmp1
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp32, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused__to_copy_abs_add_binary_cross_entropy_with_logits_gt_le_mean_mul_pow_sigmoid_sub_0[
grid(1)](buf1, arg1_1, arg0_1, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf1,
def reduce_loss(loss, reduction):
"""Reduce loss as specified.
Args:
loss (Tensor): Elementwise loss tensor.
reduction (str): Options are "none", "mean" and "sum".
Return:
Tensor: Reduced loss tensor.
"""
reduction_enum = F._Reduction.get_enum(reduction)
if reduction_enum == 0:
return loss
elif reduction_enum == 1:
return loss.mean()
elif reduction_enum == 2:
return loss.sum()
def weight_reduce_loss(loss, weight=None, reduction='mean', avg_factor=None):
"""Apply element-wise weight and reduce loss.
Args:
loss (Tensor): Element-wise loss.
weight (Tensor): Element-wise weights.
reduction (str): Same as built-in losses of PyTorch.
avg_factor (float): Avarage factor when computing the mean of losses.
Returns:
Tensor: Processed loss values.
"""
if weight is not None:
loss = loss * weight
if avg_factor is None:
loss = reduce_loss(loss, reduction)
elif reduction == 'mean':
loss = loss.sum() / avg_factor
elif reduction != 'none':
raise ValueError('avg_factor can not be used with reduction="sum"')
return loss
def varifocal_loss(pred, target, weight=None, alpha=0.75, gamma=2.0,
iou_weighted=True, reduction='mean', avg_factor=None):
"""`Varifocal Loss <https://arxiv.org/abs/2008.13367>`_
Args:
pred (torch.Tensor): The prediction with shape (N, C), C is the
number of classes
target (torch.Tensor): The learning target of the iou-aware
classification score with shape (N, C), C is the number of classes.
weight (torch.Tensor, optional): The weight of loss for each
prediction. Defaults to None.
alpha (float, optional): A balance factor for the negative part of
Varifocal Loss, which is different from the alpha of Focal Loss.
Defaults to 0.75.
gamma (float, optional): The gamma for calculating the modulating
factor. Defaults to 2.0.
iou_weighted (bool, optional): Whether to weight the loss of the
positive example with the iou target. Defaults to True.
reduction (str, optional): The method used to reduce the loss into
a scalar. Defaults to 'mean'. Options are "none", "mean" and
"sum".
avg_factor (int, optional): Average factor that is used to average
the loss. Defaults to None.
"""
assert pred.size() == target.size()
pred_sigmoid = pred.sigmoid()
target = target.type_as(pred)
if iou_weighted:
focal_weight = target * (target > 0.0).float() + alpha * (pred_sigmoid
- target).abs().pow(gamma) * (target <= 0.0).float()
else:
focal_weight = (target > 0.0).float() + alpha * (pred_sigmoid - target
).abs().pow(gamma) * (target <= 0.0).float()
loss = F.binary_cross_entropy_with_logits(pred, target, reduction='none'
) * focal_weight
loss = weight_reduce_loss(loss, weight, reduction, avg_factor)
return loss
class VarifocalLossNew(nn.Module):
def __init__(self, use_sigmoid=True, alpha=0.75, gamma=2.0,
iou_weighted=True, reduction='mean', loss_weight=1.0):
"""`Varifocal Loss <https://arxiv.org/abs/2008.13367>`_
Args:
use_sigmoid (bool, optional): Whether the prediction is
used for sigmoid or softmax. Defaults to True.
alpha (float, optional): A balance factor for the negative part of
Varifocal Loss, which is different from the alpha of Focal
Loss. Defaults to 0.75.
gamma (float, optional): The gamma for calculating the modulating
factor. Defaults to 2.0.
iou_weighted (bool, optional): Whether to weight the loss of the
positive examples with the iou target. Defaults to True.
reduction (str, optional): The method used to reduce the loss into
a scalar. Defaults to 'mean'. Options are "none", "mean" and
"sum".
loss_weight (float, optional): Weight of loss. Defaults to 1.0.
"""
super(VarifocalLossNew, self).__init__()
assert use_sigmoid is True, 'Only sigmoid varifocal loss supported now.'
assert alpha >= 0.0
self.use_sigmoid = use_sigmoid
self.alpha = alpha
self.gamma = gamma
self.iou_weighted = iou_weighted
self.reduction = reduction
self.loss_weight = loss_weight
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
Sundrops/mmdetection
|
VarifocalLoss
| false | 14,447 |
[
"Apache-2.0"
] | 549 |
d3cf38d91c454b1a6881e8c36c1e4a66dc5521b8
|
https://github.com/Sundrops/mmdetection/tree/d3cf38d91c454b1a6881e8c36c1e4a66dc5521b8
|
ContextPooler
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/yy/cyya3js6wt64vdji3sfisvrqyfvqxwkwqq5mzg5bqjl2crzjs4t3.py
# Topologically Sorted Source Nodes: [pooled_output], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# pooled_output => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%select,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = (xindex // 16)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (64*x1)), xmask)
tl.store(out_ptr0 + (x2), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/cq/ccqt6ier6syhskiutqqh3rk35u6wmkgcqhhpnliua5yw6a6ivnch.py
# Topologically Sorted Source Nodes: [pooled_output, mul, pow_1, mul_1, add, mul_2, tanh, add_1, pooled_output_1], Original ATen: [aten.add, aten.mul, aten.pow, aten.tanh]
# Source node to ATen node mapping:
# add => add_1
# add_1 => add_2
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# pooled_output => add
# pooled_output_1 => mul_3
# pow_1 => pow_1
# tanh => tanh
# Graph fragment:
# %add : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_1, %primals_3), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add, 0.5), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%add, 3.0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%pow_1, 0.044715), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add, %mul_1), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_1, 0.7978845608028654), kwargs = {})
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%mul_2,), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%tanh, 1.0), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %add_2), kwargs = {})
triton_poi_fused_add_mul_pow_tanh_1 = async_compile.triton('triton_poi_fused_add_mul_pow_tanh_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_pow_tanh_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_pow_tanh_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp5 = tmp2 * tmp2
tmp6 = tmp5 * tmp2
tmp7 = 0.044715
tmp8 = tmp6 * tmp7
tmp9 = tmp2 + tmp8
tmp10 = 0.7978845608028654
tmp11 = tmp9 * tmp10
tmp12 = libdevice.tanh(tmp11)
tmp13 = 1.0
tmp14 = tmp12 + tmp13
tmp15 = tmp4 * tmp14
tl.store(out_ptr0 + (x2), tmp15, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pooled_output], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(primals_1, buf0, 64, grid=grid(64), stream=stream0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pooled_output], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (16, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf1)
del primals_2
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pooled_output, mul, pow_1, mul_1, add, mul_2, tanh, add_1, pooled_output_1], Original ATen: [aten.add, aten.mul, aten.pow, aten.tanh]
triton_poi_fused_add_mul_pow_tanh_1.run(buf1, primals_3, buf2, 64, grid=grid(64), stream=stream0)
return (buf2, primals_3, reinterpret_tensor(buf0, (16, 4), (4, 1), 0), buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
from _paritybench_helpers import _mock_config
import math
import torch
from torch import nn
def get_mask(input, local_context):
if not isinstance(local_context, DropoutContext):
dropout = local_context
mask = None
else:
dropout = local_context.dropout
dropout *= local_context.scale
mask = local_context.mask if local_context.reuse_mask else None
if dropout > 0 and mask is None:
mask = (1 - torch.empty_like(input).bernoulli_(1 - dropout)).bool()
if isinstance(local_context, DropoutContext):
if local_context.mask is None:
local_context.mask = mask
return mask, dropout
def gelu(x):
"""
Implementation of the GELU activation function currently in Google BERT repo (identical to OpenAI GPT). Also see
the Gaussian Error Linear Units paper: https://arxiv.org/abs/1606.08415
"""
return 0.5 * x * (1.0 + torch.tanh(math.sqrt(2.0 / math.pi) * (x +
0.044715 * torch.pow(x, 3.0))))
class DropoutContext(object):
def __init__(self):
self.dropout = 0
self.mask = None
self.scale = 1
self.reuse_mask = True
class XDropout(torch.autograd.Function):
"""Optimized dropout function to save computation and memory by using mask operation instead of multiplication."""
@staticmethod
def forward(ctx, input, local_ctx):
mask, dropout = get_mask(input, local_ctx)
ctx.scale = 1.0 / (1 - dropout)
if dropout > 0:
ctx.save_for_backward(mask)
return input.masked_fill(mask, 0) * ctx.scale
else:
return input
@staticmethod
def backward(ctx, grad_output):
if ctx.scale > 1:
mask, = ctx.saved_tensors
return grad_output.masked_fill(mask, 0) * ctx.scale, None
else:
return grad_output, None
class StableDropout(torch.nn.Module):
"""
Optimized dropout module for stabilizing the training
Args:
drop_prob (float): the dropout probabilities
"""
def __init__(self, drop_prob):
super().__init__()
self.drop_prob = drop_prob
self.count = 0
self.context_stack = None
def forward(self, x):
"""
Call the module
Args:
x (:obj:`torch.tensor`): The input tensor to apply dropout
"""
if self.training and self.drop_prob > 0:
return XDropout.apply(x, self.get_context())
return x
def clear_context(self):
self.count = 0
self.context_stack = None
def init_context(self, reuse_mask=True, scale=1):
if self.context_stack is None:
self.context_stack = []
self.count = 0
for c in self.context_stack:
c.reuse_mask = reuse_mask
c.scale = scale
def get_context(self):
if self.context_stack is not None:
if self.count >= len(self.context_stack):
self.context_stack.append(DropoutContext())
ctx = self.context_stack[self.count]
ctx.dropout = self.drop_prob
self.count += 1
return ctx
else:
return self.drop_prob
class ContextPooler(nn.Module):
def __init__(self, config):
super().__init__()
try:
pooler_hidden_size = config.pooler_hidden_size
pooler_dropout = config.pooler_dropout
except AttributeError:
pooler_hidden_size = config.hidden_size
pooler_dropout = 0.1
self.dense = nn.Linear(pooler_hidden_size, pooler_hidden_size)
self.dropout = StableDropout(pooler_dropout)
self.config = config
def forward(self, hidden_states):
context_token = hidden_states[:, 0]
context_token = self.dropout(context_token)
pooled_output = self.dense(context_token)
pooled_output = gelu(pooled_output)
return pooled_output
@property
def output_dim(self):
return self.config.hidden_size
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'config': _mock_config(pooler_hidden_size=4,
pooler_dropout=0.5, hidden_size=4)}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import math
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = xindex // 16
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 64 * x1), xmask)
tl.store(out_ptr0 + x2, tmp0, xmask)
@triton.jit
def triton_poi_fused_add_mul_pow_tanh_1(in_ptr0, in_ptr1, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp5 = tmp2 * tmp2
tmp6 = tmp5 * tmp2
tmp7 = 0.044715
tmp8 = tmp6 * tmp7
tmp9 = tmp2 + tmp8
tmp10 = 0.7978845608028654
tmp11 = tmp9 * tmp10
tmp12 = libdevice.tanh(tmp11)
tmp13 = 1.0
tmp14 = tmp12 + tmp13
tmp15 = tmp4 * tmp14
tl.store(out_ptr0 + x2, tmp15, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(64)](primals_1, buf0, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf1)
del primals_2
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_add_mul_pow_tanh_1[grid(64)](buf1, primals_3, buf2,
64, XBLOCK=64, num_warps=1, num_stages=1)
return buf2, primals_3, reinterpret_tensor(buf0, (16, 4), (4, 1), 0), buf1
def get_mask(input, local_context):
if not isinstance(local_context, DropoutContext):
dropout = local_context
mask = None
else:
dropout = local_context.dropout
dropout *= local_context.scale
mask = local_context.mask if local_context.reuse_mask else None
if dropout > 0 and mask is None:
mask = (1 - torch.empty_like(input).bernoulli_(1 - dropout)).bool()
if isinstance(local_context, DropoutContext):
if local_context.mask is None:
local_context.mask = mask
return mask, dropout
def gelu(x):
"""
Implementation of the GELU activation function currently in Google BERT repo (identical to OpenAI GPT). Also see
the Gaussian Error Linear Units paper: https://arxiv.org/abs/1606.08415
"""
return 0.5 * x * (1.0 + torch.tanh(math.sqrt(2.0 / math.pi) * (x +
0.044715 * torch.pow(x, 3.0))))
class DropoutContext(object):
def __init__(self):
self.dropout = 0
self.mask = None
self.scale = 1
self.reuse_mask = True
class XDropout(torch.autograd.Function):
"""Optimized dropout function to save computation and memory by using mask operation instead of multiplication."""
@staticmethod
def forward(ctx, input, local_ctx):
mask, dropout = get_mask(input, local_ctx)
ctx.scale = 1.0 / (1 - dropout)
if dropout > 0:
ctx.save_for_backward(mask)
return input.masked_fill(mask, 0) * ctx.scale
else:
return input
@staticmethod
def backward(ctx, grad_output):
if ctx.scale > 1:
mask, = ctx.saved_tensors
return grad_output.masked_fill(mask, 0) * ctx.scale, None
else:
return grad_output, None
class StableDropout(torch.nn.Module):
"""
Optimized dropout module for stabilizing the training
Args:
drop_prob (float): the dropout probabilities
"""
def __init__(self, drop_prob):
super().__init__()
self.drop_prob = drop_prob
self.count = 0
self.context_stack = None
def forward(self, x):
"""
Call the module
Args:
x (:obj:`torch.tensor`): The input tensor to apply dropout
"""
if self.training and self.drop_prob > 0:
return XDropout.apply(x, self.get_context())
return x
def clear_context(self):
self.count = 0
self.context_stack = None
def init_context(self, reuse_mask=True, scale=1):
if self.context_stack is None:
self.context_stack = []
self.count = 0
for c in self.context_stack:
c.reuse_mask = reuse_mask
c.scale = scale
def get_context(self):
if self.context_stack is not None:
if self.count >= len(self.context_stack):
self.context_stack.append(DropoutContext())
ctx = self.context_stack[self.count]
ctx.dropout = self.drop_prob
self.count += 1
return ctx
else:
return self.drop_prob
class ContextPoolerNew(nn.Module):
def __init__(self, config):
super().__init__()
try:
pooler_hidden_size = config.pooler_hidden_size
pooler_dropout = config.pooler_dropout
except AttributeError:
pooler_hidden_size = config.hidden_size
pooler_dropout = 0.1
self.dense = nn.Linear(pooler_hidden_size, pooler_hidden_size)
self.dropout = StableDropout(pooler_dropout)
self.config = config
@property
def output_dim(self):
return self.config.hidden_size
def forward(self, input_0):
primals_2 = self.dense.weight
primals_3 = self.dense.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Stochastic-Adventure/ClinicalTransformerRelationExtraction
|
ContextPooler
| false | 14,448 |
[
"MIT"
] | 78 |
eef956bbfbd64b008014ef7cac5f818087816725
|
https://github.com/Stochastic-Adventure/ClinicalTransformerRelationExtraction/tree/eef956bbfbd64b008014ef7cac5f818087816725
|
MockAccuracy
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/je/cje5jcdwoopxedyiqqsslxfmfdsyzibxowjfk27uklptqpxyb4lo.py
# Topologically Sorted Source Nodes: [long, long_1, bool_acc, sum_1, truediv], Original ATen: [aten._to_copy, aten.eq, aten.sum, aten.div]
# Source node to ATen node mapping:
# bool_acc => eq
# long => convert_element_type
# long_1 => convert_element_type_1
# sum_1 => sum_1
# truediv => div
# Graph fragment:
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%arg0_1, torch.int64), kwargs = {})
# %convert_element_type_1 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%arg1_1, torch.int64), kwargs = {})
# %eq : [num_users=1] = call_function[target=torch.ops.aten.eq.Tensor](args = (%convert_element_type, %convert_element_type_1), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%eq,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sum_1, 256), kwargs = {})
triton_per_fused__to_copy_div_eq_sum_0 = async_compile.triton('triton_per_fused__to_copy_div_eq_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__to_copy_div_eq_sum_0', 'mutated_arg_names': [], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__to_copy_div_eq_sum_0(in_ptr0, in_ptr1, out_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp2 = tl.load(in_ptr1 + (r0), None)
tmp1 = tmp0.to(tl.int64)
tmp3 = tmp2.to(tl.int64)
tmp4 = tmp1 == tmp3
tmp5 = tmp4.to(tl.int64)
tmp6 = tl.broadcast_to(tmp5, [RBLOCK])
tmp8 = triton_helpers.promote_to_tensor(tl.sum(tmp6, 0))
tmp9 = tmp8.to(tl.float32)
tmp10 = 0.00390625
tmp11 = tmp9 * tmp10
tl.store(out_ptr1 + (tl.full([1], 0, tl.int32)), tmp11, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
# Topologically Sorted Source Nodes: [long, long_1, bool_acc, sum_1, truediv], Original ATen: [aten._to_copy, aten.eq, aten.sum, aten.div]
stream0 = get_raw_stream(0)
triton_per_fused__to_copy_div_eq_sum_0.run(arg0_1, arg1_1, buf1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
class _Metric(torch.nn.Module):
def __init__(self):
super().__init__()
def forward(self, input: 'torch.Tensor', target: 'torch.Tensor'):
raise NotImplementedError()
class Accuracy(_Metric):
def __init__(self):
super().__init__()
def forward(self, input: 'torch.Tensor', target: 'torch.Tensor'):
"""
:param input: [B, L]
:param target: [B, L]
:return:
"""
bool_acc = input.long() == target.long()
return bool_acc.sum() / bool_acc.numel()
class MockAccuracy(Accuracy):
def __init__(self):
super().__init__()
def forward(self, input: 'torch.Tensor', target: 'torch.Tensor'):
return super().forward(input, target)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused__to_copy_div_eq_sum_0(in_ptr0, in_ptr1, out_ptr1,
xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp2 = tl.load(in_ptr1 + r0, None)
tmp1 = tmp0.to(tl.int64)
tmp3 = tmp2.to(tl.int64)
tmp4 = tmp1 == tmp3
tmp5 = tmp4.to(tl.int64)
tmp6 = tl.broadcast_to(tmp5, [RBLOCK])
tmp8 = triton_helpers.promote_to_tensor(tl.sum(tmp6, 0))
tmp9 = tmp8.to(tl.float32)
tmp10 = 0.00390625
tmp11 = tmp9 * tmp10
tl.store(out_ptr1 + tl.full([1], 0, tl.int32), tmp11, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
get_raw_stream(0)
triton_per_fused__to_copy_div_eq_sum_0[grid(1)](arg0_1, arg1_1,
buf1, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf1,
class _Metric(torch.nn.Module):
def __init__(self):
super().__init__()
def forward(self, input: 'torch.Tensor', target: 'torch.Tensor'):
raise NotImplementedError()
class Accuracy(_Metric):
def __init__(self):
super().__init__()
def forward(self, input: 'torch.Tensor', target: 'torch.Tensor'):
"""
:param input: [B, L]
:param target: [B, L]
:return:
"""
bool_acc = input.long() == target.long()
return bool_acc.sum() / bool_acc.numel()
class MockAccuracyNew(Accuracy):
def __init__(self):
super().__init__()
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
Stillerman/MusicTransformer-pytorch
|
MockAccuracy
| false | 14,449 |
[
"MIT"
] | 170 |
73abb7cab271beba042b7b6fc06a6a9aaee82e8c
|
https://github.com/Stillerman/MusicTransformer-pytorch/tree/73abb7cab271beba042b7b6fc06a6a9aaee82e8c
|
ConvolutionBlock
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/6p/c6pufjuk3umqhiodebvlshdigugbnw5bywi3x3sbs3xphvcmccuv.py
# Topologically Sorted Source Nodes: [outputs], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# outputs => gt, mul, where
# Graph fragment:
# %gt : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%primals_1, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_1, 0.2), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %primals_1, %mul), kwargs = {})
triton_poi_fused_leaky_relu_0 = async_compile.triton('triton_poi_fused_leaky_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.0
tmp2 = tmp0 > tmp1
tmp3 = 0.2
tmp4 = tmp0 * tmp3
tmp5 = tl.where(tmp2, tmp0, tmp4)
tl.store(out_ptr0 + (x0), tmp5, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/2m/c2mt7vvxypcyg4roj4r3bns7fqblbouce2ybektelf7rsc62boym.py
# Topologically Sorted Source Nodes: [outputs_1], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# outputs_1 => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%unsqueeze, %primals_2, %primals_3, [1], [1], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 3), (12, 3, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [outputs], Original ATen: [aten.leaky_relu]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_0.run(primals_1, buf0, 16, grid=grid(16), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [outputs_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(reinterpret_tensor(buf0, (1, 4, 4), (0, 4, 1), 0), primals_2, stride=(1,), padding=(1,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf1, (1, 4, 4), (16, 4, 1))
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [outputs_1], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf2, primals_3, 16, grid=grid(16), stream=stream0)
del primals_3
return (reinterpret_tensor(buf2, (4, 4), (4, 1), 0), primals_2, reinterpret_tensor(buf0, (1, 4, 4), (16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 3), (12, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
class BaseModule(torch.nn.Module):
def __init__(self):
super(BaseModule, self).__init__()
@property
def nparams(self):
return sum(p.numel() for p in self.parameters() if p.requires_grad)
class Conv1dWithInitialization(BaseModule):
def __init__(self, **kwargs):
super(Conv1dWithInitialization, self).__init__()
self.conv1d = torch.nn.Conv1d(**kwargs)
torch.nn.init.orthogonal_(self.conv1d.weight.data, gain=1)
def forward(self, x):
return self.conv1d(x)
class ConvolutionBlock(BaseModule):
def __init__(self, in_channels, out_channels, dilation):
super(ConvolutionBlock, self).__init__()
self.leaky_relu = torch.nn.LeakyReLU(0.2)
self.convolution = Conv1dWithInitialization(in_channels=in_channels,
out_channels=out_channels, kernel_size=3, stride=1, padding=
dilation, dilation=dilation)
def forward(self, x):
outputs = self.leaky_relu(x)
outputs = self.convolution(outputs)
return outputs
def get_inputs():
return [torch.rand([4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'dilation': 1}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.0
tmp2 = tmp0 > tmp1
tmp3 = 0.2
tmp4 = tmp0 * tmp3
tmp5 = tl.where(tmp2, tmp0, tmp4)
tl.store(out_ptr0 + x0, tmp5, xmask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 3), (12, 3, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_leaky_relu_0[grid(16)](primals_1, buf0, 16, XBLOCK
=16, num_warps=1, num_stages=1)
del primals_1
buf1 = extern_kernels.convolution(reinterpret_tensor(buf0, (1, 4, 4
), (0, 4, 1), 0), primals_2, stride=(1,), padding=(1,),
dilation=(1,), transposed=False, output_padding=(0,), groups=1,
bias=None)
assert_size_stride(buf1, (1, 4, 4), (16, 4, 1))
buf2 = buf1
del buf1
triton_poi_fused_convolution_1[grid(16)](buf2, primals_3, 16,
XBLOCK=16, num_warps=1, num_stages=1)
del primals_3
return reinterpret_tensor(buf2, (4, 4), (4, 1), 0
), primals_2, reinterpret_tensor(buf0, (1, 4, 4), (16, 4, 1), 0)
class BaseModule(torch.nn.Module):
def __init__(self):
super(BaseModule, self).__init__()
@property
def nparams(self):
return sum(p.numel() for p in self.parameters() if p.requires_grad)
class Conv1dWithInitialization(BaseModule):
def __init__(self, **kwargs):
super(Conv1dWithInitialization, self).__init__()
self.conv1d = torch.nn.Conv1d(**kwargs)
torch.nn.init.orthogonal_(self.conv1d.weight.data, gain=1)
def forward(self, x):
return self.conv1d(x)
class ConvolutionBlockNew(BaseModule):
def __init__(self, in_channels, out_channels, dilation):
super(ConvolutionBlockNew, self).__init__()
self.leaky_relu = torch.nn.LeakyReLU(0.2)
self.convolution = Conv1dWithInitialization(in_channels=in_channels,
out_channels=out_channels, kernel_size=3, stride=1, padding=
dilation, dilation=dilation)
def forward(self, input_0):
primals_2 = self.convolution.conv1d.weight
primals_3 = self.convolution.conv1d.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Seungwoo0326/WaveGrad2-1
|
ConvolutionBlock
| false | 14,450 |
[
"MIT"
] | 45 |
3b202201348449b89353f28bce1596ca7939a810
|
https://github.com/Seungwoo0326/WaveGrad2-1/tree/3b202201348449b89353f28bce1596ca7939a810
|
ClassificationCircleLoss
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/fn/cfnbjifotifsbkjqwcrjvmwysi2u65kyzrt5izbv7n7pgfrnvakl.py
# Topologically Sorted Source Nodes: [mask, sub, positive_weighting, mul, sub_2, mul_1, sub_1, add, negative_weighting, mul_2, sub_3, sub_4, mul_3, logits, loss], Original ATen: [aten.scatter, aten.rsub, aten.clamp, aten.mul, aten.sub, aten.add, aten.where, aten._log_softmax]
# Source node to ATen node mapping:
# add => add
# logits => where
# loss => amax, exp, sub_5, sum_1
# mask => scatter_upon_const_tensor
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# mul_3 => mul_3
# negative_weighting => clamp_min_1
# positive_weighting => clamp_min
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# sub_3 => sub_3
# sub_4 => sub_4
# Graph fragment:
# %scatter_upon_const_tensor : [num_users=1] = call_function[target=torch._inductor.fx_passes.post_grad.scatter_upon_const_tensor](args = (), kwargs = {shape: [4, 4], background_val: False, dtype: torch.bool, dim: 1, selector: %unsqueeze, val: 1})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1.25, %arg0_1), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%clamp_min, 256.0), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, 0.75), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %sub_2), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, 0), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sub_1, 0.25), kwargs = {})
# %clamp_min_1 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add, 0), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%clamp_min_1, 256.0), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, 0), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub_3, 0.25), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, %sub_4), kwargs = {})
# %where : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%scatter_upon_const_tensor, %mul_1, %mul_3), kwargs = {})
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%where, [1], True), kwargs = {})
# %sub_5 : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where, %amax), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub_5,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_0 = async_compile.triton('triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: '*i64', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp6 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp48 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp67 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 == tmp1
tmp3 = tl.full([1], True, tl.int1)
tmp4 = tl.full([1], False, tl.int1)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp7 = tmp6.to(tl.float32)
tmp8 = 1.25
tmp9 = tmp8 - tmp7
tmp10 = 0.0
tmp11 = triton_helpers.maximum(tmp9, tmp10)
tmp12 = 256.0
tmp13 = tmp11 * tmp12
tmp14 = 0.75
tmp15 = tmp7 - tmp14
tmp16 = tmp13 * tmp15
tmp17 = tmp6 - tmp1
tmp18 = tmp17.to(tl.float32)
tmp19 = 0.25
tmp20 = tmp18 + tmp19
tmp21 = triton_helpers.maximum(tmp20, tmp10)
tmp22 = tmp21 * tmp12
tmp23 = tmp18 - tmp19
tmp24 = tmp22 * tmp23
tmp25 = tl.where(tmp5, tmp16, tmp24)
tmp26 = tl.full([1], 1, tl.int64)
tmp27 = tmp0 == tmp26
tmp28 = tl.where(tmp27, tmp3, tmp4)
tmp30 = tmp29.to(tl.float32)
tmp31 = tmp8 - tmp30
tmp32 = triton_helpers.maximum(tmp31, tmp10)
tmp33 = tmp32 * tmp12
tmp34 = tmp30 - tmp14
tmp35 = tmp33 * tmp34
tmp36 = tmp29 - tmp1
tmp37 = tmp36.to(tl.float32)
tmp38 = tmp37 + tmp19
tmp39 = triton_helpers.maximum(tmp38, tmp10)
tmp40 = tmp39 * tmp12
tmp41 = tmp37 - tmp19
tmp42 = tmp40 * tmp41
tmp43 = tl.where(tmp28, tmp35, tmp42)
tmp44 = triton_helpers.maximum(tmp25, tmp43)
tmp45 = tl.full([1], 2, tl.int64)
tmp46 = tmp0 == tmp45
tmp47 = tl.where(tmp46, tmp3, tmp4)
tmp49 = tmp48.to(tl.float32)
tmp50 = tmp8 - tmp49
tmp51 = triton_helpers.maximum(tmp50, tmp10)
tmp52 = tmp51 * tmp12
tmp53 = tmp49 - tmp14
tmp54 = tmp52 * tmp53
tmp55 = tmp48 - tmp1
tmp56 = tmp55.to(tl.float32)
tmp57 = tmp56 + tmp19
tmp58 = triton_helpers.maximum(tmp57, tmp10)
tmp59 = tmp58 * tmp12
tmp60 = tmp56 - tmp19
tmp61 = tmp59 * tmp60
tmp62 = tl.where(tmp47, tmp54, tmp61)
tmp63 = triton_helpers.maximum(tmp44, tmp62)
tmp64 = tl.full([1], 3, tl.int64)
tmp65 = tmp0 == tmp64
tmp66 = tl.where(tmp65, tmp3, tmp4)
tmp68 = tmp67.to(tl.float32)
tmp69 = tmp8 - tmp68
tmp70 = triton_helpers.maximum(tmp69, tmp10)
tmp71 = tmp70 * tmp12
tmp72 = tmp68 - tmp14
tmp73 = tmp71 * tmp72
tmp74 = tmp67 - tmp1
tmp75 = tmp74.to(tl.float32)
tmp76 = tmp75 + tmp19
tmp77 = triton_helpers.maximum(tmp76, tmp10)
tmp78 = tmp77 * tmp12
tmp79 = tmp75 - tmp19
tmp80 = tmp78 * tmp79
tmp81 = tl.where(tmp66, tmp73, tmp80)
tmp82 = triton_helpers.maximum(tmp63, tmp81)
tmp83 = tmp25 - tmp82
tmp84 = tl_math.exp(tmp83)
tmp85 = tmp43 - tmp82
tmp86 = tl_math.exp(tmp85)
tmp87 = tmp84 + tmp86
tmp88 = tmp62 - tmp82
tmp89 = tl_math.exp(tmp88)
tmp90 = tmp87 + tmp89
tmp91 = tmp81 - tmp82
tmp92 = tl_math.exp(tmp91)
tmp93 = tmp90 + tmp92
tl.store(out_ptr0 + (x0), tmp82, xmask)
tl.store(out_ptr1 + (x0), tmp93, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/f7/cf76gdye3s6mqx3jjsxgcjijmfytmkiiff5dl6dyksihxbvh3muu.py
# Topologically Sorted Source Nodes: [mask, sub, positive_weighting, mul, sub_2, mul_1, sub_1, add, negative_weighting, mul_2, sub_3, sub_4, mul_3, logits, loss], Original ATen: [aten.scatter, aten.rsub, aten.clamp, aten.mul, aten.sub, aten.add, aten.where, aten._log_softmax]
# Source node to ATen node mapping:
# add => add
# logits => where
# loss => log, sub_5, sub_6
# mask => scatter_upon_const_tensor
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# mul_3 => mul_3
# negative_weighting => clamp_min_1
# positive_weighting => clamp_min
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# sub_3 => sub_3
# sub_4 => sub_4
# Graph fragment:
# %scatter_upon_const_tensor : [num_users=1] = call_function[target=torch._inductor.fx_passes.post_grad.scatter_upon_const_tensor](args = (), kwargs = {shape: [4, 4], background_val: False, dtype: torch.bool, dim: 1, selector: %unsqueeze, val: 1})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1.25, %arg0_1), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%clamp_min, 256.0), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, 0.75), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %sub_2), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, 0), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sub_1, 0.25), kwargs = {})
# %clamp_min_1 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add, 0), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%clamp_min_1, 256.0), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, 0), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub_3, 0.25), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, %sub_4), kwargs = {})
# %where : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%scatter_upon_const_tensor, %mul_1, %mul_3), kwargs = {})
# %sub_5 : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where, %amax), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_6 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub_5, %log), kwargs = {})
triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_1 = async_compile.triton('triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: '*i64', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4)
x0 = xindex % 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + (x2), xmask)
tmp27 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp1 = x0
tmp2 = tmp0 == tmp1
tmp3 = tl.full([1], True, tl.int1)
tmp4 = tl.full([1], False, tl.int1)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp7 = tmp6.to(tl.float32)
tmp8 = 1.25
tmp9 = tmp8 - tmp7
tmp10 = 0.0
tmp11 = triton_helpers.maximum(tmp9, tmp10)
tmp12 = 256.0
tmp13 = tmp11 * tmp12
tmp14 = 0.75
tmp15 = tmp7 - tmp14
tmp16 = tmp13 * tmp15
tmp17 = tl.full([1], 0, tl.int64)
tmp18 = tmp6 - tmp17
tmp19 = tmp18.to(tl.float32)
tmp20 = 0.25
tmp21 = tmp19 + tmp20
tmp22 = triton_helpers.maximum(tmp21, tmp10)
tmp23 = tmp22 * tmp12
tmp24 = tmp19 - tmp20
tmp25 = tmp23 * tmp24
tmp26 = tl.where(tmp5, tmp16, tmp25)
tmp28 = tmp26 - tmp27
tmp30 = tl_math.log(tmp29)
tmp31 = tmp28 - tmp30
tl.store(out_ptr0 + (x2), tmp31, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/bz/cbzq7t26pihe3ec3rnmx4wscu3agefng6kzj2qs5lh462cc5t6ij.py
# Topologically Sorted Source Nodes: [loss], Original ATen: [aten.nll_loss_forward]
# Source node to ATen node mapping:
# loss => convert_element_type, div, full_default_2, ne_1, ne_2, neg, sum_2, sum_3, where_2
# Graph fragment:
# %ne_1 : [num_users=1] = call_function[target=torch.ops.aten.ne.Scalar](args = (%arg1_1, -100), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%squeeze,), kwargs = {})
# %full_default_2 : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where_2 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%ne_1, %neg, %full_default_2), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%where_2,), kwargs = {})
# %ne_2 : [num_users=1] = call_function[target=torch.ops.aten.ne.Scalar](args = (%arg1_1, -100), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%ne_2,), kwargs = {})
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%sum_2, torch.float32), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sum_3, %convert_element_type), kwargs = {})
triton_per_fused_nll_loss_forward_2 = async_compile.triton('triton_per_fused_nll_loss_forward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 4],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i64', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_nll_loss_forward_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 2, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_nll_loss_forward_2(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 4
RBLOCK: tl.constexpr = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = tl.full([1, 1], -100, tl.int64)
tmp2 = tmp0 != tmp1
tmp3 = tl.full([1, 1], 0, tl.int64)
tmp4 = tl.where(tmp2, tmp0, tmp3)
tmp5 = tl.full([XBLOCK, RBLOCK], 4, tl.int32)
tmp6 = tmp4 + tmp5
tmp7 = tmp4 < 0
tmp8 = tl.where(tmp7, tmp6, tmp4)
tl.device_assert((0 <= tmp8) & (tmp8 < 4), "index out of bounds: 0 <= tmp8 < 4")
tmp10 = tl.load(in_ptr1 + (tmp8 + (4*r0)), None, eviction_policy='evict_last')
tmp11 = -tmp10
tmp12 = 0.0
tmp13 = tl.where(tmp2, tmp11, tmp12)
tmp14 = tl.broadcast_to(tmp13, [XBLOCK, RBLOCK])
tmp16 = tl.sum(tmp14, 1)[:, None]
tmp17 = tmp2.to(tl.int64)
tmp18 = tl.broadcast_to(tmp17, [XBLOCK, RBLOCK])
tmp20 = tl.sum(tmp18, 1)[:, None]
tmp21 = tmp20.to(tl.float32)
tmp22 = tmp16 / tmp21
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp22, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4), (4, 1))
assert_size_stride(arg1_1, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf1 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
# Topologically Sorted Source Nodes: [mask, sub, positive_weighting, mul, sub_2, mul_1, sub_1, add, negative_weighting, mul_2, sub_3, sub_4, mul_3, logits, loss], Original ATen: [aten.scatter, aten.rsub, aten.clamp, aten.mul, aten.sub, aten.add, aten.where, aten._log_softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_0.run(arg1_1, arg0_1, buf0, buf1, 4, grid=grid(4), stream=stream0)
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mask, sub, positive_weighting, mul, sub_2, mul_1, sub_1, add, negative_weighting, mul_2, sub_3, sub_4, mul_3, logits, loss], Original ATen: [aten.scatter, aten.rsub, aten.clamp, aten.mul, aten.sub, aten.add, aten.where, aten._log_softmax]
triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_1.run(arg1_1, arg0_1, buf0, buf1, buf2, 16, grid=grid(16), stream=stream0)
del arg0_1
del buf0
del buf1
buf3 = empty_strided_cuda((), (), torch.float32)
buf5 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [loss], Original ATen: [aten.nll_loss_forward]
triton_per_fused_nll_loss_forward_2.run(buf5, arg1_1, buf2, 1, 4, grid=grid(1), stream=stream0)
del arg1_1
del buf2
return (buf5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.int64)
arg1_1 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.int64)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.utils.data
from typing import Tuple
from torch.nn.functional import cross_entropy
from itertools import product as product
from math import sqrt as sqrt
class ClassificationCircleLoss(nn.Module):
"""Circle loss for class-level labels as described in the paper
`"Circle Loss: A Unified Perspective of Pair Similarity Optimization" <#>`_
Args:
scale (float): the scale factor. Default: 256.0
margin (float): the relax margin value. Default: 0.25
circle_center (tuple[float]): the center of the circle (logit_ap, logit_an). Default: (1, 0)
reduction (string, optional): Specifies the reduction to apply to the output:
``'none'`` | ``'mean'`` | ``'sum'``. ``'none'``: no reduction will be applied,
``'mean'``: the sum of the output will be divided by the number of
elements in the output, ``'sum'``: the output will be summed. Default: ``'mean'``
"""
def __init__(self, scale: 'float'=256.0, margin: 'float'=0.25,
circle_center: 'Tuple[float, float]'=(1, 0), reduction: 'str'='mean'
) ->None:
super(ClassificationCircleLoss, self).__init__()
self.scale = scale
self.margin = margin
self.circle_center = circle_center
self.reduction = reduction
def forward(self, logits: 'torch.Tensor', targets: 'torch.LongTensor'
) ->torch.Tensor:
"""
Args:
logits (torch.Tensor): The predicted logits before softmax,
namely :math:`\\cos \\theta` in the above equation, with shape of :math:`(N, C)`
targets (torch.LongTensor): The ground-truth label long vector,
namely :math:`y` in the above equation, with shape of :math:`(N,)`
Returns:
torch.Tensor: loss
the computed loss
"""
mask = torch.zeros(logits.shape, dtype=torch.bool, device=logits.device
).scatter_(dim=1, index=targets.unsqueeze(1), value=1)
positive_weighting = torch.clamp(self.circle_center[0] + self.
margin - logits.detach(), min=0)
negative_weighting = torch.clamp(logits.detach() - self.
circle_center[1] + self.margin, min=0)
logits = torch.where(mask, self.scale * positive_weighting * (
logits - (self.circle_center[0] - self.margin)), self.scale *
negative_weighting * (logits - self.circle_center[1] - self.margin)
)
loss = cross_entropy(input=logits, target=targets, reduction=self.
reduction)
return loss
def get_inputs():
return [torch.ones([4, 4], dtype=torch.int64), torch.ones([4], dtype=
torch.int64)]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
import torch.utils.data
from typing import Tuple
from itertools import product as product
from math import sqrt as sqrt
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_0(
in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp6 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp48 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp67 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 == tmp1
tmp3 = tl.full([1], True, tl.int1)
tmp4 = tl.full([1], False, tl.int1)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp7 = tmp6.to(tl.float32)
tmp8 = 1.25
tmp9 = tmp8 - tmp7
tmp10 = 0.0
tmp11 = triton_helpers.maximum(tmp9, tmp10)
tmp12 = 256.0
tmp13 = tmp11 * tmp12
tmp14 = 0.75
tmp15 = tmp7 - tmp14
tmp16 = tmp13 * tmp15
tmp17 = tmp6 - tmp1
tmp18 = tmp17.to(tl.float32)
tmp19 = 0.25
tmp20 = tmp18 + tmp19
tmp21 = triton_helpers.maximum(tmp20, tmp10)
tmp22 = tmp21 * tmp12
tmp23 = tmp18 - tmp19
tmp24 = tmp22 * tmp23
tmp25 = tl.where(tmp5, tmp16, tmp24)
tmp26 = tl.full([1], 1, tl.int64)
tmp27 = tmp0 == tmp26
tmp28 = tl.where(tmp27, tmp3, tmp4)
tmp30 = tmp29.to(tl.float32)
tmp31 = tmp8 - tmp30
tmp32 = triton_helpers.maximum(tmp31, tmp10)
tmp33 = tmp32 * tmp12
tmp34 = tmp30 - tmp14
tmp35 = tmp33 * tmp34
tmp36 = tmp29 - tmp1
tmp37 = tmp36.to(tl.float32)
tmp38 = tmp37 + tmp19
tmp39 = triton_helpers.maximum(tmp38, tmp10)
tmp40 = tmp39 * tmp12
tmp41 = tmp37 - tmp19
tmp42 = tmp40 * tmp41
tmp43 = tl.where(tmp28, tmp35, tmp42)
tmp44 = triton_helpers.maximum(tmp25, tmp43)
tmp45 = tl.full([1], 2, tl.int64)
tmp46 = tmp0 == tmp45
tmp47 = tl.where(tmp46, tmp3, tmp4)
tmp49 = tmp48.to(tl.float32)
tmp50 = tmp8 - tmp49
tmp51 = triton_helpers.maximum(tmp50, tmp10)
tmp52 = tmp51 * tmp12
tmp53 = tmp49 - tmp14
tmp54 = tmp52 * tmp53
tmp55 = tmp48 - tmp1
tmp56 = tmp55.to(tl.float32)
tmp57 = tmp56 + tmp19
tmp58 = triton_helpers.maximum(tmp57, tmp10)
tmp59 = tmp58 * tmp12
tmp60 = tmp56 - tmp19
tmp61 = tmp59 * tmp60
tmp62 = tl.where(tmp47, tmp54, tmp61)
tmp63 = triton_helpers.maximum(tmp44, tmp62)
tmp64 = tl.full([1], 3, tl.int64)
tmp65 = tmp0 == tmp64
tmp66 = tl.where(tmp65, tmp3, tmp4)
tmp68 = tmp67.to(tl.float32)
tmp69 = tmp8 - tmp68
tmp70 = triton_helpers.maximum(tmp69, tmp10)
tmp71 = tmp70 * tmp12
tmp72 = tmp68 - tmp14
tmp73 = tmp71 * tmp72
tmp74 = tmp67 - tmp1
tmp75 = tmp74.to(tl.float32)
tmp76 = tmp75 + tmp19
tmp77 = triton_helpers.maximum(tmp76, tmp10)
tmp78 = tmp77 * tmp12
tmp79 = tmp75 - tmp19
tmp80 = tmp78 * tmp79
tmp81 = tl.where(tmp66, tmp73, tmp80)
tmp82 = triton_helpers.maximum(tmp63, tmp81)
tmp83 = tmp25 - tmp82
tmp84 = tl_math.exp(tmp83)
tmp85 = tmp43 - tmp82
tmp86 = tl_math.exp(tmp85)
tmp87 = tmp84 + tmp86
tmp88 = tmp62 - tmp82
tmp89 = tl_math.exp(tmp88)
tmp90 = tmp87 + tmp89
tmp91 = tmp81 - tmp82
tmp92 = tl_math.exp(tmp91)
tmp93 = tmp90 + tmp92
tl.store(out_ptr0 + x0, tmp82, xmask)
tl.store(out_ptr1 + x0, tmp93, xmask)
@triton.jit
def triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_1(
in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4
x0 = xindex % 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + x2, xmask)
tmp27 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp1 = x0
tmp2 = tmp0 == tmp1
tmp3 = tl.full([1], True, tl.int1)
tmp4 = tl.full([1], False, tl.int1)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp7 = tmp6.to(tl.float32)
tmp8 = 1.25
tmp9 = tmp8 - tmp7
tmp10 = 0.0
tmp11 = triton_helpers.maximum(tmp9, tmp10)
tmp12 = 256.0
tmp13 = tmp11 * tmp12
tmp14 = 0.75
tmp15 = tmp7 - tmp14
tmp16 = tmp13 * tmp15
tmp17 = tl.full([1], 0, tl.int64)
tmp18 = tmp6 - tmp17
tmp19 = tmp18.to(tl.float32)
tmp20 = 0.25
tmp21 = tmp19 + tmp20
tmp22 = triton_helpers.maximum(tmp21, tmp10)
tmp23 = tmp22 * tmp12
tmp24 = tmp19 - tmp20
tmp25 = tmp23 * tmp24
tmp26 = tl.where(tmp5, tmp16, tmp25)
tmp28 = tmp26 - tmp27
tmp30 = tl_math.log(tmp29)
tmp31 = tmp28 - tmp30
tl.store(out_ptr0 + x2, tmp31, xmask)
@triton.jit
def triton_per_fused_nll_loss_forward_2(in_out_ptr0, in_ptr0, in_ptr1,
xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 4
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = tl.full([1, 1], -100, tl.int64)
tmp2 = tmp0 != tmp1
tmp3 = tl.full([1, 1], 0, tl.int64)
tmp4 = tl.where(tmp2, tmp0, tmp3)
tmp5 = tl.full([XBLOCK, RBLOCK], 4, tl.int32)
tmp6 = tmp4 + tmp5
tmp7 = tmp4 < 0
tmp8 = tl.where(tmp7, tmp6, tmp4)
tl.device_assert((0 <= tmp8) & (tmp8 < 4),
'index out of bounds: 0 <= tmp8 < 4')
tmp10 = tl.load(in_ptr1 + (tmp8 + 4 * r0), None, eviction_policy=
'evict_last')
tmp11 = -tmp10
tmp12 = 0.0
tmp13 = tl.where(tmp2, tmp11, tmp12)
tmp14 = tl.broadcast_to(tmp13, [XBLOCK, RBLOCK])
tmp16 = tl.sum(tmp14, 1)[:, None]
tmp17 = tmp2.to(tl.int64)
tmp18 = tl.broadcast_to(tmp17, [XBLOCK, RBLOCK])
tmp20 = tl.sum(tmp18, 1)[:, None]
tmp21 = tmp20.to(tl.float32)
tmp22 = tmp16 / tmp21
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp22, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4), (4, 1))
assert_size_stride(arg1_1, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf1 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
get_raw_stream(0)
triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_0[
grid(4)](arg1_1, arg0_1, buf0, buf1, 4, XBLOCK=4, num_warps=1,
num_stages=1)
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
triton_poi_fused__log_softmax_add_clamp_mul_rsub_scatter_sub_where_1[
grid(16)](arg1_1, arg0_1, buf0, buf1, buf2, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del arg0_1
del buf0
del buf1
buf3 = empty_strided_cuda((), (), torch.float32)
buf5 = buf3
del buf3
triton_per_fused_nll_loss_forward_2[grid(1)](buf5, arg1_1, buf2, 1,
4, XBLOCK=1, num_warps=2, num_stages=1)
del arg1_1
del buf2
return buf5,
class ClassificationCircleLossNew(nn.Module):
"""Circle loss for class-level labels as described in the paper
`"Circle Loss: A Unified Perspective of Pair Similarity Optimization" <#>`_
Args:
scale (float): the scale factor. Default: 256.0
margin (float): the relax margin value. Default: 0.25
circle_center (tuple[float]): the center of the circle (logit_ap, logit_an). Default: (1, 0)
reduction (string, optional): Specifies the reduction to apply to the output:
``'none'`` | ``'mean'`` | ``'sum'``. ``'none'``: no reduction will be applied,
``'mean'``: the sum of the output will be divided by the number of
elements in the output, ``'sum'``: the output will be summed. Default: ``'mean'``
"""
def __init__(self, scale: 'float'=256.0, margin: 'float'=0.25,
circle_center: 'Tuple[float, float]'=(1, 0), reduction: 'str'='mean'
) ->None:
super(ClassificationCircleLossNew, self).__init__()
self.scale = scale
self.margin = margin
self.circle_center = circle_center
self.reduction = reduction
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
StevenGrove/DynamicHead
|
ClassificationCircleLoss
| false | 14,451 |
[
"Apache-2.0"
] | 69 |
d62aa84e1d1c6a0c74d46258ad77b11413c10bef
|
https://github.com/StevenGrove/DynamicHead/tree/d62aa84e1d1c6a0c74d46258ad77b11413c10bef
|
SoftmaxLayer
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/td/ctdj5kazgiki6gdaadhqtp2x7tq2ee5ey5hqqdcoqmp54jyhf74f.py
# Topologically Sorted Source Nodes: [cross_entropy], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# cross_entropy => amax, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%view_1, [1], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_1, %amax), kwargs = {})
triton_poi_fused__log_softmax_0 = async_compile.triton('triton_poi_fused__log_softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + (x3), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/gi/cgithwybw2tzjapvn6gjr3k5f6k5xouk6su56n4ctmox3wk6syln.py
# Topologically Sorted Source Nodes: [cross_entropy], Original ATen: [aten._log_softmax, aten.mul, aten.sum, aten.neg]
# Source node to ATen node mapping:
# cross_entropy => exp, log, mul, neg, sub_1, sum_1, sum_2
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %primals_4), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%mul,), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%sum_2,), kwargs = {})
triton_per_fused__log_softmax_mul_neg_sum_1 = async_compile.triton('triton_per_fused__log_softmax_mul_neg_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__log_softmax_mul_neg_sum_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 6, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__log_softmax_mul_neg_sum_1(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 16
r2 = (rindex // 64)
tmp0 = tl.load(in_ptr0 + (r3), None)
tmp1 = tl.load(in_ptr0 + (r0 + (64*r2)), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr1 + (r3), None)
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tmp15 = tmp13 * tmp14
tmp16 = tl.broadcast_to(tmp15, [RBLOCK])
tmp18 = triton_helpers.promote_to_tensor(tl.sum(tmp16, 0))
tmp19 = -tmp18
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp19, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [tag_scores], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cross_entropy], Original ATen: [aten._log_softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__log_softmax_0.run(buf0, buf1, 256, grid=grid(256), stream=stream0)
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [cross_entropy], Original ATen: [aten._log_softmax, aten.mul, aten.sum, aten.neg]
triton_per_fused__log_softmax_mul_neg_sum_1.run(buf3, buf1, primals_4, 1, 256, grid=grid(1), stream=stream0)
del buf1
return (buf3, primals_4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class SoftmaxLayer(nn.Module):
""" Naive softmax-layer """
def __init__(self, output_dim, n_class):
"""
:param output_dim: int
:param n_class: int
"""
super(SoftmaxLayer, self).__init__()
self.hidden2tag = nn.Linear(output_dim, n_class)
self.criterion = nn.CrossEntropyLoss(size_average=False)
def forward(self, x, y):
"""
:param x: torch.Tensor
:param y: torch.Tensor
:return:
"""
tag_scores = self.hidden2tag(x)
return self.criterion(tag_scores, y)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'output_dim': 4, 'n_class': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__log_softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + x3, tmp8, xmask)
@triton.jit
def triton_per_fused__log_softmax_mul_neg_sum_1(in_out_ptr0, in_ptr0,
in_ptr1, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 16
r2 = rindex // 64
tmp0 = tl.load(in_ptr0 + r3, None)
tmp1 = tl.load(in_ptr0 + (r0 + 64 * r2), None, eviction_policy='evict_last'
)
tmp3 = tl.load(in_ptr0 + (16 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (32 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (48 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp14 = tl.load(in_ptr1 + r3, None)
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tmp15 = tmp13 * tmp14
tmp16 = tl.broadcast_to(tmp15, [RBLOCK])
tmp18 = triton_helpers.promote_to_tensor(tl.sum(tmp16, 0))
tmp19 = -tmp18
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp19, None)
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__log_softmax_0[grid(256)](buf0, buf1, 256, XBLOCK=
128, num_warps=4, num_stages=1)
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2
del buf2
triton_per_fused__log_softmax_mul_neg_sum_1[grid(1)](buf3, buf1,
primals_4, 1, 256, num_warps=2, num_stages=1)
del buf1
return buf3, primals_4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf0
class SoftmaxLayerNew(nn.Module):
""" Naive softmax-layer """
def __init__(self, output_dim, n_class):
"""
:param output_dim: int
:param n_class: int
"""
super(SoftmaxLayerNew, self).__init__()
self.hidden2tag = nn.Linear(output_dim, n_class)
self.criterion = nn.CrossEntropyLoss(size_average=False)
def forward(self, input_0, input_1):
primals_1 = self.hidden2tag.weight
primals_2 = self.hidden2tag.bias
primals_3 = input_0
primals_4 = input_1
output = call([primals_1, primals_2, primals_3, primals_4])
return output[0]
|
Sy-Zhang/ELMoForManyLangs
|
SoftmaxLayer
| false | 14,452 |
[
"MIT"
] | 1,414 |
f82bf0fef80df617e39d34baa3e46d9857e94e65
|
https://github.com/Sy-Zhang/ELMoForManyLangs/tree/f82bf0fef80df617e39d34baa3e46d9857e94e65
|
disparityregression
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/5s/c5squiykk4px43qubk7u6sa2t45czr633uswoju5nbxaxiulyd63.py
# Topologically Sorted Source Nodes: [mul, out], Original ATen: [aten.mul, aten.sum]
# Source node to ATen node mapping:
# mul => mul
# out => sum_1
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, %unsqueeze_2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [1]), kwargs = {})
triton_poi_fused_mul_sum_0 = async_compile.triton('triton_poi_fused_mul_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_sum_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (64 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + (128 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr1 + (192 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp4 = tmp0 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp0 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp0 * tmp9
tmp11 = tmp8 + tmp10
tl.store(out_ptr0 + (x2), tmp11, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((1, 1, 4, 4, 4, 4), (256, 256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, out], Original ATen: [aten.mul, aten.sum]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_sum_0.run(arg1_1, arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
del arg1_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
from _paritybench_helpers import _mock_config
import torch
import numpy as np
from torch import nn
import torch.utils.data
from torch.autograd import Variable
import torch.nn.parallel
import torch.utils.data.distributed
class disparityregression(nn.Module):
def __init__(self, maxdisp, cfg):
super(disparityregression, self).__init__()
self.disp = Variable(torch.Tensor(np.array(range(maxdisp))),
requires_grad=False)
def forward(self, x, depth):
out = torch.sum(x * depth[None, :, None, None], 1)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'maxdisp': 4, 'cfg': _mock_config()}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import numpy as np
from torch import nn
import torch.utils.data
from torch.autograd import Variable
import torch.nn.parallel
import torch.utils.data.distributed
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_mul_sum_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (64 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + (128 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr1 + (192 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp4 = tmp0 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp0 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp0 * tmp9
tmp11 = tmp8 + tmp10
tl.store(out_ptr0 + x2, tmp11, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((1, 1, 4, 4, 4, 4), (256, 256, 64, 16, 4,
1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_sum_0[grid(256)](arg1_1, arg0_1, buf0, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del arg0_1
del arg1_1
return buf0,
class disparityregressionNew(nn.Module):
def __init__(self, maxdisp, cfg):
super(disparityregressionNew, self).__init__()
self.disp = Variable(torch.Tensor(np.array(range(maxdisp))),
requires_grad=False)
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
Sarah20187/X-StereoLab
|
disparityregression
| false | 14,453 |
[
"MIT"
] | 192 |
9ae8c1413307e7df91b14a7f31e8a95f9e5754f9
|
https://github.com/Sarah20187/X-StereoLab/tree/9ae8c1413307e7df91b14a7f31e8a95f9e5754f9
|
RGBBlock
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/kt/cktl2ajvyoekfgd2slcnhbzdao66dnx5hyhjxyij65lxsksggkjb.py
# Topologically Sorted Source Nodes: [add, weights], Original ATen: [aten.add, aten.mul]
# Source node to ATen node mapping:
# add => add
# weights => mul
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%unsqueeze_2, 1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%unsqueeze_3, %add), kwargs = {})
triton_poi_fused_add_mul_0 = async_compile.triton('triton_poi_fused_add_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp4 = tmp0 * tmp3
tl.store(out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/f7/cf73pggcer4jxxnhdhgfke4kylkxhbikngtjw54ezv5h4guxeidb.py
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten._to_copy]
# Source node to ATen node mapping:
# x_4 => convert_element_type_1
# Graph fragment:
# %convert_element_type_1 : [num_users=5] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%view_3, torch.int64), kwargs = {})
triton_poi_fused__to_copy_1 = async_compile.triton('triton_poi_fused__to_copy_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0,), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_1(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tl.store(out_ptr0 + (x0), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/7j/c7jycyq644ggynst23klcavbh7pgjlmhdje7mmnxmblckhihehbp.py
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.add, aten.clamp]
# Source node to ATen node mapping:
# x_4 => add_3, clamp_max
# Graph fragment:
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convert_element_type_1, 1), kwargs = {})
# %clamp_max : [num_users=3] = call_function[target=torch.ops.aten.clamp_max.default](args = (%add_3, 3), kwargs = {})
triton_poi_fused_add_clamp_2 = async_compile.triton('triton_poi_fused_add_clamp_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0,), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_2(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.full([1], 1, tl.int64)
tmp10 = tmp8 + tmp9
tmp11 = tl.full([1], 3, tl.int64)
tmp12 = triton_helpers.minimum(tmp10, tmp11)
tl.store(out_ptr0 + (x0), tmp12, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/f2/cf2eaf34xy4hwkwpj455spstljzp6g25wzwkehr426mrz5o6d6zy.py
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
# Source node to ATen node mapping:
# x_4 => add_2, clamp_max_2, clamp_min, clamp_min_2, convert_element_type, iota, mul_1, sub, sub_2
# Graph fragment:
# %iota : [num_users=1] = call_function[target=torch.ops.prims.iota.default](args = (8,), kwargs = {start: 0, step: 1, dtype: torch.int64, device: cuda:0, requires_grad: False})
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%iota, torch.float32), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convert_element_type, 0.5), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_2, 0.5), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_1, 0.5), kwargs = {})
# %clamp_min : [num_users=3] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub, 0.0), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%clamp_min, %convert_element_type_3), kwargs = {})
# %clamp_min_2 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub_2, 0.0), kwargs = {})
# %clamp_max_2 : [num_users=3] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min_2, 1.0), kwargs = {})
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3 = async_compile.triton('triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0,), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 - tmp9
tmp11 = triton_helpers.maximum(tmp10, tmp6)
tmp12 = 1.0
tmp13 = triton_helpers.minimum(tmp11, tmp12)
tl.store(out_ptr0 + (x0), tmp13, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/mp/cmpyq4btx6boiyxgscv2tljljuyadiezmyogegfdyykyfrun3xju.py
# Topologically Sorted Source Nodes: [x_3, x_4], Original ATen: [aten.add, aten._unsafe_index, aten.sub, aten.mul]
# Source node to ATen node mapping:
# x_3 => add_1
# x_4 => _unsafe_index, _unsafe_index_1, _unsafe_index_2, _unsafe_index_3, add_6, add_7, add_8, mul_3, mul_4, mul_5, sub_3, sub_4, sub_6
# Graph fragment:
# %add_1 : [num_users=4] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_2, %primals_6), kwargs = {})
# %_unsafe_index : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%add_1, [None, None, %convert_element_type_1, %convert_element_type_3]), kwargs = {})
# %_unsafe_index_1 : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%add_1, [None, None, %convert_element_type_1, %clamp_max_1]), kwargs = {})
# %_unsafe_index_2 : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%add_1, [None, None, %clamp_max, %convert_element_type_3]), kwargs = {})
# %_unsafe_index_3 : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%add_1, [None, None, %clamp_max, %clamp_max_1]), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%_unsafe_index_1, %_unsafe_index), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_3, %clamp_max_2), kwargs = {})
# %add_6 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%_unsafe_index, %mul_3), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%_unsafe_index_3, %_unsafe_index_2), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_4, %clamp_max_2), kwargs = {})
# %add_7 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%_unsafe_index_2, %mul_4), kwargs = {})
# %sub_6 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_7, %add_6), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_6, %clamp_max_3), kwargs = {})
# %add_8 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_6, %mul_5), kwargs = {})
triton_poi_fused__unsafe_index_add_mul_sub_4 = async_compile.triton('triton_poi_fused__unsafe_index_add_mul_sub_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i64', 2: '*i64', 3: '*fp32', 4: '*fp32', 5: '*i64', 6: '*fp32', 7: '*i64', 8: '*fp32', 9: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__unsafe_index_add_mul_sub_4', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__unsafe_index_add_mul_sub_4(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 8) % 8
x0 = xindex % 8
x2 = (xindex // 64)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr5 + (x0), xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr6 + (x1), xmask, eviction_policy='evict_last')
tmp37 = tl.load(in_ptr7 + (x1), xmask, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 4, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr2 + (tmp8 + (4*tmp4) + (16*x2)), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr3 + (tmp8 + (4*tmp4)), xmask, eviction_policy='evict_last')
tmp11 = tmp9 + tmp10
tmp13 = tmp12 + tmp1
tmp14 = tmp12 < 0
tmp15 = tl.where(tmp14, tmp13, tmp12)
tmp16 = tl.load(in_ptr2 + (tmp15 + (4*tmp4) + (16*x2)), xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr3 + (tmp15 + (4*tmp4)), xmask, eviction_policy='evict_last')
tmp18 = tmp16 + tmp17
tmp19 = tmp18 - tmp11
tmp21 = tmp19 * tmp20
tmp22 = tmp11 + tmp21
tmp24 = tmp23 + tmp1
tmp25 = tmp23 < 0
tmp26 = tl.where(tmp25, tmp24, tmp23)
tmp27 = tl.load(in_ptr2 + (tmp8 + (4*tmp26) + (16*x2)), xmask, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr3 + (tmp8 + (4*tmp26)), xmask, eviction_policy='evict_last')
tmp29 = tmp27 + tmp28
tmp30 = tl.load(in_ptr2 + (tmp15 + (4*tmp26) + (16*x2)), xmask, eviction_policy='evict_last')
tmp31 = tl.load(in_ptr3 + (tmp15 + (4*tmp26)), xmask, eviction_policy='evict_last')
tmp32 = tmp30 + tmp31
tmp33 = tmp32 - tmp29
tmp34 = tmp33 * tmp20
tmp35 = tmp29 + tmp34
tmp36 = tmp35 - tmp22
tmp38 = tmp36 * tmp37
tmp39 = tmp22 + tmp38
tl.store(in_out_ptr0 + (x4), tmp39, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [style], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_3, primals_4, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_2
del primals_3
buf1 = empty_strided_cuda((4, 1, 4, 1, 1), (4, 1, 1, 4, 4), torch.float32)
# Topologically Sorted Source Nodes: [add, weights], Original ATen: [aten.add, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_add_mul_0.run(primals_5, buf0, buf1, 16, grid=grid(16), stream=stream0)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1, 16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf1, (4, 4, 1, 1), (4, 1, 0, 0), 0), stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=4, bias=None)
assert_size_stride(buf2, (1, 4, 4, 4), (64, 16, 4, 1))
buf3 = empty_strided_cuda((8, 1), (1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten._to_copy]
triton_poi_fused__to_copy_1.run(buf3, 8, grid=grid(8), stream=stream0)
buf4 = empty_strided_cuda((8, 1), (1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.add, aten.clamp]
triton_poi_fused_add_clamp_2.run(buf4, 8, grid=grid(8), stream=stream0)
buf5 = empty_strided_cuda((8, ), (1, ), torch.int64)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
triton_poi_fused__to_copy_1.run(buf5, 8, grid=grid(8), stream=stream0)
buf6 = empty_strided_cuda((8, ), (1, ), torch.int64)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.add, aten.clamp]
triton_poi_fused_add_clamp_2.run(buf6, 8, grid=grid(8), stream=stream0)
buf7 = empty_strided_cuda((8, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3.run(buf7, 8, grid=grid(8), stream=stream0)
buf9 = empty_strided_cuda((8, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.sub, aten.clamp]
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3.run(buf9, 8, grid=grid(8), stream=stream0)
buf10 = empty_strided_cuda((4, 1, 8, 8), (64, 256, 8, 1), torch.float32)
buf11 = reinterpret_tensor(buf10, (4, 1, 8, 8), (64, 64, 8, 1), 0); del buf10 # reuse
# Topologically Sorted Source Nodes: [x_3, x_4], Original ATen: [aten.add, aten._unsafe_index, aten.sub, aten.mul]
triton_poi_fused__unsafe_index_add_mul_sub_4.run(buf11, buf3, buf5, buf2, primals_6, buf6, buf7, buf4, buf9, 256, grid=grid(256), stream=stream0)
del buf2
del primals_6
return (buf11, primals_4, primals_5, buf0, reinterpret_tensor(primals_1, (1, 16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf1, (4, 4, 1, 1), (4, 1, 1, 1), 0), buf3, buf4, buf5, buf6, buf7, buf9, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn.functional as F
import torch.nn as nn
class Conv2DMod(nn.Module):
def __init__(self, in_chan, out_chan, kernel, demod=True, stride=1,
dilation=1, **kwargs):
super().__init__()
self.filters = out_chan
self.demod = demod
self.kernel = kernel
self.stride = stride
self.dilation = dilation
self.weight = nn.Parameter(torch.randn((out_chan, in_chan, kernel,
kernel)))
nn.init.kaiming_normal_(self.weight, a=0, mode='fan_in',
nonlinearity='leaky_relu')
def _get_same_padding(self, size, kernel, dilation, stride):
return ((size - 1) * (stride - 1) + dilation * (kernel - 1)) // 2
def forward(self, x, y):
b, _c, h, w = x.shape
w1 = y[:, None, :, None, None]
w2 = self.weight[None, :, :, :, :]
weights = w2 * (w1 + 1)
if self.demod:
d = torch.rsqrt((weights ** 2).sum(dim=(2, 3, 4), keepdim=True) +
EPS)
weights = weights * d
x = x.reshape(1, -1, h, w)
_, _, *ws = weights.shape
weights = weights.reshape(b * self.filters, *ws)
padding = self._get_same_padding(h, self.kernel, self.dilation,
self.stride)
x = F.conv2d(x, weights, padding=padding, groups=b)
x = x.reshape(-1, self.filters, h, w)
return x
class RGBBlock(nn.Module):
def __init__(self, latent_dim, input_channel, upsample, rgba=False):
super().__init__()
self.input_channel = input_channel
self.to_style = nn.Linear(latent_dim, input_channel)
out_filters = 1
self.conv = Conv2DMod(input_channel, out_filters, 1, demod=False)
self.upsample = nn.Upsample(scale_factor=2, mode='bilinear',
align_corners=False) if upsample else None
def forward(self, x, prev_rgb, istyle):
_b, _c, _h, _w = x.shape
style = self.to_style(istyle)
x = self.conv(x, style)
if prev_rgb is not None:
x = x + prev_rgb
if self.upsample is not None:
x = self.upsample(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'latent_dim': 4, 'input_channel': 4, 'upsample': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn.functional as F
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_mul_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp4 = tmp0 * tmp3
tl.store(out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused__to_copy_1(out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tl.store(out_ptr0 + x0, tmp8, xmask)
@triton.jit
def triton_poi_fused_add_clamp_2(out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.full([1], 1, tl.int64)
tmp10 = tmp8 + tmp9
tmp11 = tl.full([1], 3, tl.int64)
tmp12 = triton_helpers.minimum(tmp10, tmp11)
tl.store(out_ptr0 + x0, tmp12, xmask)
@triton.jit
def triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3(out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 - tmp9
tmp11 = triton_helpers.maximum(tmp10, tmp6)
tmp12 = 1.0
tmp13 = triton_helpers.minimum(tmp11, tmp12)
tl.store(out_ptr0 + x0, tmp13, xmask)
@triton.jit
def triton_poi_fused__unsafe_index_add_mul_sub_4(in_out_ptr0, in_ptr0,
in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 8 % 8
x0 = xindex % 8
x2 = xindex // 64
x4 = xindex
tmp0 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr5 + x0, xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr6 + x1, xmask, eviction_policy='evict_last')
tmp37 = tl.load(in_ptr7 + x1, xmask, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 4, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr2 + (tmp8 + 4 * tmp4 + 16 * x2), xmask,
eviction_policy='evict_last')
tmp10 = tl.load(in_ptr3 + (tmp8 + 4 * tmp4), xmask, eviction_policy=
'evict_last')
tmp11 = tmp9 + tmp10
tmp13 = tmp12 + tmp1
tmp14 = tmp12 < 0
tmp15 = tl.where(tmp14, tmp13, tmp12)
tmp16 = tl.load(in_ptr2 + (tmp15 + 4 * tmp4 + 16 * x2), xmask,
eviction_policy='evict_last')
tmp17 = tl.load(in_ptr3 + (tmp15 + 4 * tmp4), xmask, eviction_policy=
'evict_last')
tmp18 = tmp16 + tmp17
tmp19 = tmp18 - tmp11
tmp21 = tmp19 * tmp20
tmp22 = tmp11 + tmp21
tmp24 = tmp23 + tmp1
tmp25 = tmp23 < 0
tmp26 = tl.where(tmp25, tmp24, tmp23)
tmp27 = tl.load(in_ptr2 + (tmp8 + 4 * tmp26 + 16 * x2), xmask,
eviction_policy='evict_last')
tmp28 = tl.load(in_ptr3 + (tmp8 + 4 * tmp26), xmask, eviction_policy=
'evict_last')
tmp29 = tmp27 + tmp28
tmp30 = tl.load(in_ptr2 + (tmp15 + 4 * tmp26 + 16 * x2), xmask,
eviction_policy='evict_last')
tmp31 = tl.load(in_ptr3 + (tmp15 + 4 * tmp26), xmask, eviction_policy=
'evict_last')
tmp32 = tmp30 + tmp31
tmp33 = tmp32 - tmp29
tmp34 = tmp33 * tmp20
tmp35 = tmp29 + tmp34
tmp36 = tmp35 - tmp22
tmp38 = tmp36 * tmp37
tmp39 = tmp22 + tmp38
tl.store(in_out_ptr0 + x4, tmp39, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_3, primals_4, reinterpret_tensor(
primals_2, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_2
del primals_3
buf1 = empty_strided_cuda((4, 1, 4, 1, 1), (4, 1, 1, 4, 4), torch.
float32)
get_raw_stream(0)
triton_poi_fused_add_mul_0[grid(16)](primals_5, buf0, buf1, 16,
XBLOCK=16, num_warps=1, num_stages=1)
buf2 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1,
16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf1, (4, 4,
1, 1), (4, 1, 0, 0), 0), stride=(1, 1), padding=(0, 0),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=4, bias=None)
assert_size_stride(buf2, (1, 4, 4, 4), (64, 16, 4, 1))
buf3 = empty_strided_cuda((8, 1), (1, 1), torch.int64)
triton_poi_fused__to_copy_1[grid(8)](buf3, 8, XBLOCK=8, num_warps=1,
num_stages=1)
buf4 = empty_strided_cuda((8, 1), (1, 1), torch.int64)
triton_poi_fused_add_clamp_2[grid(8)](buf4, 8, XBLOCK=8, num_warps=
1, num_stages=1)
buf5 = empty_strided_cuda((8,), (1,), torch.int64)
triton_poi_fused__to_copy_1[grid(8)](buf5, 8, XBLOCK=8, num_warps=1,
num_stages=1)
buf6 = empty_strided_cuda((8,), (1,), torch.int64)
triton_poi_fused_add_clamp_2[grid(8)](buf6, 8, XBLOCK=8, num_warps=
1, num_stages=1)
buf7 = empty_strided_cuda((8,), (1,), torch.float32)
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3[grid(8)](buf7,
8, XBLOCK=8, num_warps=1, num_stages=1)
buf9 = empty_strided_cuda((8, 1), (1, 1), torch.float32)
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_3[grid(8)](buf9,
8, XBLOCK=8, num_warps=1, num_stages=1)
buf10 = empty_strided_cuda((4, 1, 8, 8), (64, 256, 8, 1), torch.float32
)
buf11 = reinterpret_tensor(buf10, (4, 1, 8, 8), (64, 64, 8, 1), 0)
del buf10
triton_poi_fused__unsafe_index_add_mul_sub_4[grid(256)](buf11, buf3,
buf5, buf2, primals_6, buf6, buf7, buf4, buf9, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del buf2
del primals_6
return buf11, primals_4, primals_5, buf0, reinterpret_tensor(primals_1,
(1, 16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf1, (4, 4,
1, 1), (4, 1, 1, 1), 0), buf3, buf4, buf5, buf6, buf7, buf9
class Conv2DMod(nn.Module):
def __init__(self, in_chan, out_chan, kernel, demod=True, stride=1,
dilation=1, **kwargs):
super().__init__()
self.filters = out_chan
self.demod = demod
self.kernel = kernel
self.stride = stride
self.dilation = dilation
self.weight = nn.Parameter(torch.randn((out_chan, in_chan, kernel,
kernel)))
nn.init.kaiming_normal_(self.weight, a=0, mode='fan_in',
nonlinearity='leaky_relu')
def _get_same_padding(self, size, kernel, dilation, stride):
return ((size - 1) * (stride - 1) + dilation * (kernel - 1)) // 2
def forward(self, x, y):
b, _c, h, w = x.shape
w1 = y[:, None, :, None, None]
w2 = self.weight[None, :, :, :, :]
weights = w2 * (w1 + 1)
if self.demod:
d = torch.rsqrt((weights ** 2).sum(dim=(2, 3, 4), keepdim=True) +
EPS)
weights = weights * d
x = x.reshape(1, -1, h, w)
_, _, *ws = weights.shape
weights = weights.reshape(b * self.filters, *ws)
padding = self._get_same_padding(h, self.kernel, self.dilation,
self.stride)
x = F.conv2d(x, weights, padding=padding, groups=b)
x = x.reshape(-1, self.filters, h, w)
return x
class RGBBlockNew(nn.Module):
def __init__(self, latent_dim, input_channel, upsample, rgba=False):
super().__init__()
self.input_channel = input_channel
self.to_style = nn.Linear(latent_dim, input_channel)
out_filters = 1
self.conv = Conv2DMod(input_channel, out_filters, 1, demod=False)
self.upsample = nn.Upsample(scale_factor=2, mode='bilinear',
align_corners=False) if upsample else None
def forward(self, input_0, input_1, input_2):
primals_2 = self.to_style.weight
primals_3 = self.to_style.bias
primals_5 = self.conv.weight
primals_1 = input_0
primals_4 = input_1
primals_6 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6])
return output[0]
|
SongweiGe/DoodlerGAN
|
RGBBlock
| false | 14,454 |
[
"MIT"
] | 92 |
d435d9b3c0579937cd3c22aa2051960ceb921785
|
https://github.com/SongweiGe/DoodlerGAN/tree/d435d9b3c0579937cd3c22aa2051960ceb921785
|
SpatialGate
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/zo/czobpmlyr5atbcpsuque6vcmk7nafmb3smtbzoqilz46drm7zbkm.py
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv2d => convolution
# Graph fragment:
# %convolution : [num_users=2] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/3k/c3kqucdvk4o4zpjqokxiye7xneshc77mutxurbuxezpwyz5n5xoj.py
# Topologically Sorted Source Nodes: [tanh, gate, gate_1, x, output], Original ATen: [aten.tanh, aten.clamp, aten.view, aten.mul]
# Source node to ATen node mapping:
# gate => clamp_min
# gate_1 => view_1
# output => view_2
# tanh => tanh
# x => mul
# Graph fragment:
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%convolution,), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%tanh, 0), kwargs = {})
# %view_1 : [num_users=1] = call_function[target=torch.ops.aten.reshape.default](args = (%clamp_min, [4, -1, 4, 4]), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view, %view_1), kwargs = {})
# %view_2 : [num_users=1] = call_function[target=torch.ops.aten.reshape.default](args = (%mul, [4, -1, 4, 4]), kwargs = {})
triton_poi_fused_clamp_mul_tanh_view_1 = async_compile.triton('triton_poi_fused_clamp_mul_tanh_view_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clamp_mul_tanh_view_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clamp_mul_tanh_view_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr1 + (x0 + (16*x2)), xmask, eviction_policy='evict_last')
tmp2 = libdevice.tanh(tmp1)
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = tmp0 * tmp4
tl.store(out_ptr0 + (x3), tmp5, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_2, (1, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 4, 4), (16, 16, 4, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(buf1, primals_2, 64, grid=grid(64), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [tanh, gate, gate_1, x, output], Original ATen: [aten.tanh, aten.clamp, aten.view, aten.mul]
triton_poi_fused_clamp_mul_tanh_view_1.run(primals_4, buf1, buf2, 256, grid=grid(256), stream=stream0)
return (buf2, primals_1, primals_3, primals_4, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
import torch.nn as nn
import torch.utils.data
from itertools import product as product
from math import sqrt as sqrt
class SpatialGate(nn.Module):
def __init__(self, in_channels: 'int', num_groups: 'int'=1, kernel_size:
'int'=1, padding: 'int'=0, stride: 'int'=1, gate_activation: 'str'=
'ReTanH', gate_activation_kargs: 'dict'=None, get_running_cost:
'callable'=None):
super(SpatialGate, self).__init__()
self.num_groups = num_groups
self.gate_conv = nn.Conv2d(in_channels, num_groups, kernel_size,
padding=padding, stride=stride)
self.gate_activation = gate_activation
self.gate_activation_kargs = gate_activation_kargs
if gate_activation == 'ReTanH':
self.gate_activate = lambda x: torch.tanh(x).clamp(min=0)
elif gate_activation == 'Sigmoid':
self.gate_activate = lambda x: torch.sigmoid(x)
elif gate_activation == 'GeReTanH':
assert 'tau' in gate_activation_kargs
tau = gate_activation_kargs['tau']
ttau = math.tanh(tau)
self.gate_activate = lambda x: ((torch.tanh(x - tau) + ttau) /
(1 + ttau)).clamp(min=0)
else:
raise NotImplementedError()
self.get_running_cost = get_running_cost
self.running_cost = None
self.init_parameters()
def init_parameters(self, init_gate=0.99):
if self.gate_activation == 'ReTanH':
bias_value = 0.5 * math.log((1 + init_gate) / (1 - init_gate))
elif self.gate_activation == 'Sigmoid':
bias_value = 0.5 * math.log(init_gate / (1 - init_gate))
elif self.gate_activation == 'GeReTanH':
tau = self.gate_activation_kargs['tau']
bias_value = 0.5 * math.log((1 + init_gate * math.exp(2 * tau)) /
(1 - init_gate))
nn.init.normal_(self.gate_conv.weight, std=0.01)
nn.init.constant_(self.gate_conv.bias, bias_value)
def encode(self, *inputs):
outputs = [x.view(x.shape[0] * self.num_groups, -1, *x.shape[2:]) for
x in inputs]
return outputs
def decode(self, *inputs):
outputs = [x.view(x.shape[0] // self.num_groups, -1, *x.shape[2:]) for
x in inputs]
return outputs
def update_running_cost(self, gate):
if self.get_running_cost is not None:
cost = self.get_running_cost(gate)
if self.running_cost is not None:
self.running_cost = [(x + y) for x, y in zip(self.
running_cost, cost)]
else:
self.running_cost = cost
def clear_running_cost(self):
self.running_cost = None
def forward(self, data_input, gate_input, masked_func=None):
gate = self.gate_activate(self.gate_conv(gate_input))
self.update_running_cost(gate)
if masked_func is not None:
data_input = masked_func(data_input, gate)
data, gate = self.encode(data_input, gate)
output, = self.decode(data * gate)
return output
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
import math
import torch.nn as nn
import torch.utils.data
from itertools import product as product
from math import sqrt as sqrt
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + x0, tmp3, xmask)
@triton.jit
def triton_poi_fused_clamp_mul_tanh_view_1(in_ptr0, in_ptr1, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr1 + (x0 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = libdevice.tanh(tmp1)
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = tmp0 * tmp4
tl.store(out_ptr0 + x3, tmp5, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_2, (1,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 4, 4), (16, 16, 4, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(64)](buf1, primals_2, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_clamp_mul_tanh_view_1[grid(256)](primals_4, buf1,
buf2, 256, XBLOCK=128, num_warps=4, num_stages=1)
return buf2, primals_1, primals_3, primals_4, buf1
class SpatialGateNew(nn.Module):
def __init__(self, in_channels: 'int', num_groups: 'int'=1, kernel_size:
'int'=1, padding: 'int'=0, stride: 'int'=1, gate_activation: 'str'=
'ReTanH', gate_activation_kargs: 'dict'=None, get_running_cost:
'callable'=None):
super(SpatialGateNew, self).__init__()
self.num_groups = num_groups
self.gate_conv = nn.Conv2d(in_channels, num_groups, kernel_size,
padding=padding, stride=stride)
self.gate_activation = gate_activation
self.gate_activation_kargs = gate_activation_kargs
if gate_activation == 'ReTanH':
self.gate_activate = lambda x: torch.tanh(x).clamp(min=0)
elif gate_activation == 'Sigmoid':
self.gate_activate = lambda x: torch.sigmoid(x)
elif gate_activation == 'GeReTanH':
assert 'tau' in gate_activation_kargs
tau = gate_activation_kargs['tau']
ttau = math.tanh(tau)
self.gate_activate = lambda x: ((torch.tanh(x - tau) + ttau) /
(1 + ttau)).clamp(min=0)
else:
raise NotImplementedError()
self.get_running_cost = get_running_cost
self.running_cost = None
self.init_parameters()
def init_parameters(self, init_gate=0.99):
if self.gate_activation == 'ReTanH':
bias_value = 0.5 * math.log((1 + init_gate) / (1 - init_gate))
elif self.gate_activation == 'Sigmoid':
bias_value = 0.5 * math.log(init_gate / (1 - init_gate))
elif self.gate_activation == 'GeReTanH':
tau = self.gate_activation_kargs['tau']
bias_value = 0.5 * math.log((1 + init_gate * math.exp(2 * tau)) /
(1 - init_gate))
nn.init.normal_(self.gate_conv.weight, std=0.01)
nn.init.constant_(self.gate_conv.bias, bias_value)
def encode(self, *inputs):
outputs = [x.view(x.shape[0] * self.num_groups, -1, *x.shape[2:]) for
x in inputs]
return outputs
def decode(self, *inputs):
outputs = [x.view(x.shape[0] // self.num_groups, -1, *x.shape[2:]) for
x in inputs]
return outputs
def update_running_cost(self, gate):
if self.get_running_cost is not None:
cost = self.get_running_cost(gate)
if self.running_cost is not None:
self.running_cost = [(x + y) for x, y in zip(self.
running_cost, cost)]
else:
self.running_cost = cost
def clear_running_cost(self):
self.running_cost = None
def forward(self, input_0, input_1):
primals_1 = self.gate_conv.weight
primals_2 = self.gate_conv.bias
primals_3 = input_0
primals_4 = input_1
output = call([primals_1, primals_2, primals_3, primals_4])
return output[0]
|
StevenGrove/DynamicHead
|
SpatialGate
| false | 14,455 |
[
"Apache-2.0"
] | 69 |
d62aa84e1d1c6a0c74d46258ad77b11413c10bef
|
https://github.com/StevenGrove/DynamicHead/tree/d62aa84e1d1c6a0c74d46258ad77b11413c10bef
|
VAE_Kl_Loss
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/mr/cmrobrctrz66fo5x4ya6y7y2roknh7wbnn6clbjhwr4dkcoxlen2.py
# Topologically Sorted Source Nodes: [pow_1, exp, add, sub, sub_1, mul, sum_1, loss_KL], Original ATen: [aten.pow, aten.exp, aten.add, aten.sub, aten.mul, aten.sum, aten.mean]
# Source node to ATen node mapping:
# add => add
# exp => exp
# loss_KL => mean
# mul => mul
# pow_1 => pow_1
# sub => sub
# sub_1 => sub_1
# sum_1 => sum_1
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%arg0_1, 2), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%arg1_1,), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%pow_1, %exp), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %arg1_1), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, 1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, 0.5), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_1,), kwargs = {})
triton_per_fused_add_exp_mean_mul_pow_sub_sum_0 = async_compile.triton('triton_per_fused_add_exp_mean_mul_pow_sub_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_exp_mean_mul_pow_sub_sum_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_exp_mean_mul_pow_sub_sum_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex % 16
r1 = (rindex // 16)
r2 = rindex
tmp0 = tl.load(in_ptr0 + (r0 + (64*r1)), None)
tmp2 = tl.load(in_ptr1 + (r0 + (64*r1)), None)
tmp10 = tl.load(in_ptr0 + (16 + r0 + (64*r1)), None)
tmp12 = tl.load(in_ptr1 + (16 + r0 + (64*r1)), None)
tmp19 = tl.load(in_ptr0 + (32 + r0 + (64*r1)), None)
tmp21 = tl.load(in_ptr1 + (32 + r0 + (64*r1)), None)
tmp28 = tl.load(in_ptr0 + (48 + r0 + (64*r1)), None)
tmp30 = tl.load(in_ptr1 + (48 + r0 + (64*r1)), None)
tmp1 = tmp0 * tmp0
tmp3 = tl_math.exp(tmp2)
tmp4 = tmp1 + tmp3
tmp5 = tmp4 - tmp2
tmp6 = 1.0
tmp7 = tmp5 - tmp6
tmp8 = 0.5
tmp9 = tmp7 * tmp8
tmp11 = tmp10 * tmp10
tmp13 = tl_math.exp(tmp12)
tmp14 = tmp11 + tmp13
tmp15 = tmp14 - tmp12
tmp16 = tmp15 - tmp6
tmp17 = tmp16 * tmp8
tmp18 = tmp9 + tmp17
tmp20 = tmp19 * tmp19
tmp22 = tl_math.exp(tmp21)
tmp23 = tmp20 + tmp22
tmp24 = tmp23 - tmp21
tmp25 = tmp24 - tmp6
tmp26 = tmp25 * tmp8
tmp27 = tmp18 + tmp26
tmp29 = tmp28 * tmp28
tmp31 = tl_math.exp(tmp30)
tmp32 = tmp29 + tmp31
tmp33 = tmp32 - tmp30
tmp34 = tmp33 - tmp6
tmp35 = tmp34 * tmp8
tmp36 = tmp27 + tmp35
tmp37 = tl.broadcast_to(tmp36, [XBLOCK, RBLOCK])
tmp39 = tl.sum(tmp37, 1)[:, None]
tmp40 = 64.0
tmp41 = tmp39 / tmp40
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp41, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [pow_1, exp, add, sub, sub_1, mul, sum_1, loss_KL], Original ATen: [aten.pow, aten.exp, aten.add, aten.sub, aten.mul, aten.sum, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_add_exp_mean_mul_pow_sub_sum_0.run(buf2, arg0_1, arg1_1, 1, 64, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional
import torch.nn.parallel
import torch.utils.data
import torch.optim
import torch.utils.data.distributed
class VAE_Kl_Loss(nn.Module):
def __init__(self, if_print=False):
super(VAE_Kl_Loss, self).__init__()
self.if_print = if_print
def forward(self, means, variances):
loss = self.standard_KL_loss(means, variances)
if self.if_print:
None
return loss
def standard_KL_loss(self, means, variances):
loss_KL = torch.mean(torch.sum(0.5 * (means ** 2 + torch.exp(
variances) - variances - 1), dim=1))
return loss_KL
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
import torch.nn.functional
import torch.nn.parallel
import torch.utils.data
import torch.optim
import torch.utils.data.distributed
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_add_exp_mean_mul_pow_sub_sum_0(in_out_ptr0, in_ptr0,
in_ptr1, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex % 16
r1 = rindex // 16
tmp0 = tl.load(in_ptr0 + (r0 + 64 * r1), None)
tmp2 = tl.load(in_ptr1 + (r0 + 64 * r1), None)
tmp10 = tl.load(in_ptr0 + (16 + r0 + 64 * r1), None)
tmp12 = tl.load(in_ptr1 + (16 + r0 + 64 * r1), None)
tmp19 = tl.load(in_ptr0 + (32 + r0 + 64 * r1), None)
tmp21 = tl.load(in_ptr1 + (32 + r0 + 64 * r1), None)
tmp28 = tl.load(in_ptr0 + (48 + r0 + 64 * r1), None)
tmp30 = tl.load(in_ptr1 + (48 + r0 + 64 * r1), None)
tmp1 = tmp0 * tmp0
tmp3 = tl_math.exp(tmp2)
tmp4 = tmp1 + tmp3
tmp5 = tmp4 - tmp2
tmp6 = 1.0
tmp7 = tmp5 - tmp6
tmp8 = 0.5
tmp9 = tmp7 * tmp8
tmp11 = tmp10 * tmp10
tmp13 = tl_math.exp(tmp12)
tmp14 = tmp11 + tmp13
tmp15 = tmp14 - tmp12
tmp16 = tmp15 - tmp6
tmp17 = tmp16 * tmp8
tmp18 = tmp9 + tmp17
tmp20 = tmp19 * tmp19
tmp22 = tl_math.exp(tmp21)
tmp23 = tmp20 + tmp22
tmp24 = tmp23 - tmp21
tmp25 = tmp24 - tmp6
tmp26 = tmp25 * tmp8
tmp27 = tmp18 + tmp26
tmp29 = tmp28 * tmp28
tmp31 = tl_math.exp(tmp30)
tmp32 = tmp29 + tmp31
tmp33 = tmp32 - tmp30
tmp34 = tmp33 - tmp6
tmp35 = tmp34 * tmp8
tmp36 = tmp27 + tmp35
tmp37 = tl.broadcast_to(tmp36, [XBLOCK, RBLOCK])
tmp39 = tl.sum(tmp37, 1)[:, None]
tmp40 = 64.0
tmp41 = tmp39 / tmp40
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp41, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1
del buf1
get_raw_stream(0)
triton_per_fused_add_exp_mean_mul_pow_sub_sum_0[grid(1)](buf2,
arg0_1, arg1_1, 1, 64, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf2,
class VAE_Kl_LossNew(nn.Module):
def __init__(self, if_print=False):
super(VAE_Kl_LossNew, self).__init__()
self.if_print = if_print
def standard_KL_loss(self, means, variances):
loss_KL = torch.mean(torch.sum(0.5 * (means ** 2 + torch.exp(
variances) - variances - 1), dim=1))
return loss_KL
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
TPCD/LifelongReID
|
VAE_Kl_Loss
| false | 14,456 |
[
"MIT"
] | 63 |
cb33f9c29fe398e7546db345fab1c338dda8252f
|
https://github.com/TPCD/LifelongReID/tree/cb33f9c29fe398e7546db345fab1c338dda8252f
|
CategoricalSampler
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/c5/cc5hiyrqfmfxmipscxrurz47qz3x3e4v7b3qrmfw4clinzd5btca.py
# Topologically Sorted Source Nodes: [exp], Original ATen: [aten.exp]
# Source node to ATen node mapping:
# exp => exp
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%arg0_1,), kwargs = {})
triton_poi_fused_exp_0 = async_compile.triton('triton_poi_fused_exp_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_exp_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_exp_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl_math.exp(tmp0)
tl.store(out_ptr0 + (x0), tmp1, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [exp], Original ATen: [aten.exp]
stream0 = get_raw_stream(0)
triton_poi_fused_exp_0.run(arg0_1, buf0, 16, grid=grid(16), stream=stream0)
del arg0_1
# Topologically Sorted Source Nodes: [exp, multinomial], Original ATen: [aten.exp, aten.multinomial]
buf1 = torch.ops.aten.multinomial.default(buf0, 1)
del buf0
buf2 = buf1
del buf1
return (buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class Sampler(nn.Module):
""" args; logits: (batch, n_nodes)
return; next_node: (batch, 1)
TopKSampler <=> greedy; sample one with biggest probability
CategoricalSampler <=> sampling; randomly sample one from possible distribution based on probability
"""
def __init__(self, n_samples=1, **kwargs):
super().__init__(**kwargs)
self.n_samples = n_samples
class CategoricalSampler(Sampler):
def forward(self, logits):
return torch.multinomial(logits.exp(), self.n_samples)
def get_inputs():
return [torch.rand([4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_exp_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl_math.exp(tmp0)
tl.store(out_ptr0 + x0, tmp1, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_exp_0[grid(16)](arg0_1, buf0, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del arg0_1
buf1 = torch.ops.aten.multinomial.default(buf0, 1)
del buf0
buf2 = buf1
del buf1
return buf2,
class Sampler(nn.Module):
""" args; logits: (batch, n_nodes)
return; next_node: (batch, 1)
TopKSampler <=> greedy; sample one with biggest probability
CategoricalSampler <=> sampling; randomly sample one from possible distribution based on probability
"""
def __init__(self, n_samples=1, **kwargs):
super().__init__(**kwargs)
self.n_samples = n_samples
class CategoricalSamplerNew(Sampler):
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
TSLNIHAOGIT/VRP_DRL_MHA
|
CategoricalSampler
| false | 14,457 |
[
"MIT"
] | 55 |
6a59918ffb815fbdab4d75cb78130fc638c64d69
|
https://github.com/TSLNIHAOGIT/VRP_DRL_MHA/tree/6a59918ffb815fbdab4d75cb78130fc638c64d69
|
Pointnet
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/c3/cc3viy35ukuam57kedmccz7bf2yw3dvtjy2isdmexojnafyusphq.py
# Topologically Sorted Source Nodes: [x, relu], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# relu => relu
# x => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1], [0], [1], False, [0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_0 = async_compile.triton('triton_poi_fused_convolution_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 8
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ho/cho7jpiswqd4ykerylvqdy4fxxga3zxlozzedkqqf7vpydel6nf3.py
# Topologically Sorted Source Nodes: [x_1, max_1], Original ATen: [aten.convolution, aten.max]
# Source node to ATen node mapping:
# max_1 => getitem_1, max_1
# x_1 => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=2] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1], [0], [1], False, [0], 1), kwargs = {})
# %max_1 : [num_users=2] = call_function[target=torch.ops.aten.max.dim](args = (%convolution_1, 2, True), kwargs = {})
# %getitem_1 : [num_users=1] = call_function[target=operator.getitem](args = (%max_1, 1), kwargs = {})
triton_poi_fused_convolution_max_1 = async_compile.triton('triton_poi_fused_convolution_max_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i64', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_max_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_max_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (4*x2), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x2)), xmask, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr0 + (2 + (4*x2)), xmask, eviction_policy='evict_last')
tmp36 = tl.load(in_ptr0 + (3 + (4*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp5 = tmp2 > tmp4
tmp6 = tmp2 == tmp4
tmp7 = tmp2 != tmp2
tmp8 = tmp4 != tmp4
tmp9 = tmp7 > tmp8
tmp10 = tmp5 | tmp9
tmp11 = tmp7 & tmp8
tmp12 = tmp6 | tmp11
tmp13 = tl.full([1], 0, tl.int64)
tmp14 = tl.full([1], 1, tl.int64)
tmp15 = tmp13 < tmp14
tmp16 = tmp12 & tmp15
tmp17 = tmp10 | tmp16
tmp18 = tl.where(tmp17, tmp2, tmp4)
tmp19 = tl.where(tmp17, tmp13, tmp14)
tmp21 = tmp20 + tmp1
tmp22 = tmp18 > tmp21
tmp23 = tmp18 == tmp21
tmp24 = tmp18 != tmp18
tmp25 = tmp21 != tmp21
tmp26 = tmp24 > tmp25
tmp27 = tmp22 | tmp26
tmp28 = tmp24 & tmp25
tmp29 = tmp23 | tmp28
tmp30 = tl.full([1], 2, tl.int64)
tmp31 = tmp19 < tmp30
tmp32 = tmp29 & tmp31
tmp33 = tmp27 | tmp32
tmp34 = tl.where(tmp33, tmp18, tmp21)
tmp35 = tl.where(tmp33, tmp19, tmp30)
tmp37 = tmp36 + tmp1
tmp38 = tmp34 > tmp37
tmp39 = tmp34 == tmp37
tmp40 = tmp34 != tmp34
tmp41 = tmp37 != tmp37
tmp42 = tmp40 > tmp41
tmp43 = tmp38 | tmp42
tmp44 = tmp40 & tmp41
tmp45 = tmp39 | tmp44
tmp46 = tl.full([1], 3, tl.int64)
tmp47 = tmp35 < tmp46
tmp48 = tmp45 & tmp47
tmp49 = tmp43 | tmp48
tmp50 = tl.where(tmp49, tmp34, tmp37)
tmp51 = tl.where(tmp49, tmp35, tmp46)
tmp52 = triton_helpers.maximum(tmp2, tmp4)
tmp53 = triton_helpers.maximum(tmp52, tmp21)
tmp54 = triton_helpers.maximum(tmp53, tmp37)
tl.store(out_ptr0 + (x2), tmp51, xmask)
tl.store(out_ptr1 + (x2), tmp54, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/hv/chvrghq3gq6acc5azsamgh4r2z2lwefz54hnnys5utsfff5dq7lt.py
# Topologically Sorted Source Nodes: [x_2, relu_1], Original ATen: [aten.cat, aten.relu]
# Source node to ATen node mapping:
# relu_1 => relu_1
# x_2 => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%convolution_1, %expand], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%cat,), kwargs = {})
triton_poi_fused_cat_relu_2 = async_compile.triton('triton_poi_fused_cat_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_relu_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_relu_2(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 8
x0 = xindex % 4
x2 = (xindex // 32)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (4*x1) + (16*x2)), tmp4 & xmask, other=0.0)
tmp6 = tl.load(in_ptr1 + (x1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp7 = tmp5 + tmp6
tmp8 = tl.full(tmp7.shape, 0.0, tmp7.dtype)
tmp9 = tl.where(tmp4, tmp7, tmp8)
tmp10 = tmp0 >= tmp3
tmp11 = tl.full([1], 8, tl.int64)
tmp12 = tmp0 < tmp11
tmp13 = tl.load(in_ptr2 + ((4*x2) + ((-4) + x1)), tmp10 & xmask, eviction_policy='evict_last', other=0.0)
tmp14 = tl.where(tmp4, tmp9, tmp13)
tmp15 = tl.full([1], 0, tl.int32)
tmp16 = triton_helpers.maximum(tmp15, tmp14)
tl.store(out_ptr0 + (x3), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/mr/cmru2zu43k2oz2h4g44qjlddpct5ciovqgi673jkmq5bndjjakmk.py
# Topologically Sorted Source Nodes: [x_7, max_4], Original ATen: [aten.convolution, aten.max]
# Source node to ATen node mapping:
# max_4 => getitem_6, max_4
# x_7 => convolution_4
# Graph fragment:
# %convolution_4 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_3, %primals_10, %primals_11, [1], [0], [1], False, [0], 1), kwargs = {})
# %max_4 : [num_users=2] = call_function[target=torch.ops.aten.max.dim](args = (%convolution_4, 2), kwargs = {})
# %getitem_6 : [num_users=2] = call_function[target=operator.getitem](args = (%max_4, 0), kwargs = {})
triton_poi_fused_convolution_max_3 = async_compile.triton('triton_poi_fused_convolution_max_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i64', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_max_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_max_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (4*x2), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp5 = triton_helpers.maximum(tmp2, tmp4)
tmp7 = tmp6 + tmp1
tmp8 = triton_helpers.maximum(tmp5, tmp7)
tmp10 = tmp9 + tmp1
tmp11 = triton_helpers.maximum(tmp8, tmp10)
tmp12 = tmp2 > tmp4
tmp13 = tmp2 == tmp4
tmp14 = tmp2 != tmp2
tmp15 = tmp4 != tmp4
tmp16 = tmp14 > tmp15
tmp17 = tmp12 | tmp16
tmp18 = tmp14 & tmp15
tmp19 = tmp13 | tmp18
tmp20 = tl.full([1], 0, tl.int64)
tmp21 = tl.full([1], 1, tl.int64)
tmp22 = tmp20 < tmp21
tmp23 = tmp19 & tmp22
tmp24 = tmp17 | tmp23
tmp25 = tl.where(tmp24, tmp2, tmp4)
tmp26 = tl.where(tmp24, tmp20, tmp21)
tmp27 = tmp25 > tmp7
tmp28 = tmp25 == tmp7
tmp29 = tmp25 != tmp25
tmp30 = tmp7 != tmp7
tmp31 = tmp29 > tmp30
tmp32 = tmp27 | tmp31
tmp33 = tmp29 & tmp30
tmp34 = tmp28 | tmp33
tmp35 = tl.full([1], 2, tl.int64)
tmp36 = tmp26 < tmp35
tmp37 = tmp34 & tmp36
tmp38 = tmp32 | tmp37
tmp39 = tl.where(tmp38, tmp25, tmp7)
tmp40 = tl.where(tmp38, tmp26, tmp35)
tmp41 = tmp39 > tmp10
tmp42 = tmp39 == tmp10
tmp43 = tmp39 != tmp39
tmp44 = tmp10 != tmp10
tmp45 = tmp43 > tmp44
tmp46 = tmp41 | tmp45
tmp47 = tmp43 & tmp44
tmp48 = tmp42 | tmp47
tmp49 = tl.full([1], 3, tl.int64)
tmp50 = tmp40 < tmp49
tmp51 = tmp48 & tmp50
tmp52 = tmp46 | tmp51
tmp53 = tl.where(tmp52, tmp39, tmp10)
tmp54 = tl.where(tmp52, tmp40, tmp49)
tl.store(out_ptr0 + (x2), tmp11, xmask)
tl.store(out_ptr1 + (x2), tmp54, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13 = args
args.clear()
assert_size_stride(primals_1, (8, 4, 1), (4, 1, 1))
assert_size_stride(primals_2, (8, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_9, (4, ), (1, ))
assert_size_stride(primals_10, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_11, (4, ), (1, ))
assert_size_stride(primals_12, (4, 4), (4, 1))
assert_size_stride(primals_13, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4), (32, 4, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [x, relu], Original ATen: [aten.convolution, aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_0.run(buf1, primals_2, 128, grid=grid(128), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
buf3 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
buf4 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
# Topologically Sorted Source Nodes: [x_1, max_1], Original ATen: [aten.convolution, aten.max]
triton_poi_fused_convolution_max_1.run(buf2, primals_5, buf3, buf4, 16, grid=grid(16), stream=stream0)
buf5 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2, relu_1], Original ATen: [aten.cat, aten.relu]
triton_poi_fused_cat_relu_2.run(buf2, primals_5, buf4, buf5, 128, grid=grid(128), stream=stream0)
del buf2
del primals_5
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
buf6 = extern_kernels.convolution(buf5, primals_6, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf6, (4, 4, 4), (16, 4, 1))
buf7 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
buf8 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [x_3, max_2], Original ATen: [aten.convolution, aten.max]
triton_poi_fused_convolution_max_1.run(buf6, primals_7, buf7, buf8, 16, grid=grid(16), stream=stream0)
buf9 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_4, relu_2], Original ATen: [aten.cat, aten.relu]
triton_poi_fused_cat_relu_2.run(buf6, primals_7, buf8, buf9, 128, grid=grid(128), stream=stream0)
del buf6
del primals_7
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.convolution]
buf10 = extern_kernels.convolution(buf9, primals_8, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf10, (4, 4, 4), (16, 4, 1))
buf11 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
buf12 = buf8; del buf8 # reuse
# Topologically Sorted Source Nodes: [x_5, max_3], Original ATen: [aten.convolution, aten.max]
triton_poi_fused_convolution_max_1.run(buf10, primals_9, buf11, buf12, 16, grid=grid(16), stream=stream0)
buf13 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_6, relu_3], Original ATen: [aten.cat, aten.relu]
triton_poi_fused_cat_relu_2.run(buf10, primals_9, buf12, buf13, 128, grid=grid(128), stream=stream0)
del buf10
del primals_9
# Topologically Sorted Source Nodes: [x_7], Original ATen: [aten.convolution]
buf14 = extern_kernels.convolution(buf13, primals_10, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf14, (4, 4, 4), (16, 4, 1))
buf15 = reinterpret_tensor(buf12, (4, 4), (4, 1), 0); del buf12 # reuse
buf16 = empty_strided_cuda((4, 4), (4, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_7, max_4], Original ATen: [aten.convolution, aten.max]
triton_poi_fused_convolution_max_3.run(buf14, primals_11, buf15, buf16, 16, grid=grid(16), stream=stream0)
del buf14
del primals_11
buf17 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_9], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_13, buf15, reinterpret_tensor(primals_12, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf17)
del primals_13
return (buf17, primals_1, primals_3, primals_4, primals_6, primals_8, primals_10, buf1, buf3, buf5, buf7, buf9, buf11, buf13, buf15, primals_12, reinterpret_tensor(buf16, (4, 4, 1), (4, 1, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((8, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch.nn as nn
class Pointnet(nn.Module):
def __init__(self, in_channels, out_channels, hidden_dim, segmentation=
False):
super().__init__()
self.fc_in = nn.Conv1d(in_channels, 2 * hidden_dim, 1)
self.fc_0 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.fc_1 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.fc_2 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.fc_3 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.segmentation = segmentation
if segmentation:
self.fc_out = nn.Conv1d(2 * hidden_dim, out_channels, 1)
else:
self.fc_out = nn.Linear(hidden_dim, out_channels)
self.activation = nn.ReLU()
def forward(self, x):
x = self.fc_in(x)
x = self.fc_0(self.activation(x))
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
x = self.fc_1(self.activation(x))
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
x = self.fc_2(self.activation(x))
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
x = self.fc_3(self.activation(x))
if self.segmentation:
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
else:
x = torch.max(x, dim=2)[0]
x = self.fc_out(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'hidden_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.utils.data
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 8
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_max_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + 4 * x2, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x2), xmask, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr0 + (2 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp36 = tl.load(in_ptr0 + (3 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp5 = tmp2 > tmp4
tmp6 = tmp2 == tmp4
tmp7 = tmp2 != tmp2
tmp8 = tmp4 != tmp4
tmp9 = tmp7 > tmp8
tmp10 = tmp5 | tmp9
tmp11 = tmp7 & tmp8
tmp12 = tmp6 | tmp11
tmp13 = tl.full([1], 0, tl.int64)
tmp14 = tl.full([1], 1, tl.int64)
tmp15 = tmp13 < tmp14
tmp16 = tmp12 & tmp15
tmp17 = tmp10 | tmp16
tmp18 = tl.where(tmp17, tmp2, tmp4)
tmp19 = tl.where(tmp17, tmp13, tmp14)
tmp21 = tmp20 + tmp1
tmp22 = tmp18 > tmp21
tmp23 = tmp18 == tmp21
tmp24 = tmp18 != tmp18
tmp25 = tmp21 != tmp21
tmp26 = tmp24 > tmp25
tmp27 = tmp22 | tmp26
tmp28 = tmp24 & tmp25
tmp29 = tmp23 | tmp28
tmp30 = tl.full([1], 2, tl.int64)
tmp31 = tmp19 < tmp30
tmp32 = tmp29 & tmp31
tmp33 = tmp27 | tmp32
tmp34 = tl.where(tmp33, tmp18, tmp21)
tmp35 = tl.where(tmp33, tmp19, tmp30)
tmp37 = tmp36 + tmp1
tmp38 = tmp34 > tmp37
tmp39 = tmp34 == tmp37
tmp40 = tmp34 != tmp34
tmp41 = tmp37 != tmp37
tmp42 = tmp40 > tmp41
tmp43 = tmp38 | tmp42
tmp44 = tmp40 & tmp41
tmp45 = tmp39 | tmp44
tmp46 = tl.full([1], 3, tl.int64)
tmp47 = tmp35 < tmp46
tmp48 = tmp45 & tmp47
tmp49 = tmp43 | tmp48
tl.where(tmp49, tmp34, tmp37)
tmp51 = tl.where(tmp49, tmp35, tmp46)
tmp52 = triton_helpers.maximum(tmp2, tmp4)
tmp53 = triton_helpers.maximum(tmp52, tmp21)
tmp54 = triton_helpers.maximum(tmp53, tmp37)
tl.store(out_ptr0 + x2, tmp51, xmask)
tl.store(out_ptr1 + x2, tmp54, xmask)
@triton.jit
def triton_poi_fused_cat_relu_2(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 8
x0 = xindex % 4
x2 = xindex // 32
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 4 * x1 + 16 * x2), tmp4 & xmask, other=0.0)
tmp6 = tl.load(in_ptr1 + x1, tmp4 & xmask, eviction_policy='evict_last',
other=0.0)
tmp7 = tmp5 + tmp6
tmp8 = tl.full(tmp7.shape, 0.0, tmp7.dtype)
tmp9 = tl.where(tmp4, tmp7, tmp8)
tmp10 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp13 = tl.load(in_ptr2 + (4 * x2 + (-4 + x1)), tmp10 & xmask,
eviction_policy='evict_last', other=0.0)
tmp14 = tl.where(tmp4, tmp9, tmp13)
tmp15 = tl.full([1], 0, tl.int32)
tmp16 = triton_helpers.maximum(tmp15, tmp14)
tl.store(out_ptr0 + x3, tmp16, xmask)
@triton.jit
def triton_poi_fused_convolution_max_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + 4 * x2, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x2), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * x2), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * x2), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp5 = triton_helpers.maximum(tmp2, tmp4)
tmp7 = tmp6 + tmp1
tmp8 = triton_helpers.maximum(tmp5, tmp7)
tmp10 = tmp9 + tmp1
tmp11 = triton_helpers.maximum(tmp8, tmp10)
tmp12 = tmp2 > tmp4
tmp13 = tmp2 == tmp4
tmp14 = tmp2 != tmp2
tmp15 = tmp4 != tmp4
tmp16 = tmp14 > tmp15
tmp17 = tmp12 | tmp16
tmp18 = tmp14 & tmp15
tmp19 = tmp13 | tmp18
tmp20 = tl.full([1], 0, tl.int64)
tmp21 = tl.full([1], 1, tl.int64)
tmp22 = tmp20 < tmp21
tmp23 = tmp19 & tmp22
tmp24 = tmp17 | tmp23
tmp25 = tl.where(tmp24, tmp2, tmp4)
tmp26 = tl.where(tmp24, tmp20, tmp21)
tmp27 = tmp25 > tmp7
tmp28 = tmp25 == tmp7
tmp29 = tmp25 != tmp25
tmp30 = tmp7 != tmp7
tmp31 = tmp29 > tmp30
tmp32 = tmp27 | tmp31
tmp33 = tmp29 & tmp30
tmp34 = tmp28 | tmp33
tmp35 = tl.full([1], 2, tl.int64)
tmp36 = tmp26 < tmp35
tmp37 = tmp34 & tmp36
tmp38 = tmp32 | tmp37
tmp39 = tl.where(tmp38, tmp25, tmp7)
tmp40 = tl.where(tmp38, tmp26, tmp35)
tmp41 = tmp39 > tmp10
tmp42 = tmp39 == tmp10
tmp43 = tmp39 != tmp39
tmp44 = tmp10 != tmp10
tmp45 = tmp43 > tmp44
tmp46 = tmp41 | tmp45
tmp47 = tmp43 & tmp44
tmp48 = tmp42 | tmp47
tmp49 = tl.full([1], 3, tl.int64)
tmp50 = tmp40 < tmp49
tmp51 = tmp48 & tmp50
tmp52 = tmp46 | tmp51
tl.where(tmp52, tmp39, tmp10)
tmp54 = tl.where(tmp52, tmp40, tmp49)
tl.store(out_ptr0 + x2, tmp11, xmask)
tl.store(out_ptr1 + x2, tmp54, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13) = args
args.clear()
assert_size_stride(primals_1, (8, 4, 1), (4, 1, 1))
assert_size_stride(primals_2, (8,), (1,))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_9, (4,), (1,))
assert_size_stride(primals_10, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_11, (4,), (1,))
assert_size_stride(primals_12, (4, 4), (4, 1))
assert_size_stride(primals_13, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4), (32, 4, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_relu_0[grid(128)](buf1, primals_2, 128,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
buf3 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
buf4 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
triton_poi_fused_convolution_max_1[grid(16)](buf2, primals_5, buf3,
buf4, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf5 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_relu_2[grid(128)](buf2, primals_5, buf4, buf5,
128, XBLOCK=128, num_warps=4, num_stages=1)
del buf2
del primals_5
buf6 = extern_kernels.convolution(buf5, primals_6, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf6, (4, 4, 4), (16, 4, 1))
buf7 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
buf8 = buf4
del buf4
triton_poi_fused_convolution_max_1[grid(16)](buf6, primals_7, buf7,
buf8, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf9 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_relu_2[grid(128)](buf6, primals_7, buf8, buf9,
128, XBLOCK=128, num_warps=4, num_stages=1)
del buf6
del primals_7
buf10 = extern_kernels.convolution(buf9, primals_8, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf10, (4, 4, 4), (16, 4, 1))
buf11 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
buf12 = buf8
del buf8
triton_poi_fused_convolution_max_1[grid(16)](buf10, primals_9,
buf11, buf12, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf13 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_relu_2[grid(128)](buf10, primals_9, buf12,
buf13, 128, XBLOCK=128, num_warps=4, num_stages=1)
del buf10
del primals_9
buf14 = extern_kernels.convolution(buf13, primals_10, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf14, (4, 4, 4), (16, 4, 1))
buf15 = reinterpret_tensor(buf12, (4, 4), (4, 1), 0)
del buf12
buf16 = empty_strided_cuda((4, 4), (4, 1), torch.int64)
triton_poi_fused_convolution_max_3[grid(16)](buf14, primals_11,
buf15, buf16, 16, XBLOCK=16, num_warps=1, num_stages=1)
del buf14
del primals_11
buf17 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_13, buf15, reinterpret_tensor(
primals_12, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf17)
del primals_13
return (buf17, primals_1, primals_3, primals_4, primals_6, primals_8,
primals_10, buf1, buf3, buf5, buf7, buf9, buf11, buf13, buf15,
primals_12, reinterpret_tensor(buf16, (4, 4, 1), (4, 1, 1), 0))
class PointnetNew(nn.Module):
def __init__(self, in_channels, out_channels, hidden_dim, segmentation=
False):
super().__init__()
self.fc_in = nn.Conv1d(in_channels, 2 * hidden_dim, 1)
self.fc_0 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.fc_1 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.fc_2 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.fc_3 = nn.Conv1d(2 * hidden_dim, hidden_dim, 1)
self.segmentation = segmentation
if segmentation:
self.fc_out = nn.Conv1d(2 * hidden_dim, out_channels, 1)
else:
self.fc_out = nn.Linear(hidden_dim, out_channels)
self.activation = nn.ReLU()
def forward(self, input_0):
primals_1 = self.fc_in.weight
primals_2 = self.fc_in.bias
primals_4 = self.fc_0.weight
primals_5 = self.fc_0.bias
primals_6 = self.fc_1.weight
primals_7 = self.fc_1.bias
primals_8 = self.fc_2.weight
primals_9 = self.fc_2.bias
primals_10 = self.fc_3.weight
primals_11 = self.fc_3.bias
primals_12 = self.fc_out.weight
primals_13 = self.fc_out.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13])
return output[0]
|
StructuralNeurobiologyLab/LightConvPoint
|
Pointnet
| false | 14,458 |
[
"Apache-2.0"
] | 58 |
3f353f45e9e910fa390a74520dfd478e3e88f104
|
https://github.com/StructuralNeurobiologyLab/LightConvPoint/tree/3f353f45e9e910fa390a74520dfd478e3e88f104
|
ConvPlus
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/g6/cg6ggbgmjcrpyy6skkvmdrhjstximmnocghnc7q5sven6g5bhojo.py
# Topologically Sorted Source Nodes: [conv2d, conv2d_1, add], Original ATen: [aten.convolution, aten.add]
# Source node to ATen node mapping:
# add => add
# conv2d => convolution
# conv2d_1 => convolution_1
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [1, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_4, %primals_5, [1, 1], [0, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convolution, %convolution_1), kwargs = {})
triton_poi_fused_add_convolution_0 = async_compile.triton('triton_poi_fused_add_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_convolution_0(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (x3), xmask)
tmp4 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tl.store(in_out_ptr0 + (x3), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 1), (12, 3, 1, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 1, 3), (12, 3, 3, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(1, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4, 4), (64, 16, 4, 1))
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(primals_3, primals_4, stride=(1, 1), padding=(0, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
buf2 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv2d, conv2d_1, add], Original ATen: [aten.convolution, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_convolution_0.run(buf2, primals_2, buf1, primals_5, 256, grid=grid(256), stream=stream0)
del buf1
del primals_2
del primals_5
return (buf2, primals_1, primals_3, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 3, 1), (12, 3, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 1, 3), (12, 3, 3, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class ConvPlus(nn.Module):
def __init__(self, c1, c2, k=3, s=1, g=1, bias=True):
super(ConvPlus, self).__init__()
self.cv1 = nn.Conv2d(c1, c2, (k, 1), s, (k // 2, 0), groups=g, bias
=bias)
self.cv2 = nn.Conv2d(c1, c2, (1, k), s, (0, k // 2), groups=g, bias
=bias)
def forward(self, x):
return self.cv1(x) + self.cv2(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'c1': 4, 'c2': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
@triton.jit
def triton_poi_fused_add_convolution_0(in_out_ptr0, in_ptr0, in_ptr1,
in_ptr2, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + x3, xmask)
tmp4 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tl.store(in_out_ptr0 + x3, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 1), (12, 3, 1, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 1, 3), (12, 3, 3, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(1, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4, 4), (64, 16, 4, 1))
buf1 = extern_kernels.convolution(primals_3, primals_4, stride=(1,
1), padding=(0, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
buf2 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_add_convolution_0[grid(256)](buf2, primals_2, buf1,
primals_5, 256, XBLOCK=256, num_warps=4, num_stages=1)
del buf1
del primals_2
del primals_5
return buf2, primals_1, primals_3, primals_4
class ConvPlusNew(nn.Module):
def __init__(self, c1, c2, k=3, s=1, g=1, bias=True):
super(ConvPlusNew, self).__init__()
self.cv1 = nn.Conv2d(c1, c2, (k, 1), s, (k // 2, 0), groups=g, bias
=bias)
self.cv2 = nn.Conv2d(c1, c2, (1, k), s, (0, k // 2), groups=g, bias
=bias)
def forward(self, input_0):
primals_1 = self.cv1.weight
primals_2 = self.cv1.bias
primals_4 = self.cv2.weight
primals_5 = self.cv2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
Syencil/mobile-yolov5-pruning-distillation
|
ConvPlus
| false | 14,459 |
[
"MIT"
] | 554 |
5d52454bb397ae49677b5da398e4192abc681325
|
https://github.com/Syencil/mobile-yolov5-pruning-distillation/tree/5d52454bb397ae49677b5da398e4192abc681325
|
attentionLayer
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/3l/c3lu4ccbjruychszpewk67ythz75gaj4rslgmbux6fatrywe7g7t.py
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.add]
# Source node to ATen node mapping:
# multi_head_attention_forward => add_2
# Graph fragment:
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_5, %getitem_5), kwargs = {})
triton_poi_fused_add_0 = async_compile.triton('triton_poi_fused_add_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (8 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ol/colbiyeeegfdyyzeckjnylgg3xt3rkh3aadcz7fjtfx5472nedsg.py
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# multi_head_attention_forward => mul
# Graph fragment:
# %mul : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%permute_5, 1.0), kwargs = {})
triton_poi_fused_mul_1 = async_compile.triton('triton_poi_fused_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/c5/cc5cm2utkmzhcjdhw5qgs7t254ixwfil74kthoebdprvhlljdmul.py
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.add]
# Source node to ATen node mapping:
# multi_head_attention_forward => add_1
# Graph fragment:
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_3, %getitem_4), kwargs = {})
triton_poi_fused_add_2 = async_compile.triton('triton_poi_fused_add_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/7s/c7spagnqvsgjrukyw5jujzjmswxuigeuvpyhxgdob766q2gfvgzr.py
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# multi_head_attention_forward => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%bmm, [-1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%bmm, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_3 = async_compile.triton('triton_poi_fused__softmax_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/dw/cdwqsjnh2osfmjr2utzzaqdg2vrfivzkuhareq3urgidllj2bsvr.py
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# multi_head_attention_forward => div, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_4 = async_compile.triton('triton_poi_fused__softmax_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/y5/cy5gjrtl7netbzcjhig66pdorub2vbq2qvwmv3tamld2ehimmlz7.py
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# multi_head_attention_forward => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%permute_9,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_5 = async_compile.triton('triton_poi_fused_clone_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 4
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x1)), xmask & ymask)
tl.store(out_ptr0 + (x1 + (4*y0)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/vv/cvvbofouoyxjzf3vusx62ssynyztxp4c2lk76gekn6hxry2vdjht.py
# Topologically Sorted Source Nodes: [src_1, src_2], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# src_1 => add_3
# src_2 => clone_2
# Graph fragment:
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%permute, %squeeze), kwargs = {})
# %clone_2 : [num_users=3] = call_function[target=torch.ops.aten.clone.default](args = (%add_3,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_add_native_layer_norm_6 = async_compile.triton('triton_poi_fused_add_native_layer_norm_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4, 4], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_6', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_6(in_out_ptr0, in_ptr0, in_ptr1, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 4
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x1)), xmask & ymask)
tmp1 = tl.load(in_out_ptr0 + (x1 + (4*y0)), xmask & ymask)
tmp2 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + (x1 + (4*y0)), tmp4, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/a4/ca4uo2wxiyozqosox66kdczthghpe6qlxcx3xgdhcs2i6efpvbaa.py
# Topologically Sorted Source Nodes: [src_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# src_2 => add_4, rsqrt, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%clone_2, [1]), kwargs = {correction: 0, keepdim: True})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_6, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_4,), kwargs = {})
triton_poi_fused_native_layer_norm_7 = async_compile.triton('triton_poi_fused_native_layer_norm_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_7(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/z2/cz2x6uoruxdkowfiawvku3nxpx4vgbvkygpozaunzozrsmzk6q75.py
# Topologically Sorted Source Nodes: [src_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# src_2 => add_4, add_5, mul_1, mul_2, rsqrt, sub_1, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%clone_2, [1]), kwargs = {correction: 0, keepdim: True})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_6, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_4,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%clone_2, %getitem_7), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %rsqrt), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_1, %primals_7), kwargs = {})
# %add_5 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_2, %primals_8), kwargs = {})
triton_poi_fused_native_layer_norm_8 = async_compile.triton('triton_poi_fused_native_layer_norm_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_8(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/q6/cq6bj5cbehpfrboy2hqptxwsffz2qanka4axd2qviiot545u347l.py
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# relu => relu
# Graph fragment:
# %add_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default_1, %primals_10), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor_1,), kwargs = {})
triton_poi_fused_relu_9 = async_compile.triton('triton_poi_fused_relu_9', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_9', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_9(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/4e/c4escp7v3hikmzkxuyjmzdj3juzy4ty3yheqrkztwu2gvsteirte.py
# Topologically Sorted Source Nodes: [src_3], Original ATen: [aten.add]
# Source node to ATen node mapping:
# src_3 => add_6
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_12), kwargs = {})
# %add_6 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_5, %add_tensor), kwargs = {})
triton_poi_fused_add_10 = async_compile.triton('triton_poi_fused_add_10', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_10', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_10(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_out_ptr0 + (x2), xmask)
tmp2 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (12, 4), (4, 1))
assert_size_stride(primals_4, (12, ), (1, ))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, ), (1, ))
assert_size_stride(primals_9, (16, 4), (4, 1))
assert_size_stride(primals_10, (16, ), (1, ))
assert_size_stride(primals_11, (4, 16), (16, 1))
assert_size_stride(primals_12, (4, ), (1, ))
assert_size_stride(primals_13, (4, ), (1, ))
assert_size_stride(primals_14, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf0)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 16), out=buf1)
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 32), out=buf2)
del primals_3
buf3 = reinterpret_tensor(buf2, (4, 1, 4), (4, 4, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_0.run(buf3, primals_4, 16, grid=grid(16), stream=stream0)
buf4 = reinterpret_tensor(buf0, (4, 4, 1), (1, 4, 16), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.mul]
triton_poi_fused_mul_1.run(buf4, primals_4, 16, grid=grid(16), stream=stream0)
buf5 = reinterpret_tensor(buf1, (4, 1, 4), (4, 4, 1), 0); del buf1 # reuse
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.add]
triton_poi_fused_add_2.run(buf5, primals_4, 16, grid=grid(16), stream=stream0)
del primals_4
buf6 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.bmm]
extern_kernels.bmm(buf4, reinterpret_tensor(buf5, (4, 1, 4), (1, 0, 4), 0), out=buf6)
buf7 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf6, buf7, 64, grid=grid(64), stream=stream0)
buf8 = buf6; del buf6 # reuse
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten._softmax]
triton_poi_fused__softmax_4.run(buf7, buf8, 64, grid=grid(64), stream=stream0)
buf9 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.bmm]
extern_kernels.bmm(buf8, reinterpret_tensor(buf3, (4, 4, 1), (1, 4, 0), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [multi_head_attention_forward], Original ATen: [aten.clone]
triton_poi_fused_clone_5.run(buf9, buf10, 4, 4, grid=grid(4, 4), stream=stream0)
buf11 = reinterpret_tensor(buf9, (4, 4), (4, 1), 0); del buf9 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf10, (4, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf11)
buf12 = buf11; del buf11 # reuse
# Topologically Sorted Source Nodes: [src_1, src_2], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_6.run(buf12, primals_1, primals_6, 4, 4, grid=grid(4, 4), stream=stream0)
del primals_6
buf13 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf14 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
# Topologically Sorted Source Nodes: [src_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_7.run(buf12, buf13, buf14, 4, grid=grid(4), stream=stream0)
buf15 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [src_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_8.run(buf12, buf13, buf14, primals_7, primals_8, buf15, 16, grid=grid(16), stream=stream0)
del primals_8
buf16 = reinterpret_tensor(buf7, (4, 16), (16, 1), 0); del buf7 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf15, reinterpret_tensor(primals_9, (4, 16), (1, 4), 0), out=buf16)
buf17 = buf16; del buf16 # reuse
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu]
triton_poi_fused_relu_9.run(buf17, primals_10, 64, grid=grid(64), stream=stream0)
del primals_10
buf18 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf17, reinterpret_tensor(primals_11, (16, 4), (1, 16), 0), out=buf18)
buf19 = buf18; del buf18 # reuse
# Topologically Sorted Source Nodes: [src_3], Original ATen: [aten.add]
triton_poi_fused_add_10.run(buf19, buf15, primals_12, 16, grid=grid(16), stream=stream0)
del primals_12
buf20 = buf14; del buf14 # reuse
buf21 = buf13; del buf13 # reuse
# Topologically Sorted Source Nodes: [src_4], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_7.run(buf19, buf20, buf21, 4, grid=grid(4), stream=stream0)
buf22 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [src_4], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_8.run(buf19, buf20, buf21, primals_13, primals_14, buf22, 16, grid=grid(16), stream=stream0)
del buf20
del buf21
del primals_14
return (reinterpret_tensor(buf22, (4, 4), (1, 4), 0), primals_7, primals_13, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), buf8, reinterpret_tensor(buf10, (4, 4), (4, 1), 0), buf12, buf15, buf17, buf19, primals_11, primals_9, primals_5, reinterpret_tensor(buf3, (4, 1, 4), (1, 1, 4), 0), reinterpret_tensor(buf4, (4, 1, 4), (1, 1, 4), 0), reinterpret_tensor(buf5, (4, 4, 1), (1, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((12, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((12, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((16, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((16, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, 16), (16, 1), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_14 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import MultiheadAttention
from itertools import product as product
class attentionLayer(nn.Module):
def __init__(self, d_model, nhead, dropout=0.1):
super(attentionLayer, self).__init__()
self.self_attn = MultiheadAttention(d_model, nhead, dropout=dropout)
self.linear1 = nn.Linear(d_model, d_model * 4)
self.dropout = nn.Dropout(dropout)
self.linear2 = nn.Linear(d_model * 4, d_model)
self.norm1 = nn.LayerNorm(d_model)
self.norm2 = nn.LayerNorm(d_model)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.activation = F.relu
def forward(self, src, tar):
src = src.transpose(0, 1)
tar = tar.transpose(0, 1)
src2 = self.self_attn(tar, src, src, attn_mask=None,
key_padding_mask=None)[0]
src = src + self.dropout1(src2)
src = self.norm1(src)
src2 = self.linear2(self.dropout(self.activation(self.linear1(src))))
src = src + self.dropout2(src2)
src = self.norm2(src)
src = src.transpose(0, 1)
return src
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'d_model': 4, 'nhead': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import MultiheadAttention
from itertools import product as product
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + (8 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_mul_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + (4 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_clone_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 4
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x1), xmask & ymask)
tl.store(out_ptr0 + (x1 + 4 * y0), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_6(in_out_ptr0, in_ptr0, in_ptr1,
ynumel, xnumel, YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 4
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x1), xmask & ymask)
tmp1 = tl.load(in_out_ptr0 + (x1 + 4 * y0), xmask & ymask)
tmp2 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + (x1 + 4 * y0), tmp4, xmask & ymask)
@triton.jit
def triton_poi_fused_native_layer_norm_7(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_8(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_relu_9(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_10(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK:
tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_out_ptr0 + x2, xmask)
tmp2 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + x2, tmp4, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13, primals_14) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (12, 4), (4, 1))
assert_size_stride(primals_4, (12,), (1,))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4,), (1,))
assert_size_stride(primals_9, (16, 4), (4, 1))
assert_size_stride(primals_10, (16,), (1,))
assert_size_stride(primals_11, (4, 16), (16, 1))
assert_size_stride(primals_12, (4,), (1,))
assert_size_stride(primals_13, (4,), (1,))
assert_size_stride(primals_14, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (4, 4), (1, 4), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf0)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (4, 4), (1, 4), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 16), out=buf1)
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (4, 4), (1, 4), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 32), out=buf2)
del primals_3
buf3 = reinterpret_tensor(buf2, (4, 1, 4), (4, 4, 1), 0)
del buf2
get_raw_stream(0)
triton_poi_fused_add_0[grid(16)](buf3, primals_4, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf4 = reinterpret_tensor(buf0, (4, 4, 1), (1, 4, 16), 0)
del buf0
triton_poi_fused_mul_1[grid(16)](buf4, primals_4, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf5 = reinterpret_tensor(buf1, (4, 1, 4), (4, 4, 1), 0)
del buf1
triton_poi_fused_add_2[grid(16)](buf5, primals_4, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del primals_4
buf6 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(buf4, reinterpret_tensor(buf5, (4, 1, 4), (1, 0,
4), 0), out=buf6)
buf7 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_3[grid(64)](buf6, buf7, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf8 = buf6
del buf6
triton_poi_fused__softmax_4[grid(64)](buf7, buf8, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf9 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf8, reinterpret_tensor(buf3, (4, 4, 1), (1, 4,
0), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
triton_poi_fused_clone_5[grid(4, 4)](buf9, buf10, 4, 4, XBLOCK=4,
YBLOCK=4, num_warps=1, num_stages=1)
buf11 = reinterpret_tensor(buf9, (4, 4), (4, 1), 0)
del buf9
extern_kernels.mm(reinterpret_tensor(buf10, (4, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf11)
buf12 = buf11
del buf11
triton_poi_fused_add_native_layer_norm_6[grid(4, 4)](buf12,
primals_1, primals_6, 4, 4, XBLOCK=4, YBLOCK=4, num_warps=1,
num_stages=1)
del primals_6
buf13 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf14 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
triton_poi_fused_native_layer_norm_7[grid(4)](buf12, buf13, buf14,
4, XBLOCK=4, num_warps=1, num_stages=1)
buf15 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
triton_poi_fused_native_layer_norm_8[grid(16)](buf12, buf13, buf14,
primals_7, primals_8, buf15, 16, XBLOCK=16, num_warps=1,
num_stages=1)
del primals_8
buf16 = reinterpret_tensor(buf7, (4, 16), (16, 1), 0)
del buf7
extern_kernels.mm(buf15, reinterpret_tensor(primals_9, (4, 16), (1,
4), 0), out=buf16)
buf17 = buf16
del buf16
triton_poi_fused_relu_9[grid(64)](buf17, primals_10, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_10
buf18 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf17, reinterpret_tensor(primals_11, (16, 4), (1,
16), 0), out=buf18)
buf19 = buf18
del buf18
triton_poi_fused_add_10[grid(16)](buf19, buf15, primals_12, 16,
XBLOCK=16, num_warps=1, num_stages=1)
del primals_12
buf20 = buf14
del buf14
buf21 = buf13
del buf13
triton_poi_fused_native_layer_norm_7[grid(4)](buf19, buf20, buf21,
4, XBLOCK=4, num_warps=1, num_stages=1)
buf22 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
triton_poi_fused_native_layer_norm_8[grid(16)](buf19, buf20, buf21,
primals_13, primals_14, buf22, 16, XBLOCK=16, num_warps=1,
num_stages=1)
del buf20
del buf21
del primals_14
return (reinterpret_tensor(buf22, (4, 4), (1, 4), 0), primals_7,
primals_13, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), buf8,
reinterpret_tensor(buf10, (4, 4), (4, 1), 0), buf12, buf15, buf17,
buf19, primals_11, primals_9, primals_5, reinterpret_tensor(buf3, (
4, 1, 4), (1, 1, 4), 0), reinterpret_tensor(buf4, (4, 1, 4), (1, 1,
4), 0), reinterpret_tensor(buf5, (4, 4, 1), (1, 4, 1), 0))
class attentionLayerNew(nn.Module):
def __init__(self, d_model, nhead, dropout=0.1):
super(attentionLayerNew, self).__init__()
self.self_attn = MultiheadAttention(d_model, nhead, dropout=dropout)
self.linear1 = nn.Linear(d_model, d_model * 4)
self.dropout = nn.Dropout(dropout)
self.linear2 = nn.Linear(d_model * 4, d_model)
self.norm1 = nn.LayerNorm(d_model)
self.norm2 = nn.LayerNorm(d_model)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.activation = F.relu
def forward(self, input_0, input_1):
primals_3 = self.self_attn.in_proj_weight
primals_4 = self.self_attn.in_proj_bias
primals_1 = self.self_attn.out_proj.weight
primals_6 = self.self_attn.out_proj.bias
primals_9 = self.linear1.weight
primals_10 = self.linear1.bias
primals_11 = self.linear2.weight
primals_7 = self.linear2.bias
primals_8 = self.norm1.weight
primals_12 = self.norm1.bias
primals_13 = self.norm2.weight
primals_14 = self.norm2.bias
primals_2 = input_0
primals_5 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13, primals_14])
return output[0]
|
TaoRuijie/TalkNet_ASD
|
attentionLayer
| false | 14,460 |
[
"MIT"
] | 79 |
4a2bc4859ee192ab450eaf63937a799212f2b021
|
https://github.com/TaoRuijie/TalkNet_ASD/tree/4a2bc4859ee192ab450eaf63937a799212f2b021
|
GlobalLayerNorm
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ap/cap3nkzqkfj7rlygvuozblog2seuhvlea6gxfoq5gqtaswv2kdk3.py
# Topologically Sorted Source Nodes: [mean, mean_2, sub, pow_1, mean_3, var], Original ATen: [aten.mean, aten.sub, aten.pow]
# Source node to ATen node mapping:
# mean => mean
# mean_2 => mean_1
# mean_3 => mean_2
# pow_1 => pow_1
# sub => sub
# var => mean_3
# Graph fragment:
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [1], True), kwargs = {})
# %mean_1 : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%mean, [2], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean_1), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub, 2), kwargs = {})
# %mean_2 : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%pow_1, [1], True), kwargs = {})
# %mean_3 : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%mean_2, [2], True), kwargs = {})
triton_poi_fused_mean_pow_sub_0 = async_compile.triton('triton_poi_fused_mean_pow_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mean_pow_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 16, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mean_pow_sub_0(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (64*x1)), xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0 + (64*x1)), xmask)
tmp3 = tl.load(in_ptr0 + (32 + x0 + (64*x1)), xmask)
tmp5 = tl.load(in_ptr0 + (48 + x0 + (64*x1)), xmask)
tmp9 = tl.load(in_ptr0 + (4 + x0 + (64*x1)), xmask)
tmp10 = tl.load(in_ptr0 + (20 + x0 + (64*x1)), xmask)
tmp12 = tl.load(in_ptr0 + (36 + x0 + (64*x1)), xmask)
tmp14 = tl.load(in_ptr0 + (52 + x0 + (64*x1)), xmask)
tmp18 = tl.load(in_ptr0 + (8 + x0 + (64*x1)), xmask)
tmp19 = tl.load(in_ptr0 + (24 + x0 + (64*x1)), xmask)
tmp21 = tl.load(in_ptr0 + (40 + x0 + (64*x1)), xmask)
tmp23 = tl.load(in_ptr0 + (56 + x0 + (64*x1)), xmask)
tmp27 = tl.load(in_ptr0 + (12 + x0 + (64*x1)), xmask)
tmp28 = tl.load(in_ptr0 + (28 + x0 + (64*x1)), xmask)
tmp30 = tl.load(in_ptr0 + (44 + x0 + (64*x1)), xmask)
tmp32 = tl.load(in_ptr0 + (60 + x0 + (64*x1)), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp11 = tmp9 + tmp10
tmp13 = tmp11 + tmp12
tmp15 = tmp13 + tmp14
tmp16 = tmp15 / tmp7
tmp17 = tmp8 + tmp16
tmp20 = tmp18 + tmp19
tmp22 = tmp20 + tmp21
tmp24 = tmp22 + tmp23
tmp25 = tmp24 / tmp7
tmp26 = tmp17 + tmp25
tmp29 = tmp27 + tmp28
tmp31 = tmp29 + tmp30
tmp33 = tmp31 + tmp32
tmp34 = tmp33 / tmp7
tmp35 = tmp26 + tmp34
tmp36 = tmp35 / tmp7
tmp37 = tmp0 - tmp36
tmp38 = tmp37 * tmp37
tmp39 = tmp1 - tmp36
tmp40 = tmp39 * tmp39
tmp41 = tmp38 + tmp40
tmp42 = tmp3 - tmp36
tmp43 = tmp42 * tmp42
tmp44 = tmp41 + tmp43
tmp45 = tmp5 - tmp36
tmp46 = tmp45 * tmp45
tmp47 = tmp44 + tmp46
tmp48 = tmp47 / tmp7
tmp49 = tmp9 - tmp36
tmp50 = tmp49 * tmp49
tmp51 = tmp10 - tmp36
tmp52 = tmp51 * tmp51
tmp53 = tmp50 + tmp52
tmp54 = tmp12 - tmp36
tmp55 = tmp54 * tmp54
tmp56 = tmp53 + tmp55
tmp57 = tmp14 - tmp36
tmp58 = tmp57 * tmp57
tmp59 = tmp56 + tmp58
tmp60 = tmp59 / tmp7
tmp61 = tmp48 + tmp60
tmp62 = tmp18 - tmp36
tmp63 = tmp62 * tmp62
tmp64 = tmp19 - tmp36
tmp65 = tmp64 * tmp64
tmp66 = tmp63 + tmp65
tmp67 = tmp21 - tmp36
tmp68 = tmp67 * tmp67
tmp69 = tmp66 + tmp68
tmp70 = tmp23 - tmp36
tmp71 = tmp70 * tmp70
tmp72 = tmp69 + tmp71
tmp73 = tmp72 / tmp7
tmp74 = tmp61 + tmp73
tmp75 = tmp27 - tmp36
tmp76 = tmp75 * tmp75
tmp77 = tmp28 - tmp36
tmp78 = tmp77 * tmp77
tmp79 = tmp76 + tmp78
tmp80 = tmp30 - tmp36
tmp81 = tmp80 * tmp80
tmp82 = tmp79 + tmp81
tmp83 = tmp32 - tmp36
tmp84 = tmp83 * tmp83
tmp85 = tmp82 + tmp84
tmp86 = tmp85 / tmp7
tmp87 = tmp74 + tmp86
tmp88 = tmp87 / tmp7
tl.store(out_ptr0 + (x2), tmp36, xmask)
tl.store(out_ptr1 + (x2), tmp88, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/tu/ctunck2watc7guogb7nwwbhl2evcmo3mr4fhhl6ajujqj74x6txp.py
# Topologically Sorted Source Nodes: [sub, mul, add, pow_2, truediv, gLN_y], Original ATen: [aten.sub, aten.mul, aten.add, aten.pow, aten.div]
# Source node to ATen node mapping:
# add => add
# gLN_y => add_1
# mul => mul
# pow_2 => pow_2
# sub => sub
# truediv => div
# Graph fragment:
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_2, %sub), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean_3, 1e-08), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%add, 0.5), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul, %pow_2), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%div, %primals_3), kwargs = {})
triton_poi_fused_add_div_mul_pow_sub_1 = async_compile.triton('triton_poi_fused_add_div_mul_pow_sub_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_pow_sub_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mul_pow_sub_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x4 = xindex
x0 = xindex % 4
x3 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x4), xmask)
tmp2 = tl.load(in_ptr2 + (x0 + (4*x3)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0 + (4*x3)), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 - tmp2
tmp4 = tmp0 * tmp3
tmp6 = 1e-08
tmp7 = tmp5 + tmp6
tmp8 = libdevice.sqrt(tmp7)
tmp9 = tmp4 / tmp8
tmp11 = tmp9 + tmp10
tl.store(out_ptr0 + (x4), tmp11, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1), (4, 1, 1))
assert_size_stride(primals_3, (1, 4, 1), (4, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1, 1, 4), (4, 16, 16, 1), torch.float32)
buf1 = empty_strided_cuda((4, 1, 1, 4), (4, 16, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [mean, mean_2, sub, pow_1, mean_3, var], Original ATen: [aten.mean, aten.sub, aten.pow]
stream0 = get_raw_stream(0)
triton_poi_fused_mean_pow_sub_0.run(primals_1, buf0, buf1, 16, grid=grid(16), stream=stream0)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [sub, mul, add, pow_2, truediv, gLN_y], Original ATen: [aten.sub, aten.mul, aten.add, aten.pow, aten.div]
triton_poi_fused_add_div_mul_pow_sub_1.run(primals_2, primals_1, buf0, buf1, primals_3, buf2, 256, grid=grid(256), stream=stream0)
del buf0
del buf1
del primals_2
del primals_3
return (buf2, primals_1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((1, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
from itertools import product as product
class GlobalLayerNorm(nn.Module):
def __init__(self, channel_size):
super(GlobalLayerNorm, self).__init__()
self.gamma = nn.Parameter(torch.Tensor(1, channel_size, 1))
self.beta = nn.Parameter(torch.Tensor(1, channel_size, 1))
self.reset_parameters()
def reset_parameters(self):
self.gamma.data.fill_(1)
self.beta.data.zero_()
def forward(self, y):
mean = y.mean(dim=1, keepdim=True).mean(dim=2, keepdim=True)
var = torch.pow(y - mean, 2).mean(dim=1, keepdim=True).mean(dim=2,
keepdim=True)
gLN_y = self.gamma * (y - mean) / torch.pow(var + 1e-08, 0.5
) + self.beta
return gLN_y
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channel_size': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
from itertools import product as product
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_mean_pow_sub_0(in_ptr0, out_ptr0, out_ptr1, xnumel,
XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 64 * x1), xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0 + 64 * x1), xmask)
tmp3 = tl.load(in_ptr0 + (32 + x0 + 64 * x1), xmask)
tmp5 = tl.load(in_ptr0 + (48 + x0 + 64 * x1), xmask)
tmp9 = tl.load(in_ptr0 + (4 + x0 + 64 * x1), xmask)
tmp10 = tl.load(in_ptr0 + (20 + x0 + 64 * x1), xmask)
tmp12 = tl.load(in_ptr0 + (36 + x0 + 64 * x1), xmask)
tmp14 = tl.load(in_ptr0 + (52 + x0 + 64 * x1), xmask)
tmp18 = tl.load(in_ptr0 + (8 + x0 + 64 * x1), xmask)
tmp19 = tl.load(in_ptr0 + (24 + x0 + 64 * x1), xmask)
tmp21 = tl.load(in_ptr0 + (40 + x0 + 64 * x1), xmask)
tmp23 = tl.load(in_ptr0 + (56 + x0 + 64 * x1), xmask)
tmp27 = tl.load(in_ptr0 + (12 + x0 + 64 * x1), xmask)
tmp28 = tl.load(in_ptr0 + (28 + x0 + 64 * x1), xmask)
tmp30 = tl.load(in_ptr0 + (44 + x0 + 64 * x1), xmask)
tmp32 = tl.load(in_ptr0 + (60 + x0 + 64 * x1), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp11 = tmp9 + tmp10
tmp13 = tmp11 + tmp12
tmp15 = tmp13 + tmp14
tmp16 = tmp15 / tmp7
tmp17 = tmp8 + tmp16
tmp20 = tmp18 + tmp19
tmp22 = tmp20 + tmp21
tmp24 = tmp22 + tmp23
tmp25 = tmp24 / tmp7
tmp26 = tmp17 + tmp25
tmp29 = tmp27 + tmp28
tmp31 = tmp29 + tmp30
tmp33 = tmp31 + tmp32
tmp34 = tmp33 / tmp7
tmp35 = tmp26 + tmp34
tmp36 = tmp35 / tmp7
tmp37 = tmp0 - tmp36
tmp38 = tmp37 * tmp37
tmp39 = tmp1 - tmp36
tmp40 = tmp39 * tmp39
tmp41 = tmp38 + tmp40
tmp42 = tmp3 - tmp36
tmp43 = tmp42 * tmp42
tmp44 = tmp41 + tmp43
tmp45 = tmp5 - tmp36
tmp46 = tmp45 * tmp45
tmp47 = tmp44 + tmp46
tmp48 = tmp47 / tmp7
tmp49 = tmp9 - tmp36
tmp50 = tmp49 * tmp49
tmp51 = tmp10 - tmp36
tmp52 = tmp51 * tmp51
tmp53 = tmp50 + tmp52
tmp54 = tmp12 - tmp36
tmp55 = tmp54 * tmp54
tmp56 = tmp53 + tmp55
tmp57 = tmp14 - tmp36
tmp58 = tmp57 * tmp57
tmp59 = tmp56 + tmp58
tmp60 = tmp59 / tmp7
tmp61 = tmp48 + tmp60
tmp62 = tmp18 - tmp36
tmp63 = tmp62 * tmp62
tmp64 = tmp19 - tmp36
tmp65 = tmp64 * tmp64
tmp66 = tmp63 + tmp65
tmp67 = tmp21 - tmp36
tmp68 = tmp67 * tmp67
tmp69 = tmp66 + tmp68
tmp70 = tmp23 - tmp36
tmp71 = tmp70 * tmp70
tmp72 = tmp69 + tmp71
tmp73 = tmp72 / tmp7
tmp74 = tmp61 + tmp73
tmp75 = tmp27 - tmp36
tmp76 = tmp75 * tmp75
tmp77 = tmp28 - tmp36
tmp78 = tmp77 * tmp77
tmp79 = tmp76 + tmp78
tmp80 = tmp30 - tmp36
tmp81 = tmp80 * tmp80
tmp82 = tmp79 + tmp81
tmp83 = tmp32 - tmp36
tmp84 = tmp83 * tmp83
tmp85 = tmp82 + tmp84
tmp86 = tmp85 / tmp7
tmp87 = tmp74 + tmp86
tmp88 = tmp87 / tmp7
tl.store(out_ptr0 + x2, tmp36, xmask)
tl.store(out_ptr1 + x2, tmp88, xmask)
@triton.jit
def triton_poi_fused_add_div_mul_pow_sub_1(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x4 = xindex
x0 = xindex % 4
x3 = xindex // 64
tmp0 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x4, xmask)
tmp2 = tl.load(in_ptr2 + (x0 + 4 * x3), xmask, eviction_policy='evict_last'
)
tmp5 = tl.load(in_ptr3 + (x0 + 4 * x3), xmask, eviction_policy='evict_last'
)
tmp10 = tl.load(in_ptr4 + x1, xmask, eviction_policy='evict_last')
tmp3 = tmp1 - tmp2
tmp4 = tmp0 * tmp3
tmp6 = 1e-08
tmp7 = tmp5 + tmp6
tmp8 = libdevice.sqrt(tmp7)
tmp9 = tmp4 / tmp8
tmp11 = tmp9 + tmp10
tl.store(out_ptr0 + x4, tmp11, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1), (4, 1, 1))
assert_size_stride(primals_3, (1, 4, 1), (4, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1, 1, 4), (4, 16, 16, 1), torch.float32)
buf1 = empty_strided_cuda((4, 1, 1, 4), (4, 16, 16, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mean_pow_sub_0[grid(16)](primals_1, buf0, buf1, 16,
XBLOCK=16, num_warps=1, num_stages=1)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_div_mul_pow_sub_1[grid(256)](primals_2,
primals_1, buf0, buf1, primals_3, buf2, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del buf0
del buf1
del primals_2
del primals_3
return buf2, primals_1
class GlobalLayerNormNew(nn.Module):
def __init__(self, channel_size):
super(GlobalLayerNormNew, self).__init__()
self.gamma = nn.Parameter(torch.Tensor(1, channel_size, 1))
self.beta = nn.Parameter(torch.Tensor(1, channel_size, 1))
self.reset_parameters()
def reset_parameters(self):
self.gamma.data.fill_(1)
self.beta.data.zero_()
def forward(self, input_0):
primals_2 = self.gamma
primals_3 = self.beta
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
TaoRuijie/TalkNet_ASD
|
GlobalLayerNorm
| false | 14,461 |
[
"MIT"
] | 79 |
4a2bc4859ee192ab450eaf63937a799212f2b021
|
https://github.com/TaoRuijie/TalkNet_ASD/tree/4a2bc4859ee192ab450eaf63937a799212f2b021
|
adaILN
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/um/cumfxxnuff67646tii7qigo5xu4wp3dwgamqagptzcoe6havnpgp.py
# Topologically Sorted Source Nodes: [ln_mean, ln_var, add_1, sqrt_1], Original ATen: [aten.mean, aten.var, aten.add, aten.sqrt]
# Source node to ATen node mapping:
# add_1 => add_1
# ln_mean => mean_1
# ln_var => var_1
# sqrt_1 => sqrt_1
# Graph fragment:
# %mean_1 : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [1, 2, 3], True), kwargs = {})
# %var_1 : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%primals_1, [1, 2, 3]), kwargs = {correction: 1, keepdim: True})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%var_1, 1e-05), kwargs = {})
# %sqrt_1 : [num_users=2] = call_function[target=torch.ops.aten.sqrt.default](args = (%add_1,), kwargs = {})
triton_per_fused_add_mean_sqrt_var_0 = async_compile.triton('triton_per_fused_add_mean_sqrt_var_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_mean_sqrt_var_0', 'mutated_arg_names': ['in_out_ptr0', 'in_out_ptr1'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 4, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_mean_sqrt_var_0(in_out_ptr0, in_out_ptr1, in_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 64, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 64.0
tmp20 = tmp4 / tmp19
tmp21 = 63.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + (x0), tmp25, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/3n/c3ndqo7nij3t5k6ubr2vtyqzix6kntc5ziep6mlsj6ad5b7y77jj.py
# Topologically Sorted Source Nodes: [in_mean, in_var, sub, add, sqrt, out_in, sub_1, out_ln, mul, sub_2, mul_1, out], Original ATen: [aten.mean, aten.var, aten.sub, aten.add, aten.sqrt, aten.div, aten.mul, aten.rsub]
# Source node to ATen node mapping:
# add => add
# in_mean => mean
# in_var => var
# mul => mul
# mul_1 => mul_1
# out => add_2
# out_in => div
# out_ln => div_1
# sqrt => sqrt
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# Graph fragment:
# %mean : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [2, 3], True), kwargs = {})
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%primals_1, [2, 3]), kwargs = {correction: 1, keepdim: True})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%var, 1e-05), kwargs = {})
# %sqrt : [num_users=2] = call_function[target=torch.ops.aten.sqrt.default](args = (%add,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %sqrt), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean_1), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_1, %sqrt_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%expand, %div), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %expand), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_2, %div_1), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1 = async_compile.triton('triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[16, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32', 8: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1', 'mutated_arg_names': ['in_out_ptr0', 'in_out_ptr1'], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 4, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1(in_out_ptr0, in_out_ptr1, in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 16
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
x2 = xindex % 4
x3 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (r1 + (16*x0)), xmask, other=0.0)
tmp26 = tl.load(in_ptr1 + (x2), xmask, eviction_policy='evict_last')
tmp32 = tl.load(in_ptr2 + (x3), xmask, eviction_policy='evict_last')
tmp34 = tl.load(in_ptr3 + (x3), xmask, eviction_policy='evict_last')
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 16, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 16.0
tmp20 = tmp4 / tmp19
tmp21 = 15.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tmp27 = tmp0 - tmp20
tmp28 = tmp27 / tmp25
tmp29 = tmp26 * tmp28
tmp30 = 1.0
tmp31 = tmp30 - tmp26
tmp33 = tmp0 - tmp32
tmp35 = tmp33 / tmp34
tmp36 = tmp31 * tmp35
tmp37 = tmp29 + tmp36
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + (x0), tmp25, xmask)
tl.store(out_ptr0 + (r1 + (16*x0)), tmp37, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/bh/cbhkkkfowmcaigx6sts2oovquwlb7qm6qdlrdzh26rfoghoreml6.py
# Topologically Sorted Source Nodes: [sub, out_in, sub_1, out_ln, mul, sub_2, mul_1, out, mul_2, out_1], Original ATen: [aten.sub, aten.div, aten.mul, aten.rsub, aten.add]
# Source node to ATen node mapping:
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# out => add_2
# out_1 => add_3
# out_in => div
# out_ln => div_1
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %sqrt), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean_1), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_1, %sqrt_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%expand, %div), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %expand), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_2, %div_1), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_2, %unsqueeze_1), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_2, %unsqueeze_3), kwargs = {})
triton_poi_fused_add_div_mul_rsub_sub_2 = async_compile.triton('triton_poi_fused_add_div_mul_rsub_sub_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_rsub_sub_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mul_rsub_sub_2(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4096
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex % 256
x0 = xindex % 16
x2 = (xindex // 256)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x3), None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x0 + (16*x2)), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x0 + (16*x2)), None, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp4 = tmp2 + tmp3
tl.store(out_ptr0 + (x4), tmp4, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf6 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf9 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf7 = reinterpret_tensor(buf6, (4, 1, 1, 1), (1, 1, 1, 1), 0); del buf6 # reuse
buf11 = reinterpret_tensor(buf9, (4, 1, 1, 1), (1, 1, 1, 1), 0); del buf9 # reuse
# Topologically Sorted Source Nodes: [ln_mean, ln_var, add_1, sqrt_1], Original ATen: [aten.mean, aten.var, aten.add, aten.sqrt]
stream0 = get_raw_stream(0)
triton_per_fused_add_mean_sqrt_var_0.run(buf7, buf11, primals_1, 4, 64, grid=grid(4), stream=stream0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf3 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0); del buf0 # reuse
buf5 = reinterpret_tensor(buf3, (4, 4, 1, 1), (4, 1, 1, 1), 0); del buf3 # reuse
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [in_mean, in_var, sub, add, sqrt, out_in, sub_1, out_ln, mul, sub_2, mul_1, out], Original ATen: [aten.mean, aten.var, aten.sub, aten.add, aten.sqrt, aten.div, aten.mul, aten.rsub]
triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1.run(buf1, buf5, primals_1, primals_2, buf7, buf11, buf12, 16, 16, grid=grid(16), stream=stream0)
del primals_2
buf13 = empty_strided_cuda((4, 4, 4, 4, 4, 4), (1024, 256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [sub, out_in, sub_1, out_ln, mul, sub_2, mul_1, out, mul_2, out_1], Original ATen: [aten.sub, aten.div, aten.mul, aten.rsub, aten.add]
triton_poi_fused_add_div_mul_rsub_sub_2.run(buf12, primals_3, primals_4, buf13, 4096, grid=grid(4096), stream=stream0)
del buf12
del primals_4
return (buf13, primals_1, primals_3, buf1, buf5, buf7, buf11, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
from torch.nn.parameter import Parameter
class adaILN(nn.Module):
def __init__(self, num_features, eps=1e-05):
super(adaILN, self).__init__()
self.eps = eps
self.rho = Parameter(torch.Tensor(1, num_features, 1, 1))
self.rho.data.fill_(0.9)
def forward(self, input, gamma, beta):
in_mean, in_var = torch.mean(input, dim=[2, 3], keepdim=True
), torch.var(input, dim=[2, 3], keepdim=True)
out_in = (input - in_mean) / torch.sqrt(in_var + self.eps)
ln_mean, ln_var = torch.mean(input, dim=[1, 2, 3], keepdim=True
), torch.var(input, dim=[1, 2, 3], keepdim=True)
out_ln = (input - ln_mean) / torch.sqrt(ln_var + self.eps)
out = self.rho.expand(input.shape[0], -1, -1, -1) * out_in + (1 -
self.rho.expand(input.shape[0], -1, -1, -1)) * out_ln
out = out * gamma.unsqueeze(2).unsqueeze(3) + beta.unsqueeze(2
).unsqueeze(3)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {'num_features': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_add_mean_sqrt_var_0(in_out_ptr0, in_out_ptr1, in_ptr0,
xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 64, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 64.0
tmp20 = tmp4 / tmp19
tmp21 = 63.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + x0, tmp25, xmask)
@triton.jit
def triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1(in_out_ptr0,
in_out_ptr1, in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel,
rnumel, XBLOCK: tl.constexpr):
xnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
x2 = xindex % 4
x3 = xindex // 4
tmp0 = tl.load(in_ptr0 + (r1 + 16 * x0), xmask, other=0.0)
tmp26 = tl.load(in_ptr1 + x2, xmask, eviction_policy='evict_last')
tmp32 = tl.load(in_ptr2 + x3, xmask, eviction_policy='evict_last')
tmp34 = tl.load(in_ptr3 + x3, xmask, eviction_policy='evict_last')
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 16, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 16.0
tmp20 = tmp4 / tmp19
tmp21 = 15.0
tmp22 = tmp18 / tmp21
tmp23 = 1e-05
tmp24 = tmp22 + tmp23
tmp25 = libdevice.sqrt(tmp24)
tmp27 = tmp0 - tmp20
tmp28 = tmp27 / tmp25
tmp29 = tmp26 * tmp28
tmp30 = 1.0
tmp31 = tmp30 - tmp26
tmp33 = tmp0 - tmp32
tmp35 = tmp33 / tmp34
tmp36 = tmp31 * tmp35
tmp37 = tmp29 + tmp36
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp20, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + x0, tmp25, xmask)
tl.store(out_ptr0 + (r1 + 16 * x0), tmp37, xmask)
@triton.jit
def triton_poi_fused_add_div_mul_rsub_sub_2(in_ptr0, in_ptr1, in_ptr2,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex % 256
x0 = xindex % 16
x2 = xindex // 256
x4 = xindex
tmp0 = tl.load(in_ptr0 + x3, None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x0 + 16 * x2), None, eviction_policy='evict_last'
)
tmp3 = tl.load(in_ptr2 + (x0 + 16 * x2), None, eviction_policy='evict_last'
)
tmp2 = tmp0 * tmp1
tmp4 = tmp2 + tmp3
tl.store(out_ptr0 + x4, tmp4, None)
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf6 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf9 = empty_strided_cuda((4, 1, 1, 1), (1, 4, 4, 4), torch.float32)
buf7 = reinterpret_tensor(buf6, (4, 1, 1, 1), (1, 1, 1, 1), 0)
del buf6
buf11 = reinterpret_tensor(buf9, (4, 1, 1, 1), (1, 1, 1, 1), 0)
del buf9
get_raw_stream(0)
triton_per_fused_add_mean_sqrt_var_0[grid(4)](buf7, buf11,
primals_1, 4, 64, XBLOCK=1, num_warps=2, num_stages=1)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf3 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0)
del buf0
buf5 = reinterpret_tensor(buf3, (4, 4, 1, 1), (4, 1, 1, 1), 0)
del buf3
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_per_fused_add_div_mean_mul_rsub_sqrt_sub_var_1[grid(16)](buf1,
buf5, primals_1, primals_2, buf7, buf11, buf12, 16, 16, XBLOCK=
8, num_warps=2, num_stages=1)
del primals_2
buf13 = empty_strided_cuda((4, 4, 4, 4, 4, 4), (1024, 256, 64, 16,
4, 1), torch.float32)
triton_poi_fused_add_div_mul_rsub_sub_2[grid(4096)](buf12,
primals_3, primals_4, buf13, 4096, XBLOCK=256, num_warps=4,
num_stages=1)
del buf12
del primals_4
return buf13, primals_1, primals_3, buf1, buf5, buf7, buf11
class adaILNNew(nn.Module):
def __init__(self, num_features, eps=1e-05):
super(adaILNNew, self).__init__()
self.eps = eps
self.rho = Parameter(torch.Tensor(1, num_features, 1, 1))
self.rho.data.fill_(0.9)
def forward(self, input_0, input_1, input_2):
primals_2 = self.rho
primals_1 = input_0
primals_3 = input_1
primals_4 = input_2
output = call([primals_1, primals_2, primals_3, primals_4])
return output[0]
|
SubZero12556/Cats2dogs_ONNX
|
adaILN
| false | 14,462 |
[
"MIT"
] | 2,519 |
52a6a60d519e23b02f0847f0fa9f9ead89ca5f4e
|
https://github.com/SubZero12556/Cats2dogs_ONNX/tree/52a6a60d519e23b02f0847f0fa9f9ead89ca5f4e
|
Unit1D
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/wx/cwxdyif5yfmfhla2radb4uz5a3ut6zt5zvji46osgriepgl66sbk.py
# Topologically Sorted Source Nodes: [x_1, x_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x_1 => convolution
# x_2 => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_1, %primals_2, %primals_3, [1], [0], [1], False, [0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
tl.store(out_ptr0 + (x3), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4), (16, 4, 1))
buf1 = buf0; del buf0 # reuse
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_1, x_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_threshold_backward_0.run(buf1, primals_3, buf2, 64, grid=grid(64), stream=stream0)
del primals_3
return (buf1, primals_1, primals_2, buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
class Unit1D(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=1, stride
=1, padding='same', activation_fn=F.relu, use_bias=True):
super(Unit1D, self).__init__()
self.conv1d = nn.Conv1d(in_channels, output_channels, kernel_shape,
stride, padding=0, bias=use_bias)
self._activation_fn = activation_fn
self._padding = padding
self._stride = stride
self._kernel_shape = kernel_shape
def compute_pad(self, t):
if t % self._stride == 0:
return max(self._kernel_shape - self._stride, 0)
else:
return max(self._kernel_shape - t % self._stride, 0)
def forward(self, x):
if self._padding == 'same':
_batch, _channel, t = x.size()
pad_t = self.compute_pad(t)
pad_t_f = pad_t // 2
pad_t_b = pad_t - pad_t_f
x = F.pad(x, [pad_t_f, pad_t_b])
x = self.conv1d(x)
if self._activation_fn is not None:
x = self._activation_fn(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'output_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_0(in_out_ptr0,
in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x3, tmp4, xmask)
tl.store(out_ptr0 + x3, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4), (16, 4, 1))
buf1 = buf0
del buf0
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_convolution_relu_threshold_backward_0[grid(64)](buf1,
primals_3, buf2, 64, XBLOCK=64, num_warps=1, num_stages=1)
del primals_3
return buf1, primals_1, primals_2, buf2
class Unit1DNew(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=1, stride
=1, padding='same', activation_fn=F.relu, use_bias=True):
super(Unit1DNew, self).__init__()
self.conv1d = nn.Conv1d(in_channels, output_channels, kernel_shape,
stride, padding=0, bias=use_bias)
self._activation_fn = activation_fn
self._padding = padding
self._stride = stride
self._kernel_shape = kernel_shape
def compute_pad(self, t):
if t % self._stride == 0:
return max(self._kernel_shape - self._stride, 0)
else:
return max(self._kernel_shape - t % self._stride, 0)
def forward(self, input_0):
primals_2 = self.conv1d.weight
primals_3 = self.conv1d.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
TencentYoutuResearch/ActionDetection-AFSD
|
Unit1D
| false | 14,463 |
[
"BSD-3-Clause"
] | 112 |
ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
https://github.com/TencentYoutuResearch/ActionDetection-AFSD/tree/ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
ScaleExp
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ij/cijvenr6q47ivjlilbssdsnluzb744hqbjirlgwwyakwrhdphene.py
# Topologically Sorted Source Nodes: [mul, exp], Original ATen: [aten.mul, aten.exp]
# Source node to ATen node mapping:
# exp => exp
# mul => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_2, %primals_1), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%mul,), kwargs = {})
triton_poi_fused_exp_mul_0 = async_compile.triton('triton_poi_fused_exp_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_exp_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_exp_mul_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 * tmp2
tmp4 = tl_math.exp(tmp3)
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (1, ), (1, ))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, exp], Original ATen: [aten.mul, aten.exp]
stream0 = get_raw_stream(0)
triton_poi_fused_exp_mul_0.run(primals_2, primals_1, buf0, 256, grid=grid(256), stream=stream0)
del primals_1
return (buf0, primals_2, buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class ScaleExp(nn.Module):
def __init__(self, init_value=1.0):
super(ScaleExp, self).__init__()
self.scale = nn.Parameter(torch.FloatTensor([init_value]))
def forward(self, input):
return torch.exp(input * self.scale)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_exp_mul_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 * tmp2
tmp4 = tl_math.exp(tmp3)
tl.store(out_ptr0 + x0, tmp4, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (1,), (1,))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_exp_mul_0[grid(256)](primals_2, primals_1, buf0,
256, XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
return buf0, primals_2, buf0
class ScaleExpNew(nn.Module):
def __init__(self, init_value=1.0):
super(ScaleExpNew, self).__init__()
self.scale = nn.Parameter(torch.FloatTensor([init_value]))
def forward(self, input_0):
primals_1 = self.scale
primals_2 = input_0
output = call([primals_1, primals_2])
return output[0]
|
TencentYoutuResearch/ActionDetection-AFSD
|
ScaleExp
| false | 14,464 |
[
"BSD-3-Clause"
] | 112 |
ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
https://github.com/TencentYoutuResearch/ActionDetection-AFSD/tree/ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
GAT
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/t4/ct4b67pdo2nqkmug5ve6psoz6ptovf44cjwac2selnsbhojvain4.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%view, %repeat_1], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 8
x1 = (xindex // 8)
x2 = xindex
tmp0 = x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((4*((((4*x1) + x0) // 16) % 4)) + ((((4*x1) + x0) % 16) % 4)), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr0 + ((4*(x1 % 4)) + ((-4) + x0)), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + (x2), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/rs/crsikvivgof4u6qcelh3gov7oade5uaprup6quh2r4qjgsssen7k.py
# Topologically Sorted Source Nodes: [e], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# e => gt
# Graph fragment:
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze, 0), kwargs = {})
triton_poi_fused_leaky_relu_1 = async_compile.triton('triton_poi_fused_leaky_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i1', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.0
tmp2 = tmp0 > tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ma/cmadpgjcvhgmx4wckytb6z6iivymlbid6aqweieuugx3flkeqbta.py
# Topologically Sorted Source Nodes: [e, zero_vec_1, attention, attention_1, e_1, attention_3, attention_4, e_2, attention_6, attention_7, e_3, attention_9, attention_10], Original ATen: [aten.leaky_relu, aten.fill, aten.where, aten._softmax]
# Source node to ATen node mapping:
# attention => where_1
# attention_1 => amax, exp, sub, sum_1
# attention_10 => amax_3, exp_3, sub_3, sum_4
# attention_3 => where_4
# attention_4 => amax_1, exp_1, sub_1, sum_2
# attention_6 => where_7
# attention_7 => amax_2, exp_2, sub_2, sum_3
# attention_9 => where_10
# e => mul, where
# e_1 => mul_4, where_3
# e_2 => mul_8, where_6
# e_3 => mul_12, where_9
# zero_vec_1 => full_default
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze, 4), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %squeeze, %mul), kwargs = {})
# %full_default : [num_users=4] = call_function[target=torch.ops.aten.full.default](args = ([4, 4], 9e-15), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where_1 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where, %full_default), kwargs = {})
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%where_1, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_1, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_1, 4), kwargs = {})
# %where_3 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_3, %squeeze_1, %mul_4), kwargs = {})
# %where_4 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where_3, %full_default), kwargs = {})
# %amax_1 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%where_4, [1], True), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_4, %amax_1), kwargs = {})
# %exp_1 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_1,), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_1, [1], True), kwargs = {})
# %mul_8 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_2, 4), kwargs = {})
# %where_6 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_6, %squeeze_2, %mul_8), kwargs = {})
# %where_7 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where_6, %full_default), kwargs = {})
# %amax_2 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%where_7, [1], True), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_7, %amax_2), kwargs = {})
# %exp_2 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_2,), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_2, [1], True), kwargs = {})
# %mul_12 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_3, 4), kwargs = {})
# %where_9 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_9, %squeeze_3, %mul_12), kwargs = {})
# %where_10 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where_9, %full_default), kwargs = {})
# %amax_3 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%where_10, [1], True), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_10, %amax_3), kwargs = {})
# %exp_3 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_3,), kwargs = {})
# %sum_4 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_3, [1], True), kwargs = {})
triton_poi_fused__softmax_fill_leaky_relu_where_2 = async_compile.triton('triton_poi_fused__softmax_fill_leaky_relu_where_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*i1', 1: '*i1', 2: '*fp32', 3: '*i1', 4: '*fp32', 5: '*i1', 6: '*fp32', 7: '*i1', 8: '*fp32', 9: '*fp32', 10: '*fp32', 11: '*fp32', 12: '*fp32', 13: '*fp32', 14: '*fp32', 15: '*fp32', 16: '*fp32', 17: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_fill_leaky_relu_where_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 36, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_fill_leaky_relu_where_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, in_ptr8, out_ptr0, out_ptr1, out_ptr2, out_ptr3, out_ptr4, out_ptr5, out_ptr6, out_ptr7, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last').to(tl.int1)
tmp1 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last').to(tl.int1)
tmp2 = tl.load(in_ptr2 + (4*x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp9 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp10 = tl.load(in_ptr2 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp16 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp17 = tl.load(in_ptr2 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp23 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp24 = tl.load(in_ptr2 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp40 = tl.load(in_ptr3 + (4*x0), xmask, eviction_policy='evict_last').to(tl.int1)
tmp41 = tl.load(in_ptr4 + (4*x0), xmask, eviction_policy='evict_last')
tmp45 = tl.load(in_ptr3 + (1 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp46 = tl.load(in_ptr4 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp51 = tl.load(in_ptr3 + (2 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp52 = tl.load(in_ptr4 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr3 + (3 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp58 = tl.load(in_ptr4 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp74 = tl.load(in_ptr5 + (4*x0), xmask, eviction_policy='evict_last').to(tl.int1)
tmp75 = tl.load(in_ptr6 + (4*x0), xmask, eviction_policy='evict_last')
tmp79 = tl.load(in_ptr5 + (1 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp80 = tl.load(in_ptr6 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp85 = tl.load(in_ptr5 + (2 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp86 = tl.load(in_ptr6 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp91 = tl.load(in_ptr5 + (3 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp92 = tl.load(in_ptr6 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp108 = tl.load(in_ptr7 + (4*x0), xmask, eviction_policy='evict_last').to(tl.int1)
tmp109 = tl.load(in_ptr8 + (4*x0), xmask, eviction_policy='evict_last')
tmp113 = tl.load(in_ptr7 + (1 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp114 = tl.load(in_ptr8 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp119 = tl.load(in_ptr7 + (2 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp120 = tl.load(in_ptr8 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp125 = tl.load(in_ptr7 + (3 + (4*x0)), xmask, eviction_policy='evict_last').to(tl.int1)
tmp126 = tl.load(in_ptr8 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = 4.0
tmp4 = tmp2 * tmp3
tmp5 = tl.where(tmp1, tmp2, tmp4)
tmp6 = 9e-15
tmp7 = tl.where(tmp0, tmp5, tmp6)
tmp11 = tmp10 * tmp3
tmp12 = tl.where(tmp9, tmp10, tmp11)
tmp13 = tl.where(tmp8, tmp12, tmp6)
tmp14 = triton_helpers.maximum(tmp7, tmp13)
tmp18 = tmp17 * tmp3
tmp19 = tl.where(tmp16, tmp17, tmp18)
tmp20 = tl.where(tmp15, tmp19, tmp6)
tmp21 = triton_helpers.maximum(tmp14, tmp20)
tmp25 = tmp24 * tmp3
tmp26 = tl.where(tmp23, tmp24, tmp25)
tmp27 = tl.where(tmp22, tmp26, tmp6)
tmp28 = triton_helpers.maximum(tmp21, tmp27)
tmp29 = tmp7 - tmp28
tmp30 = tl_math.exp(tmp29)
tmp31 = tmp13 - tmp28
tmp32 = tl_math.exp(tmp31)
tmp33 = tmp30 + tmp32
tmp34 = tmp20 - tmp28
tmp35 = tl_math.exp(tmp34)
tmp36 = tmp33 + tmp35
tmp37 = tmp27 - tmp28
tmp38 = tl_math.exp(tmp37)
tmp39 = tmp36 + tmp38
tmp42 = tmp41 * tmp3
tmp43 = tl.where(tmp40, tmp41, tmp42)
tmp44 = tl.where(tmp0, tmp43, tmp6)
tmp47 = tmp46 * tmp3
tmp48 = tl.where(tmp45, tmp46, tmp47)
tmp49 = tl.where(tmp8, tmp48, tmp6)
tmp50 = triton_helpers.maximum(tmp44, tmp49)
tmp53 = tmp52 * tmp3
tmp54 = tl.where(tmp51, tmp52, tmp53)
tmp55 = tl.where(tmp15, tmp54, tmp6)
tmp56 = triton_helpers.maximum(tmp50, tmp55)
tmp59 = tmp58 * tmp3
tmp60 = tl.where(tmp57, tmp58, tmp59)
tmp61 = tl.where(tmp22, tmp60, tmp6)
tmp62 = triton_helpers.maximum(tmp56, tmp61)
tmp63 = tmp44 - tmp62
tmp64 = tl_math.exp(tmp63)
tmp65 = tmp49 - tmp62
tmp66 = tl_math.exp(tmp65)
tmp67 = tmp64 + tmp66
tmp68 = tmp55 - tmp62
tmp69 = tl_math.exp(tmp68)
tmp70 = tmp67 + tmp69
tmp71 = tmp61 - tmp62
tmp72 = tl_math.exp(tmp71)
tmp73 = tmp70 + tmp72
tmp76 = tmp75 * tmp3
tmp77 = tl.where(tmp74, tmp75, tmp76)
tmp78 = tl.where(tmp0, tmp77, tmp6)
tmp81 = tmp80 * tmp3
tmp82 = tl.where(tmp79, tmp80, tmp81)
tmp83 = tl.where(tmp8, tmp82, tmp6)
tmp84 = triton_helpers.maximum(tmp78, tmp83)
tmp87 = tmp86 * tmp3
tmp88 = tl.where(tmp85, tmp86, tmp87)
tmp89 = tl.where(tmp15, tmp88, tmp6)
tmp90 = triton_helpers.maximum(tmp84, tmp89)
tmp93 = tmp92 * tmp3
tmp94 = tl.where(tmp91, tmp92, tmp93)
tmp95 = tl.where(tmp22, tmp94, tmp6)
tmp96 = triton_helpers.maximum(tmp90, tmp95)
tmp97 = tmp78 - tmp96
tmp98 = tl_math.exp(tmp97)
tmp99 = tmp83 - tmp96
tmp100 = tl_math.exp(tmp99)
tmp101 = tmp98 + tmp100
tmp102 = tmp89 - tmp96
tmp103 = tl_math.exp(tmp102)
tmp104 = tmp101 + tmp103
tmp105 = tmp95 - tmp96
tmp106 = tl_math.exp(tmp105)
tmp107 = tmp104 + tmp106
tmp110 = tmp109 * tmp3
tmp111 = tl.where(tmp108, tmp109, tmp110)
tmp112 = tl.where(tmp0, tmp111, tmp6)
tmp115 = tmp114 * tmp3
tmp116 = tl.where(tmp113, tmp114, tmp115)
tmp117 = tl.where(tmp8, tmp116, tmp6)
tmp118 = triton_helpers.maximum(tmp112, tmp117)
tmp121 = tmp120 * tmp3
tmp122 = tl.where(tmp119, tmp120, tmp121)
tmp123 = tl.where(tmp15, tmp122, tmp6)
tmp124 = triton_helpers.maximum(tmp118, tmp123)
tmp127 = tmp126 * tmp3
tmp128 = tl.where(tmp125, tmp126, tmp127)
tmp129 = tl.where(tmp22, tmp128, tmp6)
tmp130 = triton_helpers.maximum(tmp124, tmp129)
tmp131 = tmp112 - tmp130
tmp132 = tl_math.exp(tmp131)
tmp133 = tmp117 - tmp130
tmp134 = tl_math.exp(tmp133)
tmp135 = tmp132 + tmp134
tmp136 = tmp123 - tmp130
tmp137 = tl_math.exp(tmp136)
tmp138 = tmp135 + tmp137
tmp139 = tmp129 - tmp130
tmp140 = tl_math.exp(tmp139)
tmp141 = tmp138 + tmp140
tl.store(out_ptr0 + (x0), tmp28, xmask)
tl.store(out_ptr1 + (x0), tmp39, xmask)
tl.store(out_ptr2 + (x0), tmp62, xmask)
tl.store(out_ptr3 + (x0), tmp73, xmask)
tl.store(out_ptr4 + (x0), tmp96, xmask)
tl.store(out_ptr5 + (x0), tmp107, xmask)
tl.store(out_ptr6 + (x0), tmp130, xmask)
tl.store(out_ptr7 + (x0), tmp141, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/5m/c5m6l3ct22vl4fsrxep7r7xms46lwcvm5sdoe5jyb77ksclkqcb3.py
# Topologically Sorted Source Nodes: [e, zero_vec_1, attention, attention_1, e_1, attention_3, attention_4, e_2, attention_6, attention_7, e_3, attention_9, attention_10], Original ATen: [aten.leaky_relu, aten.fill, aten.where, aten._softmax]
# Source node to ATen node mapping:
# attention => where_1
# attention_1 => div, exp, sub
# attention_10 => div_3, exp_3, sub_3
# attention_3 => where_4
# attention_4 => div_1, exp_1, sub_1
# attention_6 => where_7
# attention_7 => div_2, exp_2, sub_2
# attention_9 => where_10
# e => mul, where
# e_1 => mul_4, where_3
# e_2 => mul_8, where_6
# e_3 => mul_12, where_9
# zero_vec_1 => full_default
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze, 4), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %squeeze, %mul), kwargs = {})
# %full_default : [num_users=4] = call_function[target=torch.ops.aten.full.default](args = ([4, 4], 9e-15), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where_1 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where, %full_default), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_1, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %div : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_1, 4), kwargs = {})
# %where_3 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_3, %squeeze_1, %mul_4), kwargs = {})
# %where_4 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where_3, %full_default), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_4, %amax_1), kwargs = {})
# %exp_1 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_1,), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_1, %sum_2), kwargs = {})
# %mul_8 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_2, 4), kwargs = {})
# %where_6 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_6, %squeeze_2, %mul_8), kwargs = {})
# %where_7 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where_6, %full_default), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_7, %amax_2), kwargs = {})
# %exp_2 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_2,), kwargs = {})
# %div_2 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_2, %sum_3), kwargs = {})
# %mul_12 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_3, 4), kwargs = {})
# %where_9 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_9, %squeeze_3, %mul_12), kwargs = {})
# %where_10 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %where_9, %full_default), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_10, %amax_3), kwargs = {})
# %exp_3 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_3,), kwargs = {})
# %div_3 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_3, %sum_4), kwargs = {})
triton_poi_fused__softmax_fill_leaky_relu_where_3 = async_compile.triton('triton_poi_fused__softmax_fill_leaky_relu_where_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*i1', 5: '*i1', 6: '*fp32', 7: '*fp32', 8: '*i1', 9: '*fp32', 10: '*fp32', 11: '*i1', 12: '*fp32', 13: '*fp32', 14: '*i1', 15: '*fp32', 16: '*fp32', 17: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_fill_leaky_relu_where_3', 'mutated_arg_names': ['in_out_ptr0', 'in_out_ptr1', 'in_out_ptr2', 'in_out_ptr3'], 'no_x_dim': False, 'num_load': 17, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_fill_leaky_relu_where_3(in_out_ptr0, in_out_ptr1, in_out_ptr2, in_out_ptr3, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, in_ptr8, in_ptr9, in_ptr10, in_ptr11, in_ptr12, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask).to(tl.int1)
tmp1 = tl.load(in_ptr1 + (x2), xmask).to(tl.int1)
tmp2 = tl.load(in_out_ptr0 + (x2), xmask)
tmp8 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp13 = tl.load(in_ptr4 + (x2), xmask).to(tl.int1)
tmp14 = tl.load(in_out_ptr1 + (x2), xmask)
tmp18 = tl.load(in_ptr5 + (x1), xmask, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr6 + (x1), xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr7 + (x2), xmask).to(tl.int1)
tmp24 = tl.load(in_out_ptr2 + (x2), xmask)
tmp28 = tl.load(in_ptr8 + (x1), xmask, eviction_policy='evict_last')
tmp31 = tl.load(in_ptr9 + (x1), xmask, eviction_policy='evict_last')
tmp33 = tl.load(in_ptr10 + (x2), xmask).to(tl.int1)
tmp34 = tl.load(in_out_ptr3 + (x2), xmask)
tmp38 = tl.load(in_ptr11 + (x1), xmask, eviction_policy='evict_last')
tmp41 = tl.load(in_ptr12 + (x1), xmask, eviction_policy='evict_last')
tmp3 = 4.0
tmp4 = tmp2 * tmp3
tmp5 = tl.where(tmp1, tmp2, tmp4)
tmp6 = 9e-15
tmp7 = tl.where(tmp0, tmp5, tmp6)
tmp9 = tmp7 - tmp8
tmp10 = tl_math.exp(tmp9)
tmp12 = tmp10 / tmp11
tmp15 = tmp14 * tmp3
tmp16 = tl.where(tmp13, tmp14, tmp15)
tmp17 = tl.where(tmp0, tmp16, tmp6)
tmp19 = tmp17 - tmp18
tmp20 = tl_math.exp(tmp19)
tmp22 = tmp20 / tmp21
tmp25 = tmp24 * tmp3
tmp26 = tl.where(tmp23, tmp24, tmp25)
tmp27 = tl.where(tmp0, tmp26, tmp6)
tmp29 = tmp27 - tmp28
tmp30 = tl_math.exp(tmp29)
tmp32 = tmp30 / tmp31
tmp35 = tmp34 * tmp3
tmp36 = tl.where(tmp33, tmp34, tmp35)
tmp37 = tl.where(tmp0, tmp36, tmp6)
tmp39 = tmp37 - tmp38
tmp40 = tl_math.exp(tmp39)
tmp42 = tmp40 / tmp41
tl.store(in_out_ptr0 + (x2), tmp12, xmask)
tl.store(in_out_ptr1 + (x2), tmp22, xmask)
tl.store(in_out_ptr2 + (x2), tmp32, xmask)
tl.store(in_out_ptr3 + (x2), tmp42, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/6f/c6fg755hkzgmiizoydcu7wlmcvduiztugqjkietqkvpoph4vrtad.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# x_1 => cat_4
# Graph fragment:
# %cat_4 : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%where_2, %where_5, %where_8, %where_11], 1), kwargs = {})
triton_poi_fused_cat_4 = async_compile.triton('triton_poi_fused_cat_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = (xindex // 16)
x2 = xindex
tmp0 = x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((4*x1) + x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = 0.0
tmp7 = tmp5 > tmp6
tmp8 = 1.0
tmp9 = tmp5 * tmp8
tmp10 = libdevice.expm1(tmp9)
tmp11 = tmp10 * tmp8
tmp12 = tl.where(tmp7, tmp9, tmp11)
tmp13 = tl.full(tmp12.shape, 0.0, tmp12.dtype)
tmp14 = tl.where(tmp4, tmp12, tmp13)
tmp15 = tmp0 >= tmp3
tmp16 = tl.full([1], 8, tl.int64)
tmp17 = tmp0 < tmp16
tmp18 = tmp15 & tmp17
tmp19 = tl.load(in_ptr1 + ((4*x1) + ((-4) + x0)), tmp18 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tmp19 > tmp6
tmp21 = tmp19 * tmp8
tmp22 = libdevice.expm1(tmp21)
tmp23 = tmp22 * tmp8
tmp24 = tl.where(tmp20, tmp21, tmp23)
tmp25 = tl.full(tmp24.shape, 0.0, tmp24.dtype)
tmp26 = tl.where(tmp18, tmp24, tmp25)
tmp27 = tmp0 >= tmp16
tmp28 = tl.full([1], 12, tl.int64)
tmp29 = tmp0 < tmp28
tmp30 = tmp27 & tmp29
tmp31 = tl.load(in_ptr2 + ((4*x1) + ((-8) + x0)), tmp30 & xmask, eviction_policy='evict_last', other=0.0)
tmp32 = tmp31 > tmp6
tmp33 = tmp31 * tmp8
tmp34 = libdevice.expm1(tmp33)
tmp35 = tmp34 * tmp8
tmp36 = tl.where(tmp32, tmp33, tmp35)
tmp37 = tl.full(tmp36.shape, 0.0, tmp36.dtype)
tmp38 = tl.where(tmp30, tmp36, tmp37)
tmp39 = tmp0 >= tmp28
tmp40 = tl.full([1], 16, tl.int64)
tmp41 = tmp0 < tmp40
tmp42 = tl.load(in_ptr3 + ((4*x1) + ((-12) + x0)), tmp39 & xmask, eviction_policy='evict_last', other=0.0)
tmp43 = tmp42 > tmp6
tmp44 = tmp42 * tmp8
tmp45 = libdevice.expm1(tmp44)
tmp46 = tmp45 * tmp8
tmp47 = tl.where(tmp43, tmp44, tmp46)
tmp48 = tl.full(tmp47.shape, 0.0, tmp47.dtype)
tmp49 = tl.where(tmp39, tmp47, tmp48)
tmp50 = tl.where(tmp30, tmp38, tmp49)
tmp51 = tl.where(tmp18, tmp26, tmp50)
tmp52 = tl.where(tmp4, tmp14, tmp51)
tl.store(out_ptr0 + (x2), tmp52, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (8, 1), (1, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (8, 1), (1, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (8, 1), (1, 1))
assert_size_stride(primals_9, (4, 4), (4, 1))
assert_size_stride(primals_10, (8, 1), (1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.native_dropout]
buf0 = torch.ops.aten.native_dropout.default(primals_1, 0.5, True)
del primals_1
buf1 = buf0[0]
del buf0
buf3 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h], Original ATen: [aten.mm]
extern_kernels.mm(buf1, primals_2, out=buf3)
del primals_2
buf4 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(buf3, buf4, 128, grid=grid(128), stream=stream0)
buf5 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.mm]
extern_kernels.mm(buf4, primals_3, out=buf5)
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [e], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_1.run(buf5, buf6, 16, grid=grid(16), stream=stream0)
buf7 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [gt], Original ATen: [aten.gt]
triton_poi_fused_leaky_relu_1.run(primals_4, buf7, 16, grid=grid(16), stream=stream0)
del primals_4
buf12 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h_1], Original ATen: [aten.mm]
extern_kernels.mm(buf1, primals_5, out=buf12)
del primals_5
buf13 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat_1], Original ATen: [aten.cat]
triton_poi_fused_cat_0.run(buf12, buf13, 128, grid=grid(128), stream=stream0)
buf14 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.mm]
extern_kernels.mm(buf13, primals_6, out=buf14)
buf15 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [e_1], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_1.run(buf14, buf15, 16, grid=grid(16), stream=stream0)
buf20 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h_2], Original ATen: [aten.mm]
extern_kernels.mm(buf1, primals_7, out=buf20)
del primals_7
buf21 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat_2], Original ATen: [aten.cat]
triton_poi_fused_cat_0.run(buf20, buf21, 128, grid=grid(128), stream=stream0)
buf22 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_4], Original ATen: [aten.mm]
extern_kernels.mm(buf21, primals_8, out=buf22)
buf23 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [e_2], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_1.run(buf22, buf23, 16, grid=grid(16), stream=stream0)
buf28 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h_3], Original ATen: [aten.mm]
extern_kernels.mm(buf1, primals_9, out=buf28)
del primals_9
buf29 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat_3], Original ATen: [aten.cat]
triton_poi_fused_cat_0.run(buf28, buf29, 128, grid=grid(128), stream=stream0)
buf30 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_6], Original ATen: [aten.mm]
extern_kernels.mm(buf29, primals_10, out=buf30)
buf31 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [e_3], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_1.run(buf30, buf31, 16, grid=grid(16), stream=stream0)
buf8 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf9 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf16 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf17 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf24 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf25 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf32 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf33 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
# Topologically Sorted Source Nodes: [e, zero_vec_1, attention, attention_1, e_1, attention_3, attention_4, e_2, attention_6, attention_7, e_3, attention_9, attention_10], Original ATen: [aten.leaky_relu, aten.fill, aten.where, aten._softmax]
triton_poi_fused__softmax_fill_leaky_relu_where_2.run(buf7, buf6, buf5, buf15, buf14, buf23, buf22, buf31, buf30, buf8, buf9, buf16, buf17, buf24, buf25, buf32, buf33, 4, grid=grid(4), stream=stream0)
buf10 = reinterpret_tensor(buf5, (4, 4), (4, 1), 0); del buf5 # reuse
buf18 = reinterpret_tensor(buf14, (4, 4), (4, 1), 0); del buf14 # reuse
buf26 = reinterpret_tensor(buf22, (4, 4), (4, 1), 0); del buf22 # reuse
buf34 = reinterpret_tensor(buf30, (4, 4), (4, 1), 0); del buf30 # reuse
# Topologically Sorted Source Nodes: [e, zero_vec_1, attention, attention_1, e_1, attention_3, attention_4, e_2, attention_6, attention_7, e_3, attention_9, attention_10], Original ATen: [aten.leaky_relu, aten.fill, aten.where, aten._softmax]
triton_poi_fused__softmax_fill_leaky_relu_where_3.run(buf10, buf18, buf26, buf34, buf7, buf6, buf8, buf9, buf15, buf16, buf17, buf23, buf24, buf25, buf31, buf32, buf33, 16, grid=grid(16), stream=stream0)
del buf16
del buf17
del buf24
del buf25
del buf32
del buf33
del buf8
del buf9
buf11 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h_prime], Original ATen: [aten.mm]
extern_kernels.mm(buf10, buf3, out=buf11)
buf19 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h_prime_1], Original ATen: [aten.mm]
extern_kernels.mm(buf18, buf12, out=buf19)
buf27 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h_prime_2], Original ATen: [aten.mm]
extern_kernels.mm(buf26, buf20, out=buf27)
buf35 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h_prime_3], Original ATen: [aten.mm]
extern_kernels.mm(buf34, buf28, out=buf35)
buf36 = empty_strided_cuda((4, 16), (16, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.cat]
triton_poi_fused_cat_4.run(buf11, buf19, buf27, buf35, buf36, 64, grid=grid(64), stream=stream0)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.native_dropout]
buf37 = torch.ops.aten.native_dropout.default(buf36, 0.5, True)
del buf36
buf38 = buf37[0]
buf39 = buf37[1]
del buf37
return (buf38, buf6, buf7, buf10, buf11, buf15, buf18, buf19, buf23, buf26, buf27, buf31, buf34, buf35, buf39, reinterpret_tensor(buf28, (4, 4), (1, 4), 0), reinterpret_tensor(buf29, (8, 16), (1, 8), 0), reinterpret_tensor(primals_10, (1, 8), (1, 1), 0), reinterpret_tensor(buf1, (4, 4), (1, 4), 0), reinterpret_tensor(buf20, (4, 4), (1, 4), 0), reinterpret_tensor(buf21, (8, 16), (1, 8), 0), reinterpret_tensor(primals_8, (1, 8), (1, 1), 0), reinterpret_tensor(buf12, (4, 4), (1, 4), 0), reinterpret_tensor(buf13, (8, 16), (1, 8), 0), reinterpret_tensor(primals_6, (1, 8), (1, 1), 0), reinterpret_tensor(buf3, (4, 4), (1, 4), 0), reinterpret_tensor(buf4, (8, 16), (1, 8), 0), reinterpret_tensor(primals_3, (1, 8), (1, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((8, 1), (1, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((8, 1), (1, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((8, 1), (1, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((8, 1), (1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import numpy as np
import torch.nn as nn
import torch.nn.functional as F
class GraphAttentionLayer(nn.Module):
"""
Simple GAT layer, similar to https://arxiv.org/abs/1710.10903
"""
def __init__(self, in_features, out_features, dropout, alpha, concat=False
):
super(GraphAttentionLayer, self).__init__()
self.dropout = dropout
self.in_features = in_features
self.out_features = out_features
self.alpha = alpha
self.concat = concat
self.W = nn.Parameter(nn.init.xavier_uniform_(torch.Tensor(
in_features, out_features).type(torch.FloatTensor if torch.cuda
.is_available() else torch.FloatTensor), gain=np.sqrt(2.0)),
requires_grad=True)
self.a = nn.Parameter(nn.init.xavier_uniform_(torch.Tensor(2 *
out_features, 1).type(torch.FloatTensor if torch.cuda.
is_available() else torch.FloatTensor), gain=np.sqrt(2.0)),
requires_grad=True)
self.leakyrelu = nn.LeakyReLU(self.alpha)
def forward(self, input, adj):
h = torch.mm(input, self.W)
N = h.size()[0]
a_input = torch.cat([h.repeat(1, N).view(N * N, -1), h.repeat(N, 1)
], dim=1).view(N, -1, 2 * self.out_features)
e = self.leakyrelu(torch.matmul(a_input, self.a).squeeze(2))
zero_vec = torch.zeros_like(e)
zero_vec = zero_vec.fill_(9e-15)
attention = torch.where(adj > 0, e, zero_vec)
attention = F.softmax(attention, dim=1)
attention = F.dropout(attention, self.dropout, training=self.training)
h_prime = torch.matmul(attention, h)
if self.concat:
return F.elu(h_prime)
else:
return h_prime
def __repr__(self):
return self.__class__.__name__ + ' (' + str(self.in_features
) + ' -> ' + str(self.out_features) + ')'
class GAT(nn.Module):
def __init__(self, nfeat, nhid, dropout, alpha, nheads):
super(GAT, self).__init__()
self.dropout = dropout
self.attentions = [GraphAttentionLayer(nfeat, nhid, dropout=dropout,
alpha=alpha, concat=True) for _ in range(nheads)]
for i, attention in enumerate(self.attentions):
self.add_module('attention_{}'.format(i), attention)
def forward(self, x, adj):
x = F.dropout(x, self.dropout)
x = torch.cat([att(x, adj) for att in self.attentions], dim=1)
x = F.dropout(x, self.dropout)
return x
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'nfeat': 4, 'nhid': 4, 'dropout': 0.5, 'alpha': 4,
'nheads': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import numpy as np
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 8
x1 = xindex // 8
x2 = xindex
tmp0 = x0
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (4 * ((4 * x1 + x0) // 16 % 4) + (4 * x1 + x0) %
16 % 4), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp9 = tl.load(in_ptr0 + (4 * (x1 % 4) + (-4 + x0)), tmp6 & xmask,
eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + x2, tmp10, xmask)
@triton.jit
def triton_poi_fused_leaky_relu_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.0
tmp2 = tmp0 > tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_poi_fused__softmax_fill_leaky_relu_where_2(in_ptr0, in_ptr1,
in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, in_ptr8, out_ptr0,
out_ptr1, out_ptr2, out_ptr3, out_ptr4, out_ptr5, out_ptr6, out_ptr7,
xnumel, XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last').to(tl
.int1)
tmp1 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last').to(tl
.int1)
tmp2 = tl.load(in_ptr2 + 4 * x0, xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp9 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp10 = tl.load(in_ptr2 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp15 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp16 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp17 = tl.load(in_ptr2 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp22 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp23 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp24 = tl.load(in_ptr2 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp40 = tl.load(in_ptr3 + 4 * x0, xmask, eviction_policy='evict_last').to(
tl.int1)
tmp41 = tl.load(in_ptr4 + 4 * x0, xmask, eviction_policy='evict_last')
tmp45 = tl.load(in_ptr3 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp46 = tl.load(in_ptr4 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp51 = tl.load(in_ptr3 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp52 = tl.load(in_ptr4 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp57 = tl.load(in_ptr3 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp58 = tl.load(in_ptr4 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp74 = tl.load(in_ptr5 + 4 * x0, xmask, eviction_policy='evict_last').to(
tl.int1)
tmp75 = tl.load(in_ptr6 + 4 * x0, xmask, eviction_policy='evict_last')
tmp79 = tl.load(in_ptr5 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp80 = tl.load(in_ptr6 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp85 = tl.load(in_ptr5 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp86 = tl.load(in_ptr6 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp91 = tl.load(in_ptr5 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
).to(tl.int1)
tmp92 = tl.load(in_ptr6 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp108 = tl.load(in_ptr7 + 4 * x0, xmask, eviction_policy='evict_last').to(
tl.int1)
tmp109 = tl.load(in_ptr8 + 4 * x0, xmask, eviction_policy='evict_last')
tmp113 = tl.load(in_ptr7 + (1 + 4 * x0), xmask, eviction_policy=
'evict_last').to(tl.int1)
tmp114 = tl.load(in_ptr8 + (1 + 4 * x0), xmask, eviction_policy=
'evict_last')
tmp119 = tl.load(in_ptr7 + (2 + 4 * x0), xmask, eviction_policy=
'evict_last').to(tl.int1)
tmp120 = tl.load(in_ptr8 + (2 + 4 * x0), xmask, eviction_policy=
'evict_last')
tmp125 = tl.load(in_ptr7 + (3 + 4 * x0), xmask, eviction_policy=
'evict_last').to(tl.int1)
tmp126 = tl.load(in_ptr8 + (3 + 4 * x0), xmask, eviction_policy=
'evict_last')
tmp3 = 4.0
tmp4 = tmp2 * tmp3
tmp5 = tl.where(tmp1, tmp2, tmp4)
tmp6 = 9e-15
tmp7 = tl.where(tmp0, tmp5, tmp6)
tmp11 = tmp10 * tmp3
tmp12 = tl.where(tmp9, tmp10, tmp11)
tmp13 = tl.where(tmp8, tmp12, tmp6)
tmp14 = triton_helpers.maximum(tmp7, tmp13)
tmp18 = tmp17 * tmp3
tmp19 = tl.where(tmp16, tmp17, tmp18)
tmp20 = tl.where(tmp15, tmp19, tmp6)
tmp21 = triton_helpers.maximum(tmp14, tmp20)
tmp25 = tmp24 * tmp3
tmp26 = tl.where(tmp23, tmp24, tmp25)
tmp27 = tl.where(tmp22, tmp26, tmp6)
tmp28 = triton_helpers.maximum(tmp21, tmp27)
tmp29 = tmp7 - tmp28
tmp30 = tl_math.exp(tmp29)
tmp31 = tmp13 - tmp28
tmp32 = tl_math.exp(tmp31)
tmp33 = tmp30 + tmp32
tmp34 = tmp20 - tmp28
tmp35 = tl_math.exp(tmp34)
tmp36 = tmp33 + tmp35
tmp37 = tmp27 - tmp28
tmp38 = tl_math.exp(tmp37)
tmp39 = tmp36 + tmp38
tmp42 = tmp41 * tmp3
tmp43 = tl.where(tmp40, tmp41, tmp42)
tmp44 = tl.where(tmp0, tmp43, tmp6)
tmp47 = tmp46 * tmp3
tmp48 = tl.where(tmp45, tmp46, tmp47)
tmp49 = tl.where(tmp8, tmp48, tmp6)
tmp50 = triton_helpers.maximum(tmp44, tmp49)
tmp53 = tmp52 * tmp3
tmp54 = tl.where(tmp51, tmp52, tmp53)
tmp55 = tl.where(tmp15, tmp54, tmp6)
tmp56 = triton_helpers.maximum(tmp50, tmp55)
tmp59 = tmp58 * tmp3
tmp60 = tl.where(tmp57, tmp58, tmp59)
tmp61 = tl.where(tmp22, tmp60, tmp6)
tmp62 = triton_helpers.maximum(tmp56, tmp61)
tmp63 = tmp44 - tmp62
tmp64 = tl_math.exp(tmp63)
tmp65 = tmp49 - tmp62
tmp66 = tl_math.exp(tmp65)
tmp67 = tmp64 + tmp66
tmp68 = tmp55 - tmp62
tmp69 = tl_math.exp(tmp68)
tmp70 = tmp67 + tmp69
tmp71 = tmp61 - tmp62
tmp72 = tl_math.exp(tmp71)
tmp73 = tmp70 + tmp72
tmp76 = tmp75 * tmp3
tmp77 = tl.where(tmp74, tmp75, tmp76)
tmp78 = tl.where(tmp0, tmp77, tmp6)
tmp81 = tmp80 * tmp3
tmp82 = tl.where(tmp79, tmp80, tmp81)
tmp83 = tl.where(tmp8, tmp82, tmp6)
tmp84 = triton_helpers.maximum(tmp78, tmp83)
tmp87 = tmp86 * tmp3
tmp88 = tl.where(tmp85, tmp86, tmp87)
tmp89 = tl.where(tmp15, tmp88, tmp6)
tmp90 = triton_helpers.maximum(tmp84, tmp89)
tmp93 = tmp92 * tmp3
tmp94 = tl.where(tmp91, tmp92, tmp93)
tmp95 = tl.where(tmp22, tmp94, tmp6)
tmp96 = triton_helpers.maximum(tmp90, tmp95)
tmp97 = tmp78 - tmp96
tmp98 = tl_math.exp(tmp97)
tmp99 = tmp83 - tmp96
tmp100 = tl_math.exp(tmp99)
tmp101 = tmp98 + tmp100
tmp102 = tmp89 - tmp96
tmp103 = tl_math.exp(tmp102)
tmp104 = tmp101 + tmp103
tmp105 = tmp95 - tmp96
tmp106 = tl_math.exp(tmp105)
tmp107 = tmp104 + tmp106
tmp110 = tmp109 * tmp3
tmp111 = tl.where(tmp108, tmp109, tmp110)
tmp112 = tl.where(tmp0, tmp111, tmp6)
tmp115 = tmp114 * tmp3
tmp116 = tl.where(tmp113, tmp114, tmp115)
tmp117 = tl.where(tmp8, tmp116, tmp6)
tmp118 = triton_helpers.maximum(tmp112, tmp117)
tmp121 = tmp120 * tmp3
tmp122 = tl.where(tmp119, tmp120, tmp121)
tmp123 = tl.where(tmp15, tmp122, tmp6)
tmp124 = triton_helpers.maximum(tmp118, tmp123)
tmp127 = tmp126 * tmp3
tmp128 = tl.where(tmp125, tmp126, tmp127)
tmp129 = tl.where(tmp22, tmp128, tmp6)
tmp130 = triton_helpers.maximum(tmp124, tmp129)
tmp131 = tmp112 - tmp130
tmp132 = tl_math.exp(tmp131)
tmp133 = tmp117 - tmp130
tmp134 = tl_math.exp(tmp133)
tmp135 = tmp132 + tmp134
tmp136 = tmp123 - tmp130
tmp137 = tl_math.exp(tmp136)
tmp138 = tmp135 + tmp137
tmp139 = tmp129 - tmp130
tmp140 = tl_math.exp(tmp139)
tmp141 = tmp138 + tmp140
tl.store(out_ptr0 + x0, tmp28, xmask)
tl.store(out_ptr1 + x0, tmp39, xmask)
tl.store(out_ptr2 + x0, tmp62, xmask)
tl.store(out_ptr3 + x0, tmp73, xmask)
tl.store(out_ptr4 + x0, tmp96, xmask)
tl.store(out_ptr5 + x0, tmp107, xmask)
tl.store(out_ptr6 + x0, tmp130, xmask)
tl.store(out_ptr7 + x0, tmp141, xmask)
@triton.jit
def triton_poi_fused__softmax_fill_leaky_relu_where_3(in_out_ptr0,
in_out_ptr1, in_out_ptr2, in_out_ptr3, in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, in_ptr8, in_ptr9, in_ptr10,
in_ptr11, in_ptr12, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask).to(tl.int1)
tmp1 = tl.load(in_ptr1 + x2, xmask).to(tl.int1)
tmp2 = tl.load(in_out_ptr0 + x2, xmask)
tmp8 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp13 = tl.load(in_ptr4 + x2, xmask).to(tl.int1)
tmp14 = tl.load(in_out_ptr1 + x2, xmask)
tmp18 = tl.load(in_ptr5 + x1, xmask, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr6 + x1, xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr7 + x2, xmask).to(tl.int1)
tmp24 = tl.load(in_out_ptr2 + x2, xmask)
tmp28 = tl.load(in_ptr8 + x1, xmask, eviction_policy='evict_last')
tmp31 = tl.load(in_ptr9 + x1, xmask, eviction_policy='evict_last')
tmp33 = tl.load(in_ptr10 + x2, xmask).to(tl.int1)
tmp34 = tl.load(in_out_ptr3 + x2, xmask)
tmp38 = tl.load(in_ptr11 + x1, xmask, eviction_policy='evict_last')
tmp41 = tl.load(in_ptr12 + x1, xmask, eviction_policy='evict_last')
tmp3 = 4.0
tmp4 = tmp2 * tmp3
tmp5 = tl.where(tmp1, tmp2, tmp4)
tmp6 = 9e-15
tmp7 = tl.where(tmp0, tmp5, tmp6)
tmp9 = tmp7 - tmp8
tmp10 = tl_math.exp(tmp9)
tmp12 = tmp10 / tmp11
tmp15 = tmp14 * tmp3
tmp16 = tl.where(tmp13, tmp14, tmp15)
tmp17 = tl.where(tmp0, tmp16, tmp6)
tmp19 = tmp17 - tmp18
tmp20 = tl_math.exp(tmp19)
tmp22 = tmp20 / tmp21
tmp25 = tmp24 * tmp3
tmp26 = tl.where(tmp23, tmp24, tmp25)
tmp27 = tl.where(tmp0, tmp26, tmp6)
tmp29 = tmp27 - tmp28
tmp30 = tl_math.exp(tmp29)
tmp32 = tmp30 / tmp31
tmp35 = tmp34 * tmp3
tmp36 = tl.where(tmp33, tmp34, tmp35)
tmp37 = tl.where(tmp0, tmp36, tmp6)
tmp39 = tmp37 - tmp38
tmp40 = tl_math.exp(tmp39)
tmp42 = tmp40 / tmp41
tl.store(in_out_ptr0 + x2, tmp12, xmask)
tl.store(in_out_ptr1 + x2, tmp22, xmask)
tl.store(in_out_ptr2 + x2, tmp32, xmask)
tl.store(in_out_ptr3 + x2, tmp42, xmask)
@triton.jit
def triton_poi_fused_cat_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = xindex // 16
x2 = xindex
tmp0 = x0
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (4 * x1 + x0), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = 0.0
tmp7 = tmp5 > tmp6
tmp8 = 1.0
tmp9 = tmp5 * tmp8
tmp10 = libdevice.expm1(tmp9)
tmp11 = tmp10 * tmp8
tmp12 = tl.where(tmp7, tmp9, tmp11)
tmp13 = tl.full(tmp12.shape, 0.0, tmp12.dtype)
tmp14 = tl.where(tmp4, tmp12, tmp13)
tmp15 = tmp0 >= tmp3
tmp16 = tl.full([1], 8, tl.int64)
tmp17 = tmp0 < tmp16
tmp18 = tmp15 & tmp17
tmp19 = tl.load(in_ptr1 + (4 * x1 + (-4 + x0)), tmp18 & xmask,
eviction_policy='evict_last', other=0.0)
tmp20 = tmp19 > tmp6
tmp21 = tmp19 * tmp8
tmp22 = libdevice.expm1(tmp21)
tmp23 = tmp22 * tmp8
tmp24 = tl.where(tmp20, tmp21, tmp23)
tmp25 = tl.full(tmp24.shape, 0.0, tmp24.dtype)
tmp26 = tl.where(tmp18, tmp24, tmp25)
tmp27 = tmp0 >= tmp16
tmp28 = tl.full([1], 12, tl.int64)
tmp29 = tmp0 < tmp28
tmp30 = tmp27 & tmp29
tmp31 = tl.load(in_ptr2 + (4 * x1 + (-8 + x0)), tmp30 & xmask,
eviction_policy='evict_last', other=0.0)
tmp32 = tmp31 > tmp6
tmp33 = tmp31 * tmp8
tmp34 = libdevice.expm1(tmp33)
tmp35 = tmp34 * tmp8
tmp36 = tl.where(tmp32, tmp33, tmp35)
tmp37 = tl.full(tmp36.shape, 0.0, tmp36.dtype)
tmp38 = tl.where(tmp30, tmp36, tmp37)
tmp39 = tmp0 >= tmp28
tl.full([1], 16, tl.int64)
tmp42 = tl.load(in_ptr3 + (4 * x1 + (-12 + x0)), tmp39 & xmask,
eviction_policy='evict_last', other=0.0)
tmp43 = tmp42 > tmp6
tmp44 = tmp42 * tmp8
tmp45 = libdevice.expm1(tmp44)
tmp46 = tmp45 * tmp8
tmp47 = tl.where(tmp43, tmp44, tmp46)
tmp48 = tl.full(tmp47.shape, 0.0, tmp47.dtype)
tmp49 = tl.where(tmp39, tmp47, tmp48)
tmp50 = tl.where(tmp30, tmp38, tmp49)
tmp51 = tl.where(tmp18, tmp26, tmp50)
tmp52 = tl.where(tmp4, tmp14, tmp51)
tl.store(out_ptr0 + x2, tmp52, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (8, 1), (1, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (8, 1), (1, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (8, 1), (1, 1))
assert_size_stride(primals_9, (4, 4), (4, 1))
assert_size_stride(primals_10, (8, 1), (1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = torch.ops.aten.native_dropout.default(primals_1, 0.5, True)
del primals_1
buf1 = buf0[0]
del buf0
buf3 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf1, primals_2, out=buf3)
del primals_2
buf4 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(128)](buf3, buf4, 128, XBLOCK=128,
num_warps=4, num_stages=1)
buf5 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
extern_kernels.mm(buf4, primals_3, out=buf5)
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_leaky_relu_1[grid(16)](buf5, buf6, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf7 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_leaky_relu_1[grid(16)](primals_4, buf7, 16, XBLOCK
=16, num_warps=1, num_stages=1)
del primals_4
buf12 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf1, primals_5, out=buf12)
del primals_5
buf13 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
triton_poi_fused_cat_0[grid(128)](buf12, buf13, 128, XBLOCK=128,
num_warps=4, num_stages=1)
buf14 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
extern_kernels.mm(buf13, primals_6, out=buf14)
buf15 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_leaky_relu_1[grid(16)](buf14, buf15, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf20 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf1, primals_7, out=buf20)
del primals_7
buf21 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
triton_poi_fused_cat_0[grid(128)](buf20, buf21, 128, XBLOCK=128,
num_warps=4, num_stages=1)
buf22 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
extern_kernels.mm(buf21, primals_8, out=buf22)
buf23 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_leaky_relu_1[grid(16)](buf22, buf23, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf28 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf1, primals_9, out=buf28)
del primals_9
buf29 = empty_strided_cuda((16, 8), (8, 1), torch.float32)
triton_poi_fused_cat_0[grid(128)](buf28, buf29, 128, XBLOCK=128,
num_warps=4, num_stages=1)
buf30 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
extern_kernels.mm(buf29, primals_10, out=buf30)
buf31 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_leaky_relu_1[grid(16)](buf30, buf31, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf8 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf9 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf16 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf17 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf24 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf25 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf32 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
buf33 = empty_strided_cuda((4, 1), (1, 4), torch.float32)
triton_poi_fused__softmax_fill_leaky_relu_where_2[grid(4)](buf7,
buf6, buf5, buf15, buf14, buf23, buf22, buf31, buf30, buf8,
buf9, buf16, buf17, buf24, buf25, buf32, buf33, 4, XBLOCK=4,
num_warps=1, num_stages=1)
buf10 = reinterpret_tensor(buf5, (4, 4), (4, 1), 0)
del buf5
buf18 = reinterpret_tensor(buf14, (4, 4), (4, 1), 0)
del buf14
buf26 = reinterpret_tensor(buf22, (4, 4), (4, 1), 0)
del buf22
buf34 = reinterpret_tensor(buf30, (4, 4), (4, 1), 0)
del buf30
triton_poi_fused__softmax_fill_leaky_relu_where_3[grid(16)](buf10,
buf18, buf26, buf34, buf7, buf6, buf8, buf9, buf15, buf16,
buf17, buf23, buf24, buf25, buf31, buf32, buf33, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del buf16
del buf17
del buf24
del buf25
del buf32
del buf33
del buf8
del buf9
buf11 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf10, buf3, out=buf11)
buf19 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf18, buf12, out=buf19)
buf27 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf26, buf20, out=buf27)
buf35 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf34, buf28, out=buf35)
buf36 = empty_strided_cuda((4, 16), (16, 1), torch.float32)
triton_poi_fused_cat_4[grid(64)](buf11, buf19, buf27, buf35, buf36,
64, XBLOCK=64, num_warps=1, num_stages=1)
buf37 = torch.ops.aten.native_dropout.default(buf36, 0.5, True)
del buf36
buf38 = buf37[0]
buf39 = buf37[1]
del buf37
return (buf38, buf6, buf7, buf10, buf11, buf15, buf18, buf19, buf23,
buf26, buf27, buf31, buf34, buf35, buf39, reinterpret_tensor(buf28,
(4, 4), (1, 4), 0), reinterpret_tensor(buf29, (8, 16), (1, 8), 0),
reinterpret_tensor(primals_10, (1, 8), (1, 1), 0),
reinterpret_tensor(buf1, (4, 4), (1, 4), 0), reinterpret_tensor(
buf20, (4, 4), (1, 4), 0), reinterpret_tensor(buf21, (8, 16), (1, 8
), 0), reinterpret_tensor(primals_8, (1, 8), (1, 1), 0),
reinterpret_tensor(buf12, (4, 4), (1, 4), 0), reinterpret_tensor(
buf13, (8, 16), (1, 8), 0), reinterpret_tensor(primals_6, (1, 8), (
1, 1), 0), reinterpret_tensor(buf3, (4, 4), (1, 4), 0),
reinterpret_tensor(buf4, (8, 16), (1, 8), 0), reinterpret_tensor(
primals_3, (1, 8), (1, 1), 0))
class GraphAttentionLayer(nn.Module):
"""
Simple GAT layer, similar to https://arxiv.org/abs/1710.10903
"""
def __init__(self, in_features, out_features, dropout, alpha, concat=False
):
super(GraphAttentionLayer, self).__init__()
self.dropout = dropout
self.in_features = in_features
self.out_features = out_features
self.alpha = alpha
self.concat = concat
self.W = nn.Parameter(nn.init.xavier_uniform_(torch.Tensor(
in_features, out_features).type(torch.FloatTensor if torch.cuda
.is_available() else torch.FloatTensor), gain=np.sqrt(2.0)),
requires_grad=True)
self.a = nn.Parameter(nn.init.xavier_uniform_(torch.Tensor(2 *
out_features, 1).type(torch.FloatTensor if torch.cuda.
is_available() else torch.FloatTensor), gain=np.sqrt(2.0)),
requires_grad=True)
self.leakyrelu = nn.LeakyReLU(self.alpha)
def forward(self, input, adj):
h = torch.mm(input, self.W)
N = h.size()[0]
a_input = torch.cat([h.repeat(1, N).view(N * N, -1), h.repeat(N, 1)
], dim=1).view(N, -1, 2 * self.out_features)
e = self.leakyrelu(torch.matmul(a_input, self.a).squeeze(2))
zero_vec = torch.zeros_like(e)
zero_vec = zero_vec.fill_(9e-15)
attention = torch.where(adj > 0, e, zero_vec)
attention = F.softmax(attention, dim=1)
attention = F.dropout(attention, self.dropout, training=self.training)
h_prime = torch.matmul(attention, h)
if self.concat:
return F.elu(h_prime)
else:
return h_prime
def __repr__(self):
return self.__class__.__name__ + ' (' + str(self.in_features
) + ' -> ' + str(self.out_features) + ')'
class GATNew(nn.Module):
def __init__(self, nfeat, nhid, dropout, alpha, nheads):
super(GATNew, self).__init__()
self.dropout = dropout
self.attentions = [GraphAttentionLayer(nfeat, nhid, dropout=dropout,
alpha=alpha, concat=True) for _ in range(nheads)]
for i, attention in enumerate(self.attentions):
self.add_module('attention_{}'.format(i), attention)
def forward(self, input_0, input_1):
primals_1 = self.attention_0.W
primals_3 = self.attention_0.a
primals_2 = self.attention_1.W
primals_6 = self.attention_1.a
primals_4 = self.attention_2.W
primals_8 = self.attention_2.a
primals_5 = self.attention_3.W
primals_10 = self.attention_3.a
primals_7 = input_0
primals_9 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9, primals_10])
return output[0]
|
StrangeTcy/Q-BERT
|
GAT
| false | 14,465 |
[
"MIT"
] | 57 |
4e4cd4ddda3036d4bf7d878641592462189245d4
|
https://github.com/StrangeTcy/Q-BERT/tree/4e4cd4ddda3036d4bf7d878641592462189245d4
|
Unit3D
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/tu/ctudb4kblgaueia6crtpr2p7rsn3d22hehgavsmvoc56nw6lhlxk.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x_2 => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i1', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.full([1], 0, tl.int32)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp3 = 0.0
tmp4 = tmp2 <= tmp3
tl.store(in_out_ptr0 + (x0), tmp2, xmask)
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 1, 1, 1), (4, 1, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1, 1, 1), padding=(0, 0, 0), dilation=(1, 1, 1), transposed=False, output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
buf1 = buf0; del buf0 # reuse
buf2 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, buf2, 1024, grid=grid(1024), stream=stream0)
return (buf1, primals_1, primals_2, buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 1, 1, 1), (4, 1, 1, 1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
class Unit3D(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=(1, 1, 1),
stride=(1, 1, 1), padding='spatial_valid', activation_fn=F.relu,
use_batch_norm=False, use_bias=False):
"""Initializes Unit3D module."""
super(Unit3D, self).__init__()
self._output_channels = output_channels
self._kernel_shape = kernel_shape
self._stride = stride
self._use_batch_norm = use_batch_norm
self._activation_fn = activation_fn
self._use_bias = use_bias
self.padding = padding
if self._use_batch_norm:
self.bn = nn.BatchNorm3d(self._output_channels, eps=0.001,
momentum=0.01)
self.conv3d = nn.Conv3d(in_channels=in_channels, out_channels=self.
_output_channels, kernel_size=self._kernel_shape, stride=self.
_stride, padding=0, bias=self._use_bias)
def compute_pad(self, dim, s):
if s % self._stride[dim] == 0:
return max(self._kernel_shape[dim] - self._stride[dim], 0)
else:
return max(self._kernel_shape[dim] - s % self._stride[dim], 0)
def forward(self, x):
if self.padding == 'same':
_batch, _channel, t, h, w = x.size()
pad_t = self.compute_pad(0, t)
pad_h = self.compute_pad(1, h)
pad_w = self.compute_pad(2, w)
pad_t_f = pad_t // 2
pad_t_b = pad_t - pad_t_f
pad_h_f = pad_h // 2
pad_h_b = pad_h - pad_h_f
pad_w_f = pad_w // 2
pad_w_b = pad_w - pad_w_f
pad = [pad_w_f, pad_w_b, pad_h_f, pad_h_b, pad_t_f, pad_t_b]
x = F.pad(x, pad)
if self.padding == 'spatial_valid':
_batch, _channel, t, h, w = x.size()
pad_t = self.compute_pad(0, t)
pad_t_f = pad_t // 2
pad_t_b = pad_t - pad_t_f
pad = [0, 0, 0, 0, pad_t_f, pad_t_b]
x = F.pad(x, pad)
x = self.conv3d(x)
if self._use_batch_norm:
x = self.bn(x)
if self._activation_fn is not None:
x = self._activation_fn(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'output_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.full([1], 0, tl.int32)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp3 = 0.0
tmp4 = tmp2 <= tmp3
tl.store(in_out_ptr0 + x0, tmp2, xmask)
tl.store(out_ptr0 + x0, tmp4, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 1, 1, 1), (4, 1, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1,
1, 1), padding=(0, 0, 0), dilation=(1, 1, 1), transposed=False,
output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
buf1 = buf0
del buf0
buf2 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(1024)](buf1, buf2,
1024, XBLOCK=256, num_warps=4, num_stages=1)
return buf1, primals_1, primals_2, buf2
class Unit3DNew(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=(1, 1, 1),
stride=(1, 1, 1), padding='spatial_valid', activation_fn=F.relu,
use_batch_norm=False, use_bias=False):
"""Initializes Unit3D module."""
super(Unit3DNew, self).__init__()
self._output_channels = output_channels
self._kernel_shape = kernel_shape
self._stride = stride
self._use_batch_norm = use_batch_norm
self._activation_fn = activation_fn
self._use_bias = use_bias
self.padding = padding
if self._use_batch_norm:
self.bn = nn.BatchNorm3d(self._output_channels, eps=0.001,
momentum=0.01)
self.conv3d = nn.Conv3d(in_channels=in_channels, out_channels=self.
_output_channels, kernel_size=self._kernel_shape, stride=self.
_stride, padding=0, bias=self._use_bias)
def compute_pad(self, dim, s):
if s % self._stride[dim] == 0:
return max(self._kernel_shape[dim] - self._stride[dim], 0)
else:
return max(self._kernel_shape[dim] - s % self._stride[dim], 0)
def forward(self, input_0):
primals_2 = self.conv3d.weight
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0]
|
TencentYoutuResearch/ActionDetection-AFSD
|
Unit3D
| false | 14,466 |
[
"BSD-3-Clause"
] | 112 |
ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
https://github.com/TencentYoutuResearch/ActionDetection-AFSD/tree/ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
ResidualPointnet
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/cd/ccdx2fddfeepacucdj3letxeueyreism77yexvyvwrzrkvq6mbcn.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x => convolution
# Graph fragment:
# %convolution : [num_users=3] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1], [0], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 8
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/i3/ci3nuuurbsrmcufle642yc7udhwn4itsu6aptfssij5nzrnylpne.py
# Topologically Sorted Source Nodes: [x_1, relu], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# relu => relu
# x_1 => convolution_2
# Graph fragment:
# %convolution_2 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%convolution, %primals_6, %primals_7, [1], [0], [1], False, [0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_2,), kwargs = {})
triton_poi_fused_convolution_relu_1 = async_compile.triton('triton_poi_fused_convolution_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/3l/c3loa3lltbgdsalnzsa55tibvlplwj75kqk3tk6lhdxcmgauedq4.py
# Topologically Sorted Source Nodes: [x_short, x_2, add, x_3, max_1], Original ATen: [aten.convolution, aten.add, aten.relu, aten.max]
# Source node to ATen node mapping:
# add => add
# max_1 => getitem_1, max_1
# x_2 => convolution_3
# x_3 => relu_1
# x_short => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%convolution, %primals_4, %primals_5, [1], [0], [1], False, [0], 1), kwargs = {})
# %convolution_3 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_8, %primals_9, [1], [0], [1], False, [0], 1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convolution_3, %convolution_1), kwargs = {})
# %relu_1 : [num_users=3] = call_function[target=torch.ops.aten.relu.default](args = (%add,), kwargs = {})
# %max_1 : [num_users=2] = call_function[target=torch.ops.aten.max.dim](args = (%relu_1, 2, True), kwargs = {})
# %getitem_1 : [num_users=1] = call_function[target=operator.getitem](args = (%max_1, 1), kwargs = {})
triton_poi_fused_add_convolution_max_relu_2 = async_compile.triton('triton_poi_fused_add_convolution_max_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*i64', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_convolution_max_relu_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 10, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_convolution_max_relu_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (4*x2), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (4*x2), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (1 + (4*x2)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr2 + (1 + (4*x2)), xmask, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr0 + (2 + (4*x2)), xmask, eviction_policy='evict_last')
tmp18 = tl.load(in_ptr2 + (2 + (4*x2)), xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr0 + (3 + (4*x2)), xmask, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr2 + (3 + (4*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp7 = tl.full([1], 0, tl.int32)
tmp8 = triton_helpers.maximum(tmp7, tmp6)
tmp10 = tmp9 + tmp1
tmp12 = tmp11 + tmp4
tmp13 = tmp10 + tmp12
tmp14 = triton_helpers.maximum(tmp7, tmp13)
tmp15 = triton_helpers.maximum(tmp8, tmp14)
tmp17 = tmp16 + tmp1
tmp19 = tmp18 + tmp4
tmp20 = tmp17 + tmp19
tmp21 = triton_helpers.maximum(tmp7, tmp20)
tmp22 = triton_helpers.maximum(tmp15, tmp21)
tmp24 = tmp23 + tmp1
tmp26 = tmp25 + tmp4
tmp27 = tmp24 + tmp26
tmp28 = triton_helpers.maximum(tmp7, tmp27)
tmp29 = triton_helpers.maximum(tmp22, tmp28)
tmp30 = tmp8 > tmp14
tmp31 = tmp8 == tmp14
tmp32 = tmp8 != tmp8
tmp33 = tmp14 != tmp14
tmp34 = tmp32 > tmp33
tmp35 = tmp30 | tmp34
tmp36 = tmp32 & tmp33
tmp37 = tmp31 | tmp36
tmp38 = tl.full([1], 0, tl.int64)
tmp39 = tl.full([1], 1, tl.int64)
tmp40 = tmp38 < tmp39
tmp41 = tmp37 & tmp40
tmp42 = tmp35 | tmp41
tmp43 = tl.where(tmp42, tmp8, tmp14)
tmp44 = tl.where(tmp42, tmp38, tmp39)
tmp45 = tmp43 > tmp21
tmp46 = tmp43 == tmp21
tmp47 = tmp43 != tmp43
tmp48 = tmp21 != tmp21
tmp49 = tmp47 > tmp48
tmp50 = tmp45 | tmp49
tmp51 = tmp47 & tmp48
tmp52 = tmp46 | tmp51
tmp53 = tl.full([1], 2, tl.int64)
tmp54 = tmp44 < tmp53
tmp55 = tmp52 & tmp54
tmp56 = tmp50 | tmp55
tmp57 = tl.where(tmp56, tmp43, tmp21)
tmp58 = tl.where(tmp56, tmp44, tmp53)
tmp59 = tmp57 > tmp28
tmp60 = tmp57 == tmp28
tmp61 = tmp57 != tmp57
tmp62 = tmp28 != tmp28
tmp63 = tmp61 > tmp62
tmp64 = tmp59 | tmp63
tmp65 = tmp61 & tmp62
tmp66 = tmp60 | tmp65
tmp67 = tl.full([1], 3, tl.int64)
tmp68 = tmp58 < tmp67
tmp69 = tmp66 & tmp68
tmp70 = tmp64 | tmp69
tmp71 = tl.where(tmp70, tmp57, tmp28)
tmp72 = tl.where(tmp70, tmp58, tmp67)
tl.store(out_ptr0 + (x2), tmp29, xmask)
tl.store(out_ptr1 + (x2), tmp72, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/kz/ckzq4obr5gwbgwbozb4hwkytiwjx7chjimbruddxdjwwnnq3ag5m.py
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# x_4 => cat
# Graph fragment:
# %cat : [num_users=3] = call_function[target=torch.ops.aten.cat.default](args = ([%relu_1, %expand], 1), kwargs = {})
triton_poi_fused_cat_3 = async_compile.triton('triton_poi_fused_cat_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_3(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 8
x0 = xindex % 4
x2 = (xindex // 32)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (4*x1) + (16*x2)), tmp4 & xmask, other=0.0)
tmp6 = tl.load(in_ptr1 + (x1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp7 = tmp5 + tmp6
tmp8 = tl.load(in_ptr2 + (x0 + (4*x1) + (16*x2)), tmp4 & xmask, other=0.0)
tmp9 = tl.load(in_ptr3 + (x1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp10 = tmp8 + tmp9
tmp11 = tmp7 + tmp10
tmp12 = tl.full([1], 0, tl.int32)
tmp13 = triton_helpers.maximum(tmp12, tmp11)
tmp14 = tl.full(tmp13.shape, 0.0, tmp13.dtype)
tmp15 = tl.where(tmp4, tmp13, tmp14)
tmp16 = tmp0 >= tmp3
tmp17 = tl.full([1], 8, tl.int64)
tmp18 = tmp0 < tmp17
tmp19 = tl.load(in_ptr4 + ((4*x2) + ((-4) + x1)), tmp16 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp4, tmp15, tmp19)
tl.store(out_ptr0 + (x3), tmp20, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/w5/cw5ajqzhtyiwa2kub46dczrk6jbsvy6rnnisqji6bkga66kmkqiu.py
# Topologically Sorted Source Nodes: [x_short_4, x_18, add_4, x_19], Original ATen: [aten.convolution, aten.add, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# add_4 => add_4
# x_18 => convolution_15
# x_19 => relu_9
# x_short_4 => convolution_13
# Graph fragment:
# %convolution_13 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%cat_3, %primals_28, %primals_29, [1], [0], [1], False, [0], 1), kwargs = {})
# %convolution_15 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_8, %primals_32, %primals_33, [1], [0], [1], False, [0], 1), kwargs = {})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convolution_15, %convolution_13), kwargs = {})
# %relu_9 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_4,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_9, 0), kwargs = {})
triton_poi_fused_add_convolution_relu_threshold_backward_4 = async_compile.triton('triton_poi_fused_add_convolution_relu_threshold_backward_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*i1', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_convolution_relu_threshold_backward_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_convolution_relu_threshold_backward_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x3), xmask)
tmp4 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp7 = tl.full([1], 0, tl.int32)
tmp8 = triton_helpers.maximum(tmp7, tmp6)
tmp9 = 0.0
tmp10 = tmp8 <= tmp9
tl.store(out_ptr0 + (x3), tmp10, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15, primals_16, primals_17, primals_18, primals_19, primals_20, primals_21, primals_22, primals_23, primals_24, primals_25, primals_26, primals_27, primals_28, primals_29, primals_30, primals_31, primals_32, primals_33, primals_34, primals_35 = args
args.clear()
assert_size_stride(primals_1, (8, 4, 1), (4, 1, 1))
assert_size_stride(primals_2, (8, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_9, (4, ), (1, ))
assert_size_stride(primals_10, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_11, (4, ), (1, ))
assert_size_stride(primals_12, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_13, (4, ), (1, ))
assert_size_stride(primals_14, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_15, (4, ), (1, ))
assert_size_stride(primals_16, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_17, (4, ), (1, ))
assert_size_stride(primals_18, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_19, (4, ), (1, ))
assert_size_stride(primals_20, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_21, (4, ), (1, ))
assert_size_stride(primals_22, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_23, (4, ), (1, ))
assert_size_stride(primals_24, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_25, (4, ), (1, ))
assert_size_stride(primals_26, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_27, (4, ), (1, ))
assert_size_stride(primals_28, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_29, (4, ), (1, ))
assert_size_stride(primals_30, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_31, (4, ), (1, ))
assert_size_stride(primals_32, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_33, (4, ), (1, ))
assert_size_stride(primals_34, (4, 4), (4, 1))
assert_size_stride(primals_35, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4), (32, 4, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(buf1, primals_2, 128, grid=grid(128), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [x_short], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(buf1, primals_6, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf3, (4, 4, 4), (16, 4, 1))
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [x_1, relu], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf4, primals_7, 64, grid=grid(64), stream=stream0)
del primals_7
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf5 = extern_kernels.convolution(buf4, primals_8, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf5, (4, 4, 4), (16, 4, 1))
buf6 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf7 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_short, x_2, add, x_3, max_1], Original ATen: [aten.convolution, aten.add, aten.relu, aten.max]
triton_poi_fused_add_convolution_max_relu_2.run(buf5, primals_9, buf2, primals_5, buf6, buf7, 16, grid=grid(16), stream=stream0)
buf8 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.cat]
triton_poi_fused_cat_3.run(buf5, primals_9, buf2, primals_5, buf6, buf8, 128, grid=grid(128), stream=stream0)
# Topologically Sorted Source Nodes: [x_short_1], Original ATen: [aten.convolution]
buf9 = extern_kernels.convolution(buf8, primals_10, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf9, (4, 4, 4), (16, 4, 1))
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.convolution]
buf10 = extern_kernels.convolution(buf8, primals_12, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf10, (4, 4, 4), (16, 4, 1))
buf11 = buf10; del buf10 # reuse
# Topologically Sorted Source Nodes: [x_5, relu_2], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf11, primals_13, 64, grid=grid(64), stream=stream0)
del primals_13
# Topologically Sorted Source Nodes: [x_6], Original ATen: [aten.convolution]
buf12 = extern_kernels.convolution(buf11, primals_14, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf12, (4, 4, 4), (16, 4, 1))
buf13 = buf6; del buf6 # reuse
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_short_1, x_6, add_1, x_7, max_2], Original ATen: [aten.convolution, aten.add, aten.relu, aten.max]
triton_poi_fused_add_convolution_max_relu_2.run(buf12, primals_15, buf9, primals_11, buf13, buf14, 16, grid=grid(16), stream=stream0)
buf15 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_8], Original ATen: [aten.cat]
triton_poi_fused_cat_3.run(buf12, primals_15, buf9, primals_11, buf13, buf15, 128, grid=grid(128), stream=stream0)
# Topologically Sorted Source Nodes: [x_short_2], Original ATen: [aten.convolution]
buf16 = extern_kernels.convolution(buf15, primals_16, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf16, (4, 4, 4), (16, 4, 1))
# Topologically Sorted Source Nodes: [x_9], Original ATen: [aten.convolution]
buf17 = extern_kernels.convolution(buf15, primals_18, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf17, (4, 4, 4), (16, 4, 1))
buf18 = buf17; del buf17 # reuse
# Topologically Sorted Source Nodes: [x_9, relu_4], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf18, primals_19, 64, grid=grid(64), stream=stream0)
del primals_19
# Topologically Sorted Source Nodes: [x_10], Original ATen: [aten.convolution]
buf19 = extern_kernels.convolution(buf18, primals_20, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf19, (4, 4, 4), (16, 4, 1))
buf20 = buf13; del buf13 # reuse
buf21 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_short_2, x_10, add_2, x_11, max_3], Original ATen: [aten.convolution, aten.add, aten.relu, aten.max]
triton_poi_fused_add_convolution_max_relu_2.run(buf19, primals_21, buf16, primals_17, buf20, buf21, 16, grid=grid(16), stream=stream0)
buf22 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_12], Original ATen: [aten.cat]
triton_poi_fused_cat_3.run(buf19, primals_21, buf16, primals_17, buf20, buf22, 128, grid=grid(128), stream=stream0)
# Topologically Sorted Source Nodes: [x_short_3], Original ATen: [aten.convolution]
buf23 = extern_kernels.convolution(buf22, primals_22, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf23, (4, 4, 4), (16, 4, 1))
# Topologically Sorted Source Nodes: [x_13], Original ATen: [aten.convolution]
buf24 = extern_kernels.convolution(buf22, primals_24, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf24, (4, 4, 4), (16, 4, 1))
buf25 = buf24; del buf24 # reuse
# Topologically Sorted Source Nodes: [x_13, relu_6], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf25, primals_25, 64, grid=grid(64), stream=stream0)
del primals_25
# Topologically Sorted Source Nodes: [x_14], Original ATen: [aten.convolution]
buf26 = extern_kernels.convolution(buf25, primals_26, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf26, (4, 4, 4), (16, 4, 1))
buf27 = buf20; del buf20 # reuse
buf28 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_short_3, x_14, add_3, x_15, max_4], Original ATen: [aten.convolution, aten.add, aten.relu, aten.max]
triton_poi_fused_add_convolution_max_relu_2.run(buf26, primals_27, buf23, primals_23, buf27, buf28, 16, grid=grid(16), stream=stream0)
buf29 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_16], Original ATen: [aten.cat]
triton_poi_fused_cat_3.run(buf26, primals_27, buf23, primals_23, buf27, buf29, 128, grid=grid(128), stream=stream0)
# Topologically Sorted Source Nodes: [x_short_4], Original ATen: [aten.convolution]
buf30 = extern_kernels.convolution(buf29, primals_28, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf30, (4, 4, 4), (16, 4, 1))
# Topologically Sorted Source Nodes: [x_17], Original ATen: [aten.convolution]
buf31 = extern_kernels.convolution(buf29, primals_30, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf31, (4, 4, 4), (16, 4, 1))
buf32 = buf31; del buf31 # reuse
# Topologically Sorted Source Nodes: [x_17, relu_8], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf32, primals_31, 64, grid=grid(64), stream=stream0)
del primals_31
# Topologically Sorted Source Nodes: [x_18], Original ATen: [aten.convolution]
buf33 = extern_kernels.convolution(buf32, primals_32, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf33, (4, 4, 4), (16, 4, 1))
buf34 = reinterpret_tensor(buf27, (4, 4), (4, 1), 0); del buf27 # reuse
buf35 = empty_strided_cuda((4, 4), (4, 1), torch.int64)
# Topologically Sorted Source Nodes: [x_short_4, x_18, add_4, x_19, max_5], Original ATen: [aten.convolution, aten.add, aten.relu, aten.max]
triton_poi_fused_add_convolution_max_relu_2.run(buf33, primals_33, buf30, primals_29, buf34, buf35, 16, grid=grid(16), stream=stream0)
buf36 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_21], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_35, buf34, reinterpret_tensor(primals_34, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf36)
del primals_35
buf37 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_short_4, x_18, add_4, x_19], Original ATen: [aten.convolution, aten.add, aten.relu, aten.threshold_backward]
triton_poi_fused_add_convolution_relu_threshold_backward_4.run(buf33, primals_33, buf30, primals_29, buf37, 64, grid=grid(64), stream=stream0)
del buf30
del buf33
del primals_29
del primals_33
buf38 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_short_3, x_14, add_3, x_15], Original ATen: [aten.convolution, aten.add, aten.relu, aten.threshold_backward]
triton_poi_fused_add_convolution_relu_threshold_backward_4.run(buf26, primals_27, buf23, primals_23, buf38, 64, grid=grid(64), stream=stream0)
del buf23
del buf26
del primals_23
del primals_27
buf39 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_short_2, x_10, add_2, x_11], Original ATen: [aten.convolution, aten.add, aten.relu, aten.threshold_backward]
triton_poi_fused_add_convolution_relu_threshold_backward_4.run(buf19, primals_21, buf16, primals_17, buf39, 64, grid=grid(64), stream=stream0)
del buf16
del buf19
del primals_17
del primals_21
buf40 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_short_1, x_6, add_1, x_7], Original ATen: [aten.convolution, aten.add, aten.relu, aten.threshold_backward]
triton_poi_fused_add_convolution_relu_threshold_backward_4.run(buf12, primals_15, buf9, primals_11, buf40, 64, grid=grid(64), stream=stream0)
del buf12
del buf9
del primals_11
del primals_15
buf41 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_short, x_2, add, x_3], Original ATen: [aten.convolution, aten.add, aten.relu, aten.threshold_backward]
triton_poi_fused_add_convolution_relu_threshold_backward_4.run(buf5, primals_9, buf2, primals_5, buf41, 64, grid=grid(64), stream=stream0)
del buf2
del buf5
del primals_5
del primals_9
return (buf36, primals_1, primals_3, primals_4, primals_6, primals_8, primals_10, primals_12, primals_14, primals_16, primals_18, primals_20, primals_22, primals_24, primals_26, primals_28, primals_30, primals_32, buf1, buf4, buf7, buf8, buf11, buf14, buf15, buf18, buf21, buf22, buf25, buf28, buf29, buf32, buf34, primals_34, reinterpret_tensor(buf35, (4, 4, 1), (4, 1, 1), 0), buf37, buf38, buf39, buf40, buf41, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((8, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_14 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_15 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_16 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_17 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_18 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_19 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_20 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_21 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_22 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_23 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_24 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_25 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_26 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_27 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_28 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_29 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_30 = rand_strided((4, 8, 1), (8, 1, 1), device='cuda:0', dtype=torch.float32)
primals_31 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_32 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_33 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_34 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_35 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15, primals_16, primals_17, primals_18, primals_19, primals_20, primals_21, primals_22, primals_23, primals_24, primals_25, primals_26, primals_27, primals_28, primals_29, primals_30, primals_31, primals_32, primals_33, primals_34, primals_35])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch.nn as nn
class ResidualBlock(nn.Module):
def __init__(self, in_channels, out_channels, hidden_dim):
super().__init__()
self.fc_0 = nn.Conv1d(in_channels, hidden_dim, 1)
self.fc_1 = nn.Conv1d(hidden_dim, out_channels, 1)
self.activation = nn.ReLU()
if in_channels != out_channels:
self.shortcut = nn.Conv1d(in_channels, out_channels, 1)
else:
self.shortcut = nn.Identity()
nn.init.zeros_(self.fc_1.weight)
def forward(self, x):
x_short = self.shortcut(x)
x = self.fc_0(x)
x = self.fc_1(self.activation(x))
x = self.activation(x + x_short)
return x
class ResidualPointnet(nn.Module):
""" PointNet-based encoder network with ResNet blocks.
Args:
c_dim (int): dimension of latent code c
dim (int): input points dimension
hidden_dim (int): hidden dimension of the network
"""
def __init__(self, in_channels, out_channels, hidden_dim, segmentation=
False):
super().__init__()
self.fc_in = nn.Conv1d(in_channels, 2 * hidden_dim, 1)
self.block_0 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_1 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_2 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_3 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_4 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.segmentation = segmentation
if self.segmentation:
self.fc_out = nn.Conv1d(2 * hidden_dim, out_channels, 1)
else:
self.fc_out = nn.Linear(hidden_dim, out_channels)
def forward(self, x):
x = self.fc_in(x)
x = self.block_0(x)
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
x = self.block_1(x)
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
x = self.block_2(x)
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
x = self.block_3(x)
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
x = self.block_4(x)
if self.segmentation:
x_pool = torch.max(x, dim=2, keepdim=True)[0].expand_as(x)
x = torch.cat([x, x_pool], dim=1)
else:
x = torch.max(x, dim=2)[0]
x = self.fc_out(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'hidden_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.utils.data
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 8
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_convolution_max_relu_2(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + 4 * x2, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + 4 * x2, xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (1 + 4 * x2), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr2 + (1 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp16 = tl.load(in_ptr0 + (2 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp18 = tl.load(in_ptr2 + (2 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp23 = tl.load(in_ptr0 + (3 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp25 = tl.load(in_ptr2 + (3 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp7 = tl.full([1], 0, tl.int32)
tmp8 = triton_helpers.maximum(tmp7, tmp6)
tmp10 = tmp9 + tmp1
tmp12 = tmp11 + tmp4
tmp13 = tmp10 + tmp12
tmp14 = triton_helpers.maximum(tmp7, tmp13)
tmp15 = triton_helpers.maximum(tmp8, tmp14)
tmp17 = tmp16 + tmp1
tmp19 = tmp18 + tmp4
tmp20 = tmp17 + tmp19
tmp21 = triton_helpers.maximum(tmp7, tmp20)
tmp22 = triton_helpers.maximum(tmp15, tmp21)
tmp24 = tmp23 + tmp1
tmp26 = tmp25 + tmp4
tmp27 = tmp24 + tmp26
tmp28 = triton_helpers.maximum(tmp7, tmp27)
tmp29 = triton_helpers.maximum(tmp22, tmp28)
tmp30 = tmp8 > tmp14
tmp31 = tmp8 == tmp14
tmp32 = tmp8 != tmp8
tmp33 = tmp14 != tmp14
tmp34 = tmp32 > tmp33
tmp35 = tmp30 | tmp34
tmp36 = tmp32 & tmp33
tmp37 = tmp31 | tmp36
tmp38 = tl.full([1], 0, tl.int64)
tmp39 = tl.full([1], 1, tl.int64)
tmp40 = tmp38 < tmp39
tmp41 = tmp37 & tmp40
tmp42 = tmp35 | tmp41
tmp43 = tl.where(tmp42, tmp8, tmp14)
tmp44 = tl.where(tmp42, tmp38, tmp39)
tmp45 = tmp43 > tmp21
tmp46 = tmp43 == tmp21
tmp47 = tmp43 != tmp43
tmp48 = tmp21 != tmp21
tmp49 = tmp47 > tmp48
tmp50 = tmp45 | tmp49
tmp51 = tmp47 & tmp48
tmp52 = tmp46 | tmp51
tmp53 = tl.full([1], 2, tl.int64)
tmp54 = tmp44 < tmp53
tmp55 = tmp52 & tmp54
tmp56 = tmp50 | tmp55
tmp57 = tl.where(tmp56, tmp43, tmp21)
tmp58 = tl.where(tmp56, tmp44, tmp53)
tmp59 = tmp57 > tmp28
tmp60 = tmp57 == tmp28
tmp61 = tmp57 != tmp57
tmp62 = tmp28 != tmp28
tmp63 = tmp61 > tmp62
tmp64 = tmp59 | tmp63
tmp65 = tmp61 & tmp62
tmp66 = tmp60 | tmp65
tmp67 = tl.full([1], 3, tl.int64)
tmp68 = tmp58 < tmp67
tmp69 = tmp66 & tmp68
tmp70 = tmp64 | tmp69
tl.where(tmp70, tmp57, tmp28)
tmp72 = tl.where(tmp70, tmp58, tmp67)
tl.store(out_ptr0 + x2, tmp29, xmask)
tl.store(out_ptr1 + x2, tmp72, xmask)
@triton.jit
def triton_poi_fused_cat_3(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 8
x0 = xindex % 4
x2 = xindex // 32
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 4 * x1 + 16 * x2), tmp4 & xmask, other=0.0)
tmp6 = tl.load(in_ptr1 + x1, tmp4 & xmask, eviction_policy='evict_last',
other=0.0)
tmp7 = tmp5 + tmp6
tmp8 = tl.load(in_ptr2 + (x0 + 4 * x1 + 16 * x2), tmp4 & xmask, other=0.0)
tmp9 = tl.load(in_ptr3 + x1, tmp4 & xmask, eviction_policy='evict_last',
other=0.0)
tmp10 = tmp8 + tmp9
tmp11 = tmp7 + tmp10
tmp12 = tl.full([1], 0, tl.int32)
tmp13 = triton_helpers.maximum(tmp12, tmp11)
tmp14 = tl.full(tmp13.shape, 0.0, tmp13.dtype)
tmp15 = tl.where(tmp4, tmp13, tmp14)
tmp16 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp19 = tl.load(in_ptr4 + (4 * x2 + (-4 + x1)), tmp16 & xmask,
eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp4, tmp15, tmp19)
tl.store(out_ptr0 + x3, tmp20, xmask)
@triton.jit
def triton_poi_fused_add_convolution_relu_threshold_backward_4(in_ptr0,
in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x3, xmask)
tmp4 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp7 = tl.full([1], 0, tl.int32)
tmp8 = triton_helpers.maximum(tmp7, tmp6)
tmp9 = 0.0
tmp10 = tmp8 <= tmp9
tl.store(out_ptr0 + x3, tmp10, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13, primals_14, primals_15, primals_16, primals_17,
primals_18, primals_19, primals_20, primals_21, primals_22,
primals_23, primals_24, primals_25, primals_26, primals_27,
primals_28, primals_29, primals_30, primals_31, primals_32,
primals_33, primals_34, primals_35) = args
args.clear()
assert_size_stride(primals_1, (8, 4, 1), (4, 1, 1))
assert_size_stride(primals_2, (8,), (1,))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_9, (4,), (1,))
assert_size_stride(primals_10, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_11, (4,), (1,))
assert_size_stride(primals_12, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_13, (4,), (1,))
assert_size_stride(primals_14, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_15, (4,), (1,))
assert_size_stride(primals_16, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_17, (4,), (1,))
assert_size_stride(primals_18, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_19, (4,), (1,))
assert_size_stride(primals_20, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_21, (4,), (1,))
assert_size_stride(primals_22, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_23, (4,), (1,))
assert_size_stride(primals_24, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_25, (4,), (1,))
assert_size_stride(primals_26, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_27, (4,), (1,))
assert_size_stride(primals_28, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_29, (4,), (1,))
assert_size_stride(primals_30, (4, 8, 1), (8, 1, 1))
assert_size_stride(primals_31, (4,), (1,))
assert_size_stride(primals_32, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_33, (4,), (1,))
assert_size_stride(primals_34, (4, 4), (4, 1))
assert_size_stride(primals_35, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4), (32, 4, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(128)](buf1, primals_2, 128,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
buf3 = extern_kernels.convolution(buf1, primals_6, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf3, (4, 4, 4), (16, 4, 1))
buf4 = buf3
del buf3
triton_poi_fused_convolution_relu_1[grid(64)](buf4, primals_7, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_7
buf5 = extern_kernels.convolution(buf4, primals_8, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf5, (4, 4, 4), (16, 4, 1))
buf6 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf7 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
triton_poi_fused_add_convolution_max_relu_2[grid(16)](buf5,
primals_9, buf2, primals_5, buf6, buf7, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf8 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_3[grid(128)](buf5, primals_9, buf2, primals_5,
buf6, buf8, 128, XBLOCK=128, num_warps=4, num_stages=1)
buf9 = extern_kernels.convolution(buf8, primals_10, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf9, (4, 4, 4), (16, 4, 1))
buf10 = extern_kernels.convolution(buf8, primals_12, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf10, (4, 4, 4), (16, 4, 1))
buf11 = buf10
del buf10
triton_poi_fused_convolution_relu_1[grid(64)](buf11, primals_13, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_13
buf12 = extern_kernels.convolution(buf11, primals_14, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf12, (4, 4, 4), (16, 4, 1))
buf13 = buf6
del buf6
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
triton_poi_fused_add_convolution_max_relu_2[grid(16)](buf12,
primals_15, buf9, primals_11, buf13, buf14, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf15 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_3[grid(128)](buf12, primals_15, buf9,
primals_11, buf13, buf15, 128, XBLOCK=128, num_warps=4,
num_stages=1)
buf16 = extern_kernels.convolution(buf15, primals_16, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf16, (4, 4, 4), (16, 4, 1))
buf17 = extern_kernels.convolution(buf15, primals_18, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf17, (4, 4, 4), (16, 4, 1))
buf18 = buf17
del buf17
triton_poi_fused_convolution_relu_1[grid(64)](buf18, primals_19, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_19
buf19 = extern_kernels.convolution(buf18, primals_20, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf19, (4, 4, 4), (16, 4, 1))
buf20 = buf13
del buf13
buf21 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
triton_poi_fused_add_convolution_max_relu_2[grid(16)](buf19,
primals_21, buf16, primals_17, buf20, buf21, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf22 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_3[grid(128)](buf19, primals_21, buf16,
primals_17, buf20, buf22, 128, XBLOCK=128, num_warps=4,
num_stages=1)
buf23 = extern_kernels.convolution(buf22, primals_22, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf23, (4, 4, 4), (16, 4, 1))
buf24 = extern_kernels.convolution(buf22, primals_24, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf24, (4, 4, 4), (16, 4, 1))
buf25 = buf24
del buf24
triton_poi_fused_convolution_relu_1[grid(64)](buf25, primals_25, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_25
buf26 = extern_kernels.convolution(buf25, primals_26, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf26, (4, 4, 4), (16, 4, 1))
buf27 = buf20
del buf20
buf28 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.int64)
triton_poi_fused_add_convolution_max_relu_2[grid(16)](buf26,
primals_27, buf23, primals_23, buf27, buf28, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf29 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_3[grid(128)](buf26, primals_27, buf23,
primals_23, buf27, buf29, 128, XBLOCK=128, num_warps=4,
num_stages=1)
buf30 = extern_kernels.convolution(buf29, primals_28, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf30, (4, 4, 4), (16, 4, 1))
buf31 = extern_kernels.convolution(buf29, primals_30, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf31, (4, 4, 4), (16, 4, 1))
buf32 = buf31
del buf31
triton_poi_fused_convolution_relu_1[grid(64)](buf32, primals_31, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_31
buf33 = extern_kernels.convolution(buf32, primals_32, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf33, (4, 4, 4), (16, 4, 1))
buf34 = reinterpret_tensor(buf27, (4, 4), (4, 1), 0)
del buf27
buf35 = empty_strided_cuda((4, 4), (4, 1), torch.int64)
triton_poi_fused_add_convolution_max_relu_2[grid(16)](buf33,
primals_33, buf30, primals_29, buf34, buf35, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf36 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_35, buf34, reinterpret_tensor(
primals_34, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf36)
del primals_35
buf37 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
triton_poi_fused_add_convolution_relu_threshold_backward_4[grid(64)](
buf33, primals_33, buf30, primals_29, buf37, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf30
del buf33
del primals_29
del primals_33
buf38 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
triton_poi_fused_add_convolution_relu_threshold_backward_4[grid(64)](
buf26, primals_27, buf23, primals_23, buf38, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf23
del buf26
del primals_23
del primals_27
buf39 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
triton_poi_fused_add_convolution_relu_threshold_backward_4[grid(64)](
buf19, primals_21, buf16, primals_17, buf39, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf16
del buf19
del primals_17
del primals_21
buf40 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
triton_poi_fused_add_convolution_relu_threshold_backward_4[grid(64)](
buf12, primals_15, buf9, primals_11, buf40, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf12
del buf9
del primals_11
del primals_15
buf41 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
triton_poi_fused_add_convolution_relu_threshold_backward_4[grid(64)](
buf5, primals_9, buf2, primals_5, buf41, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf2
del buf5
del primals_5
del primals_9
return (buf36, primals_1, primals_3, primals_4, primals_6, primals_8,
primals_10, primals_12, primals_14, primals_16, primals_18,
primals_20, primals_22, primals_24, primals_26, primals_28,
primals_30, primals_32, buf1, buf4, buf7, buf8, buf11, buf14, buf15,
buf18, buf21, buf22, buf25, buf28, buf29, buf32, buf34, primals_34,
reinterpret_tensor(buf35, (4, 4, 1), (4, 1, 1), 0), buf37, buf38,
buf39, buf40, buf41)
class ResidualBlock(nn.Module):
def __init__(self, in_channels, out_channels, hidden_dim):
super().__init__()
self.fc_0 = nn.Conv1d(in_channels, hidden_dim, 1)
self.fc_1 = nn.Conv1d(hidden_dim, out_channels, 1)
self.activation = nn.ReLU()
if in_channels != out_channels:
self.shortcut = nn.Conv1d(in_channels, out_channels, 1)
else:
self.shortcut = nn.Identity()
nn.init.zeros_(self.fc_1.weight)
def forward(self, x):
x_short = self.shortcut(x)
x = self.fc_0(x)
x = self.fc_1(self.activation(x))
x = self.activation(x + x_short)
return x
class ResidualPointnetNew(nn.Module):
""" PointNet-based encoder network with ResNet blocks.
Args:
c_dim (int): dimension of latent code c
dim (int): input points dimension
hidden_dim (int): hidden dimension of the network
"""
def __init__(self, in_channels, out_channels, hidden_dim, segmentation=
False):
super().__init__()
self.fc_in = nn.Conv1d(in_channels, 2 * hidden_dim, 1)
self.block_0 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_1 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_2 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_3 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.block_4 = ResidualBlock(2 * hidden_dim, hidden_dim, hidden_dim)
self.segmentation = segmentation
if self.segmentation:
self.fc_out = nn.Conv1d(2 * hidden_dim, out_channels, 1)
else:
self.fc_out = nn.Linear(hidden_dim, out_channels)
def forward(self, input_0):
primals_1 = self.fc_in.weight
primals_2 = self.fc_in.bias
primals_4 = self.block_0.fc_0.weight
primals_5 = self.block_0.fc_0.bias
primals_8 = self.block_0.fc_1.weight
primals_7 = self.block_0.fc_1.bias
primals_6 = self.block_0.shortcut.weight
primals_9 = self.block_0.shortcut.bias
primals_10 = self.block_1.fc_0.weight
primals_11 = self.block_1.fc_0.bias
primals_14 = self.block_1.fc_1.weight
primals_13 = self.block_1.fc_1.bias
primals_12 = self.block_1.shortcut.weight
primals_15 = self.block_1.shortcut.bias
primals_16 = self.block_2.fc_0.weight
primals_17 = self.block_2.fc_0.bias
primals_20 = self.block_2.fc_1.weight
primals_19 = self.block_2.fc_1.bias
primals_18 = self.block_2.shortcut.weight
primals_21 = self.block_2.shortcut.bias
primals_22 = self.block_3.fc_0.weight
primals_23 = self.block_3.fc_0.bias
primals_26 = self.block_3.fc_1.weight
primals_25 = self.block_3.fc_1.bias
primals_24 = self.block_3.shortcut.weight
primals_27 = self.block_3.shortcut.bias
primals_28 = self.block_4.fc_0.weight
primals_29 = self.block_4.fc_0.bias
primals_32 = self.block_4.fc_1.weight
primals_31 = self.block_4.fc_1.bias
primals_30 = self.block_4.shortcut.weight
primals_33 = self.block_4.shortcut.bias
primals_34 = self.fc_out.weight
primals_35 = self.fc_out.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13, primals_14,
primals_15, primals_16, primals_17, primals_18, primals_19,
primals_20, primals_21, primals_22, primals_23, primals_24,
primals_25, primals_26, primals_27, primals_28, primals_29,
primals_30, primals_31, primals_32, primals_33, primals_34,
primals_35])
return output[0]
|
StructuralNeurobiologyLab/LightConvPoint
|
ResidualPointnet
| false | 14,467 |
[
"Apache-2.0"
] | 58 |
3f353f45e9e910fa390a74520dfd478e3e88f104
|
https://github.com/StructuralNeurobiologyLab/LightConvPoint/tree/3f353f45e9e910fa390a74520dfd478e3e88f104
|
GroupedChannelNorm
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/iw/ciwneupqau4kn2jj7timx3ubvnhllst5w6ftifgvb35ru3sw54dr.py
# Topologically Sorted Source Nodes: [mean, sub, std, add, x_norm], Original ATen: [aten.mean, aten.sub, aten.std, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# mean => mean
# std => sqrt, var
# sub => sub
# x_norm => div
# Graph fragment:
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%view, [2], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view, %mean), kwargs = {})
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%view, [2]), kwargs = {correction: 1.0, keepdim: True})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%var,), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sqrt, 1e-07), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %add), kwargs = {})
triton_poi_fused_add_div_mean_std_sub_0 = async_compile.triton('triton_poi_fused_add_div_mean_std_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mean_std_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mean_std_sub_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = 4.0
tmp9 = tmp7 / tmp8
tmp10 = tmp0 - tmp9
tmp11 = tmp1 - tmp9
tmp12 = tmp11 * tmp11
tmp13 = tmp2 - tmp9
tmp14 = tmp13 * tmp13
tmp15 = tmp12 + tmp14
tmp16 = tmp4 - tmp9
tmp17 = tmp16 * tmp16
tmp18 = tmp15 + tmp17
tmp19 = tmp6 - tmp9
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = 3.0
tmp23 = tmp21 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-07
tmp26 = tmp24 + tmp25
tmp27 = tmp10 / tmp26
tl.store(out_ptr0 + (x3), tmp27, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1, 4, 4, 4), (64, 1, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mean, sub, std, add, x_norm], Original ATen: [aten.mean, aten.sub, aten.std, aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mean_std_sub_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch
import torch.nn as nn
class GroupedChannelNorm(nn.Module):
def __init__(self, num_groups):
super().__init__()
self.num_groups = num_groups
def forward(self, x):
shape = list(x.shape)
new_shape = [shape[0], self.num_groups, shape[1] // self.num_groups
] + shape[2:]
x = x.view(*new_shape)
mean = x.mean(dim=2, keepdim=True)
std = x.std(dim=2, keepdim=True)
x_norm = (x - mean) / (std + 1e-07)
return x_norm.view(*shape)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'num_groups': 1}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.utils.data
import torch
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_div_mean_std_sub_0(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = 4.0
tmp9 = tmp7 / tmp8
tmp10 = tmp0 - tmp9
tmp11 = tmp1 - tmp9
tmp12 = tmp11 * tmp11
tmp13 = tmp2 - tmp9
tmp14 = tmp13 * tmp13
tmp15 = tmp12 + tmp14
tmp16 = tmp4 - tmp9
tmp17 = tmp16 * tmp16
tmp18 = tmp15 + tmp17
tmp19 = tmp6 - tmp9
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = 3.0
tmp23 = tmp21 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-07
tmp26 = tmp24 + tmp25
tmp27 = tmp10 / tmp26
tl.store(out_ptr0 + x3, tmp27, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1, 4, 4, 4), (64, 1, 16, 4, 1), torch
.float32)
get_raw_stream(0)
triton_poi_fused_add_div_mean_std_sub_0[grid(256)](arg0_1, buf0,
256, XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
return reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0),
class GroupedChannelNormNew(nn.Module):
def __init__(self, num_groups):
super().__init__()
self.num_groups = num_groups
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
Theomat/colorization-av-enseirb-2020
|
GroupedChannelNorm
| false | 14,468 |
[
"Apache-2.0"
] | 1,422 |
c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
https://github.com/Theomat/colorization-av-enseirb-2020/tree/c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
TransposedConv1d
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/hn/chn2p54izye7krngpydarmkqibmpimfjowvk2nuzlpg5j73vpgdh.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x_1 => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%squeeze,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 8)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3), (12, 3, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1, 4, 4), (16, 4, 1), 0), primals_1, stride=(2,), padding=(1,), dilation=(1,), transposed=True, output_padding=(1,), groups=1, bias=None)
assert_size_stride(buf0, (1, 4, 8), (32, 8, 1))
buf1 = reinterpret_tensor(buf0, (4, 8), (8, 1), 0); del buf0 # reuse
buf2 = empty_strided_cuda((4, 8), (8, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf2, 32, grid=grid(32), stream=stream0)
del primals_2
return (buf1, primals_1, reinterpret_tensor(primals_3, (1, 4, 4), (16, 4, 1), 0), buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 3), (12, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
class TransposedConv1d(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=3, stride
=2, padding=1, output_padding=1, activation_fn=F.relu,
use_batch_norm=False, use_bias=True):
super(TransposedConv1d, self).__init__()
self._use_batch_norm = use_batch_norm
self._activation_fn = activation_fn
self.transposed_conv1d = nn.ConvTranspose1d(in_channels,
output_channels, kernel_shape, stride, padding=padding,
output_padding=output_padding, bias=use_bias)
if self._use_batch_norm:
self.bn = nn.BatchNorm3d(self._output_channels, eps=0.001,
momentum=0.01)
def forward(self, x):
x = self.transposed_conv1d(x)
if self._use_batch_norm:
x = self.bn(x)
if self._activation_fn is not None:
x = self._activation_fn(x)
return x
def get_inputs():
return [torch.rand([4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'output_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 8
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3), (12, 3, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1,
4, 4), (16, 4, 1), 0), primals_1, stride=(2,), padding=(1,),
dilation=(1,), transposed=True, output_padding=(1,), groups=1,
bias=None)
assert_size_stride(buf0, (1, 4, 8), (32, 8, 1))
buf1 = reinterpret_tensor(buf0, (4, 8), (8, 1), 0)
del buf0
buf2 = empty_strided_cuda((4, 8), (8, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(32)](buf1,
primals_2, buf2, 32, XBLOCK=32, num_warps=1, num_stages=1)
del primals_2
return buf1, primals_1, reinterpret_tensor(primals_3, (1, 4, 4), (16, 4,
1), 0), buf2
class TransposedConv1dNew(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=3, stride
=2, padding=1, output_padding=1, activation_fn=F.relu,
use_batch_norm=False, use_bias=True):
super(TransposedConv1dNew, self).__init__()
self._use_batch_norm = use_batch_norm
self._activation_fn = activation_fn
self.transposed_conv1d = nn.ConvTranspose1d(in_channels,
output_channels, kernel_shape, stride, padding=padding,
output_padding=output_padding, bias=use_bias)
if self._use_batch_norm:
self.bn = nn.BatchNorm3d(self._output_channels, eps=0.001,
momentum=0.01)
def forward(self, input_0):
primals_1 = self.transposed_conv1d.weight
primals_2 = self.transposed_conv1d.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
TencentYoutuResearch/ActionDetection-AFSD
|
TransposedConv1d
| false | 14,469 |
[
"BSD-3-Clause"
] | 112 |
ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
https://github.com/TencentYoutuResearch/ActionDetection-AFSD/tree/ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
PoolingF
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ku/ckuooqx6q27vkk7tqqmkpsydmkdv3enwjhqctsyozm4owufb7yb6.py
# Topologically Sorted Source Nodes: [input_1], Original ATen: [aten.adaptive_max_pool2d]
# Source node to ATen node mapping:
# input_1 => adaptive_max_pool2d
# Graph fragment:
# %adaptive_max_pool2d : [num_users=1] = call_function[target=torch.ops.aten.adaptive_max_pool2d.default](args = (%arg0_1, [1, 1]), kwargs = {})
triton_poi_fused_adaptive_max_pool2d_0 = async_compile.triton('triton_poi_fused_adaptive_max_pool2d_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_adaptive_max_pool2d_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 16, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_adaptive_max_pool2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (16*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (16*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (16*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (16*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (4 + (16*x0)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (5 + (16*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (6 + (16*x0)), xmask, eviction_policy='evict_last')
tmp13 = tl.load(in_ptr0 + (7 + (16*x0)), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr0 + (8 + (16*x0)), xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr0 + (9 + (16*x0)), xmask, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr0 + (10 + (16*x0)), xmask, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (11 + (16*x0)), xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr0 + (12 + (16*x0)), xmask, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr0 + (13 + (16*x0)), xmask, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr0 + (14 + (16*x0)), xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr0 + (15 + (16*x0)), xmask, eviction_policy='evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp8 = triton_helpers.maximum(tmp7, tmp6)
tmp10 = triton_helpers.maximum(tmp9, tmp8)
tmp12 = triton_helpers.maximum(tmp11, tmp10)
tmp14 = triton_helpers.maximum(tmp13, tmp12)
tmp16 = triton_helpers.maximum(tmp15, tmp14)
tmp18 = triton_helpers.maximum(tmp17, tmp16)
tmp20 = triton_helpers.maximum(tmp19, tmp18)
tmp22 = triton_helpers.maximum(tmp21, tmp20)
tmp24 = triton_helpers.maximum(tmp23, tmp22)
tmp26 = triton_helpers.maximum(tmp25, tmp24)
tmp28 = triton_helpers.maximum(tmp27, tmp26)
tmp30 = triton_helpers.maximum(tmp29, tmp28)
tl.store(out_ptr0 + (x0), tmp30, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/oy/coyfug4lezxxlpdfds2pbopjbi4465nns5jojyud3gmda7742ba5.py
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, add, out], Original ATen: [aten.pow, aten.sum, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# norm => pow_2
# out => div
# pow_1 => pow_1
# sum_1 => sum_1
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%getitem, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [1], True), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sum_1, 0.5), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%pow_2, 1e-07), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%getitem, %add), kwargs = {})
triton_poi_fused_add_div_pow_sum_1 = async_compile.triton('triton_poi_fused_add_div_pow_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_pow_sum_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_pow_sum_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x2), tmp15, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
# Topologically Sorted Source Nodes: [input_1], Original ATen: [aten.adaptive_max_pool2d]
stream0 = get_raw_stream(0)
triton_poi_fused_adaptive_max_pool2d_0.run(arg0_1, buf0, 16, grid=grid(16), stream=stream0)
del arg0_1
buf1 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, add, out], Original ATen: [aten.pow, aten.sum, aten.add, aten.div]
triton_poi_fused_add_div_pow_sum_1.run(buf0, buf1, 16, grid=grid(16), stream=stream0)
del buf0
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch
import torch.nn as nn
class Normalize(nn.Module):
def __init__(self, power=2):
super(Normalize, self).__init__()
self.power = power
def forward(self, x):
norm = x.pow(self.power).sum(1, keepdim=True).pow(1.0 / self.power)
out = x.div(norm + 1e-07)
return out
class PoolingF(nn.Module):
def __init__(self):
super(PoolingF, self).__init__()
model = [nn.AdaptiveMaxPool2d(1)]
self.model = nn.Sequential(*model)
self.l2norm = Normalize(2)
def forward(self, x):
return self.l2norm(self.model(x))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
import torch.utils.data
import torch
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_adaptive_max_pool2d_0(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 16 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp3 = tl.load(in_ptr0 + (2 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp5 = tl.load(in_ptr0 + (3 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp7 = tl.load(in_ptr0 + (4 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp9 = tl.load(in_ptr0 + (5 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp11 = tl.load(in_ptr0 + (6 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp13 = tl.load(in_ptr0 + (7 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp15 = tl.load(in_ptr0 + (8 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp17 = tl.load(in_ptr0 + (9 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp19 = tl.load(in_ptr0 + (10 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp21 = tl.load(in_ptr0 + (11 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp23 = tl.load(in_ptr0 + (12 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp25 = tl.load(in_ptr0 + (13 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp27 = tl.load(in_ptr0 + (14 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp29 = tl.load(in_ptr0 + (15 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp8 = triton_helpers.maximum(tmp7, tmp6)
tmp10 = triton_helpers.maximum(tmp9, tmp8)
tmp12 = triton_helpers.maximum(tmp11, tmp10)
tmp14 = triton_helpers.maximum(tmp13, tmp12)
tmp16 = triton_helpers.maximum(tmp15, tmp14)
tmp18 = triton_helpers.maximum(tmp17, tmp16)
tmp20 = triton_helpers.maximum(tmp19, tmp18)
tmp22 = triton_helpers.maximum(tmp21, tmp20)
tmp24 = triton_helpers.maximum(tmp23, tmp22)
tmp26 = triton_helpers.maximum(tmp25, tmp24)
tmp28 = triton_helpers.maximum(tmp27, tmp26)
tmp30 = triton_helpers.maximum(tmp29, tmp28)
tl.store(out_ptr0 + x0, tmp30, xmask)
@triton.jit
def triton_poi_fused_add_div_pow_sum_1(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + x2, tmp15, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
get_raw_stream(0)
triton_poi_fused_adaptive_max_pool2d_0[grid(16)](arg0_1, buf0, 16,
XBLOCK=16, num_warps=1, num_stages=1)
del arg0_1
buf1 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
triton_poi_fused_add_div_pow_sum_1[grid(16)](buf0, buf1, 16, XBLOCK
=16, num_warps=1, num_stages=1)
del buf0
return buf1,
class Normalize(nn.Module):
def __init__(self, power=2):
super(Normalize, self).__init__()
self.power = power
def forward(self, x):
norm = x.pow(self.power).sum(1, keepdim=True).pow(1.0 / self.power)
out = x.div(norm + 1e-07)
return out
class PoolingFNew(nn.Module):
def __init__(self):
super(PoolingFNew, self).__init__()
model = [nn.AdaptiveMaxPool2d(1)]
self.model = nn.Sequential(*model)
self.l2norm = Normalize(2)
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
Theomat/colorization-av-enseirb-2020
|
PoolingF
| false | 14,470 |
[
"Apache-2.0"
] | 1,422 |
c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
https://github.com/Theomat/colorization-av-enseirb-2020/tree/c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
TransposedConv3d
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/x4/cx4iqpwvwk2udyozikfqok25vqpb2at5quhk54guhagusagsgecd.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x_1 => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%squeeze,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 128)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 3, 3), (108, 27, 9, 3, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1), 0), primals_1, stride=(2, 1, 1), padding=(1, 1, 1), dilation=(1, 1, 1), transposed=True, output_padding=(1, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (1, 4, 8, 4, 4), (512, 128, 16, 4, 1))
buf1 = reinterpret_tensor(buf0, (4, 8, 4, 4), (128, 16, 4, 1), 0); del buf0 # reuse
buf2 = empty_strided_cuda((4, 8, 4, 4), (128, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf2, 512, grid=grid(512), stream=stream0)
del primals_2
return (buf1, primals_1, reinterpret_tensor(primals_3, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1), 0), buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 3, 3, 3), (108, 27, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
class TransposedConv3d(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=(3, 3, 3),
stride=(2, 1, 1), padding=(1, 1, 1), output_padding=(1, 0, 0),
activation_fn=F.relu, use_batch_norm=False, use_bias=True):
super(TransposedConv3d, self).__init__()
self._use_batch_norm = use_batch_norm
self._activation_fn = activation_fn
self.transposed_conv3d = nn.ConvTranspose3d(in_channels,
output_channels, kernel_shape, stride, padding=padding,
output_padding=output_padding, bias=use_bias)
if self._use_batch_norm:
self.bn = nn.BatchNorm3d(self._output_channels, eps=0.001,
momentum=0.01)
def forward(self, x):
x = self.transposed_conv3d(x)
if self._use_batch_norm:
x = self.bn(x)
if self._activation_fn is not None:
x = self._activation_fn(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'output_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 128
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 3, 3), (108, 27, 9, 3, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1,
4, 4, 4, 4), (256, 64, 16, 4, 1), 0), primals_1, stride=(2, 1,
1), padding=(1, 1, 1), dilation=(1, 1, 1), transposed=True,
output_padding=(1, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (1, 4, 8, 4, 4), (512, 128, 16, 4, 1))
buf1 = reinterpret_tensor(buf0, (4, 8, 4, 4), (128, 16, 4, 1), 0)
del buf0
buf2 = empty_strided_cuda((4, 8, 4, 4), (128, 16, 4, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(512)](buf1,
primals_2, buf2, 512, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
return buf1, primals_1, reinterpret_tensor(primals_3, (1, 4, 4, 4, 4),
(256, 64, 16, 4, 1), 0), buf2
class TransposedConv3dNew(nn.Module):
def __init__(self, in_channels, output_channels, kernel_shape=(3, 3, 3),
stride=(2, 1, 1), padding=(1, 1, 1), output_padding=(1, 0, 0),
activation_fn=F.relu, use_batch_norm=False, use_bias=True):
super(TransposedConv3dNew, self).__init__()
self._use_batch_norm = use_batch_norm
self._activation_fn = activation_fn
self.transposed_conv3d = nn.ConvTranspose3d(in_channels,
output_channels, kernel_shape, stride, padding=padding,
output_padding=output_padding, bias=use_bias)
if self._use_batch_norm:
self.bn = nn.BatchNorm3d(self._output_channels, eps=0.001,
momentum=0.01)
def forward(self, input_0):
primals_1 = self.transposed_conv3d.weight
primals_2 = self.transposed_conv3d.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
TencentYoutuResearch/ActionDetection-AFSD
|
TransposedConv3d
| false | 14,471 |
[
"BSD-3-Clause"
] | 112 |
ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
https://github.com/TencentYoutuResearch/ActionDetection-AFSD/tree/ed86a0df91e58baa7d78c796ed29cff82b1f3fa6
|
SM
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/fi/cfibns5ntmtk4tzcie36w465rywyy46ysnh74xm6s7lnxi2xwts5.py
# Topologically Sorted Source Nodes: [avg_pool2d], Original ATen: [aten.avg_pool2d]
# Source node to ATen node mapping:
# avg_pool2d => avg_pool2d
# Graph fragment:
# %avg_pool2d : [num_users=1] = call_function[target=torch.ops.aten.avg_pool2d.default](args = (%arg0_1, [3, 3], [1, 1], [1, 1]), kwargs = {})
triton_poi_fused_avg_pool2d_0 = async_compile.triton('triton_poi_fused_avg_pool2d_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_avg_pool2d_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 9, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_avg_pool2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x4 = xindex
tmp0 = (-1) + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = (-1) + x0
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + ((-5) + x4), tmp10 & xmask, other=0.0)
tmp12 = x0
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + ((-4) + x4), tmp16 & xmask, other=0.0)
tmp18 = tmp17 + tmp11
tmp19 = 1 + x0
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + ((-3) + x4), tmp23 & xmask, other=0.0)
tmp25 = tmp24 + tmp18
tmp26 = x1
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + ((-1) + x4), tmp30 & xmask, other=0.0)
tmp32 = tmp31 + tmp25
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + (x4), tmp33 & xmask, other=0.0)
tmp35 = tmp34 + tmp32
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (1 + x4), tmp36 & xmask, other=0.0)
tmp38 = tmp37 + tmp35
tmp39 = 1 + x1
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (3 + x4), tmp43 & xmask, other=0.0)
tmp45 = tmp44 + tmp38
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (4 + x4), tmp46 & xmask, other=0.0)
tmp48 = tmp47 + tmp45
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (5 + x4), tmp49 & xmask, other=0.0)
tmp51 = tmp50 + tmp48
tmp52 = 1 + ((-1)*x0) + ((-1)*x1) + (x0*x1) + (((5) * ((5) <= (2 + x0)) + (2 + x0) * ((2 + x0) < (5)))*((5) * ((5) <= (2 + x1)) + (2 + x1) * ((2 + x1) < (5)))) + ((-1)*x0*((5) * ((5) <= (2 + x1)) + (2 + x1) * ((2 + x1) < (5)))) + ((-1)*x1*((5) * ((5) <= (2 + x0)) + (2 + x0) * ((2 + x0) < (5)))) + ((5) * ((5) <= (2 + x0)) + (2 + x0) * ((2 + x0) < (5))) + ((5) * ((5) <= (2 + x1)) + (2 + x1) * ((2 + x1) < (5)))
tmp53 = tmp51 / tmp52
tl.store(out_ptr0 + (x4), tmp53, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/yi/cyieqphiurllwq74ivk4vr5d2is3exwqasgwi3da5qx3v3f7y6ey.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# x => getitem
# Graph fragment:
# %getitem : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 0), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_1 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 9, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x4 = xindex
tmp0 = (-1) + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = (-1) + x0
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + ((-5) + x4), tmp10 & xmask, other=float("-inf"))
tmp12 = x0
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + ((-4) + x4), tmp16 & xmask, other=float("-inf"))
tmp18 = triton_helpers.maximum(tmp17, tmp11)
tmp19 = 1 + x0
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + ((-3) + x4), tmp23 & xmask, other=float("-inf"))
tmp25 = triton_helpers.maximum(tmp24, tmp18)
tmp26 = x1
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + ((-1) + x4), tmp30 & xmask, other=float("-inf"))
tmp32 = triton_helpers.maximum(tmp31, tmp25)
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + (x4), tmp33 & xmask, other=float("-inf"))
tmp35 = triton_helpers.maximum(tmp34, tmp32)
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (1 + x4), tmp36 & xmask, other=float("-inf"))
tmp38 = triton_helpers.maximum(tmp37, tmp35)
tmp39 = 1 + x1
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (3 + x4), tmp43 & xmask, other=float("-inf"))
tmp45 = triton_helpers.maximum(tmp44, tmp38)
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (4 + x4), tmp46 & xmask, other=float("-inf"))
tmp48 = triton_helpers.maximum(tmp47, tmp45)
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (5 + x4), tmp49 & xmask, other=float("-inf"))
tmp51 = triton_helpers.maximum(tmp50, tmp48)
tl.store(out_ptr0 + (x4), tmp51, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [avg_pool2d], Original ATen: [aten.avg_pool2d]
stream0 = get_raw_stream(0)
triton_poi_fused_avg_pool2d_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_1.run(buf0, buf1, 256, grid=grid(256), stream=stream0)
del buf0
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
def autopad(k, p=None):
if p is None:
p = k // 2 if isinstance(k, int) else [(x // 2) for x in k]
return p
class SM(nn.Module):
def __init__(self, k=3, s=1):
super(SM, self).__init__()
self.avg = nn.AvgPool2d(k, stride=s, padding=autopad(k))
self.max = nn.MaxPool2d(k, stride=s, padding=autopad(k))
def forward(self, x):
x = self.max(self.avg(x))
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_avg_pool2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x4 = xindex
tmp0 = -1 + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = -1 + x0
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + (-5 + x4), tmp10 & xmask, other=0.0)
tmp12 = x0
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + (-4 + x4), tmp16 & xmask, other=0.0)
tmp18 = tmp17 + tmp11
tmp19 = 1 + x0
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + (-3 + x4), tmp23 & xmask, other=0.0)
tmp25 = tmp24 + tmp18
tmp26 = x1
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + (-1 + x4), tmp30 & xmask, other=0.0)
tmp32 = tmp31 + tmp25
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + x4, tmp33 & xmask, other=0.0)
tmp35 = tmp34 + tmp32
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (1 + x4), tmp36 & xmask, other=0.0)
tmp38 = tmp37 + tmp35
tmp39 = 1 + x1
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (3 + x4), tmp43 & xmask, other=0.0)
tmp45 = tmp44 + tmp38
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (4 + x4), tmp46 & xmask, other=0.0)
tmp48 = tmp47 + tmp45
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (5 + x4), tmp49 & xmask, other=0.0)
tmp51 = tmp50 + tmp48
tmp52 = 1 + -1 * x0 + -1 * x1 + x0 * x1 + (5 * (5 <= 2 + x0) + (2 + x0) *
(2 + x0 < 5)) * (5 * (5 <= 2 + x1) + (2 + x1) * (2 + x1 < 5)
) + -1 * x0 * (5 * (5 <= 2 + x1) + (2 + x1) * (2 + x1 < 5)
) + -1 * x1 * (5 * (5 <= 2 + x0) + (2 + x0) * (2 + x0 < 5)) + (5 *
(5 <= 2 + x0) + (2 + x0) * (2 + x0 < 5)) + (5 * (5 <= 2 + x1) + (2 +
x1) * (2 + x1 < 5))
tmp53 = tmp51 / tmp52
tl.store(out_ptr0 + x4, tmp53, xmask)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_1(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x4 = xindex
tmp0 = -1 + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = -1 + x0
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + (-5 + x4), tmp10 & xmask, other=float('-inf'))
tmp12 = x0
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + (-4 + x4), tmp16 & xmask, other=float('-inf'))
tmp18 = triton_helpers.maximum(tmp17, tmp11)
tmp19 = 1 + x0
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + (-3 + x4), tmp23 & xmask, other=float('-inf'))
tmp25 = triton_helpers.maximum(tmp24, tmp18)
tmp26 = x1
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + (-1 + x4), tmp30 & xmask, other=float('-inf'))
tmp32 = triton_helpers.maximum(tmp31, tmp25)
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + x4, tmp33 & xmask, other=float('-inf'))
tmp35 = triton_helpers.maximum(tmp34, tmp32)
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (1 + x4), tmp36 & xmask, other=float('-inf'))
tmp38 = triton_helpers.maximum(tmp37, tmp35)
tmp39 = 1 + x1
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (3 + x4), tmp43 & xmask, other=float('-inf'))
tmp45 = triton_helpers.maximum(tmp44, tmp38)
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (4 + x4), tmp46 & xmask, other=float('-inf'))
tmp48 = triton_helpers.maximum(tmp47, tmp45)
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (5 + x4), tmp49 & xmask, other=float('-inf'))
tmp51 = triton_helpers.maximum(tmp50, tmp48)
tl.store(out_ptr0 + x4, tmp51, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_avg_pool2d_0[grid(256)](arg0_1, buf0, 256, XBLOCK=
256, num_warps=4, num_stages=1)
del arg0_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_max_pool2d_with_indices_1[grid(256)](buf0, buf1,
256, XBLOCK=256, num_warps=4, num_stages=1)
del buf0
return buf1,
def autopad(k, p=None):
if p is None:
p = k // 2 if isinstance(k, int) else [(x // 2) for x in k]
return p
class SMNew(nn.Module):
def __init__(self, k=3, s=1):
super(SMNew, self).__init__()
self.avg = nn.AvgPool2d(k, stride=s, padding=autopad(k))
self.max = nn.MaxPool2d(k, stride=s, padding=autopad(k))
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
TarikToha/NWPU-Crowd-Sample-Code-for-Localization
|
SM
| false | 14,472 |
[
"MIT"
] | 132 |
0e348b99ea41d4469eff2a78a75648454128d49a
|
https://github.com/TarikToha/NWPU-Crowd-Sample-Code-for-Localization/tree/0e348b99ea41d4469eff2a78a75648454128d49a
|
SpatialConv3D
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/v5/cv5je5prl3u6emvyoq3ehrsgzxvu5pa42gg4ztn3gzwqegf5w23d.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
# Source node to ATen node mapping:
# x_1 => gt, mul, where
# Graph fragment:
# %gt : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze, 0.01), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %squeeze, %mul), kwargs = {})
# %gt_3 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_3, 0), kwargs = {})
triton_poi_fused_leaky_relu_leaky_relu_backward_0 = async_compile.triton('triton_poi_fused_leaky_relu_leaky_relu_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_leaky_relu_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 16)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + (x2), tmp7, xmask)
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/my/cmytvt5pqlkrxcw247g5zhypkpvz2xt6lh2gd4ahqadihmdwxcsx.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
# Source node to ATen node mapping:
# x_3 => gt_1, mul_1, where_1
# Graph fragment:
# %gt_1 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_4, 0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_4, 0.01), kwargs = {})
# %where_1 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %squeeze_4, %mul_1), kwargs = {})
# %gt_2 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_9, 0), kwargs = {})
triton_poi_fused_leaky_relu_leaky_relu_backward_1 = async_compile.triton('triton_poi_fused_leaky_relu_leaky_relu_backward_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_leaky_relu_backward_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_1(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + (x2), tmp7, xmask)
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (16, 4, 1, 3, 3), (36, 9, 9, 3, 1))
assert_size_stride(primals_2, (16, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 16, 1, 3, 3), (144, 9, 9, 3, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1), 0), primals_1, stride=(1, 2, 2), padding=(0, 1, 1), dilation=(1, 1, 1), transposed=False, output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (1, 16, 4, 2, 2), (256, 16, 4, 2, 1))
buf1 = reinterpret_tensor(buf0, (16, 4, 2, 2), (16, 4, 2, 1), 0); del buf0 # reuse
buf5 = empty_strided_cuda((16, 4, 2, 2), (16, 4, 2, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_leaky_relu_backward_0.run(buf1, primals_2, buf5, 256, grid=grid(256), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(buf1, (1, 16, 4, 2, 2), (0, 16, 4, 2, 1), 0), primals_4, stride=(1, 2, 2), padding=(0, 1, 1), dilation=(1, 1, 1), transposed=False, output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf2, (1, 4, 4, 1, 1), (16, 4, 1, 1, 1))
buf3 = reinterpret_tensor(buf2, (4, 4, 1, 1), (4, 1, 16, 16), 0); del buf2 # reuse
buf4 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
triton_poi_fused_leaky_relu_leaky_relu_backward_1.run(buf3, primals_5, buf4, 16, grid=grid(16), stream=stream0)
del primals_5
return (reinterpret_tensor(buf3, (4, 4, 1, 1), (4, 1, 1, 1), 0), primals_1, primals_4, reinterpret_tensor(primals_3, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1), 0), reinterpret_tensor(buf1, (1, 16, 4, 2, 2), (256, 16, 4, 2, 1), 0), buf4, buf5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((16, 4, 1, 3, 3), (36, 9, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((16, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 16, 1, 3, 3), (144, 9, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class SpatialConv3D(nn.Module):
"""
Apply 3D conv. over an input signal composed of several input planes with distinct spatial and time axes, by performing 3D convolution over the spatiotemporal axes
rrgs:
in_channels (int): number of channels in the input tensor
out_channels (int): number of channels produced by the convolution
kernel_size (int or tuple): size of the convolution kernel
stride (int or tuple): stride
padding (int or tuple): zero-padding
"""
def __init__(self, in_channels, out_channels, kernel_size=(1, 3, 3),
stride=(1, 2, 2), padding=(0, 1, 1)):
super(SpatialConv3D, self).__init__()
self.conv1 = nn.Conv3d(in_channels, 16, kernel_size, stride, padding)
self.reLu1 = nn.LeakyReLU(inplace=True)
self.conv2 = nn.Conv3d(16, out_channels, kernel_size, stride, padding)
self.reLu2 = nn.LeakyReLU(inplace=True)
def forward(self, x):
x = self.conv1(x)
x = self.reLu1(x)
x = self.conv2(x)
x = self.reLu2(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 16
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + x2, tmp7, xmask)
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_1(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + x2, tmp7, xmask)
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (16, 4, 1, 3, 3), (36, 9, 9, 3, 1))
assert_size_stride(primals_2, (16,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 16, 1, 3, 3), (144, 9, 9, 3, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1,
4, 4, 4, 4), (256, 64, 16, 4, 1), 0), primals_1, stride=(1, 2,
2), padding=(0, 1, 1), dilation=(1, 1, 1), transposed=False,
output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (1, 16, 4, 2, 2), (256, 16, 4, 2, 1))
buf1 = reinterpret_tensor(buf0, (16, 4, 2, 2), (16, 4, 2, 1), 0)
del buf0
buf5 = empty_strided_cuda((16, 4, 2, 2), (16, 4, 2, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_leaky_relu_leaky_relu_backward_0[grid(256)](buf1,
primals_2, buf5, 256, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(reinterpret_tensor(buf1, (1, 16,
4, 2, 2), (0, 16, 4, 2, 1), 0), primals_4, stride=(1, 2, 2),
padding=(0, 1, 1), dilation=(1, 1, 1), transposed=False,
output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf2, (1, 4, 4, 1, 1), (16, 4, 1, 1, 1))
buf3 = reinterpret_tensor(buf2, (4, 4, 1, 1), (4, 1, 16, 16), 0)
del buf2
buf4 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.bool)
triton_poi_fused_leaky_relu_leaky_relu_backward_1[grid(16)](buf3,
primals_5, buf4, 16, XBLOCK=16, num_warps=1, num_stages=1)
del primals_5
return reinterpret_tensor(buf3, (4, 4, 1, 1), (4, 1, 1, 1), 0
), primals_1, primals_4, reinterpret_tensor(primals_3, (1, 4, 4, 4,
4), (256, 64, 16, 4, 1), 0), reinterpret_tensor(buf1, (1, 16, 4, 2,
2), (256, 16, 4, 2, 1), 0), buf4, buf5
class SpatialConv3DNew(nn.Module):
"""
Apply 3D conv. over an input signal composed of several input planes with distinct spatial and time axes, by performing 3D convolution over the spatiotemporal axes
rrgs:
in_channels (int): number of channels in the input tensor
out_channels (int): number of channels produced by the convolution
kernel_size (int or tuple): size of the convolution kernel
stride (int or tuple): stride
padding (int or tuple): zero-padding
"""
def __init__(self, in_channels, out_channels, kernel_size=(1, 3, 3),
stride=(1, 2, 2), padding=(0, 1, 1)):
super(SpatialConv3DNew, self).__init__()
self.conv1 = nn.Conv3d(in_channels, 16, kernel_size, stride, padding)
self.reLu1 = nn.LeakyReLU(inplace=True)
self.conv2 = nn.Conv3d(16, out_channels, kernel_size, stride, padding)
self.reLu2 = nn.LeakyReLU(inplace=True)
def forward(self, input_0):
primals_1 = self.conv1.weight
primals_2 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
Tencent/DVQA
|
SpatialConv3D
| false | 14,473 |
[
"BSD-3-Clause"
] | 408 |
21727333a6b41d54ad1a8beca1fcbe00a69ed347
|
https://github.com/Tencent/DVQA/tree/21727333a6b41d54ad1a8beca1fcbe00a69ed347
|
ModulatedConv2d
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/wi/cwiyl3lwwtancorrifw77xt3aqb4lermdintht45zvkj3bg54nbl.py
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_3, 0.5), kwargs = {})
triton_poi_fused_mul_0 = async_compile.triton('triton_poi_fused_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/2o/c2oqkq7zaubqmw7vuixxlseb2ff5jzqqbyczicxlmsahuxwdpdyp.py
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul_1 => mul_1
# Graph fragment:
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_4, 1), kwargs = {})
triton_poi_fused_mul_1 = async_compile.triton('triton_poi_fused_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ri/criuvsdl3sferb4bb6ci5zaps3wys7xxcpybz7vfo2ba4q7cuq6c.py
# Topologically Sorted Source Nodes: [mul_2, weight, pow_1, sum_1, add, demod, weight_1], Original ATen: [aten.mul, aten.pow, aten.sum, aten.add, aten.rsqrt]
# Source node to ATen node mapping:
# add => add
# demod => rsqrt
# mul_2 => mul_2
# pow_1 => pow_1
# sum_1 => sum_1
# weight => mul_3
# weight_1 => mul_4
# Graph fragment:
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_5, 0.125), kwargs = {})
# %mul_3 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, %view), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%mul_3, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [2, 3, 4]), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_1, 1e-08), kwargs = {})
# %rsqrt : [num_users=2] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_3, %view_1), kwargs = {})
triton_per_fused_add_mul_pow_rsqrt_sum_2 = async_compile.triton('triton_per_fused_add_mul_pow_rsqrt_sum_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[16, 64],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_mul_pow_rsqrt_sum_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_mul_pow_rsqrt_sum_2(in_out_ptr0, in_ptr0, in_ptr1, out_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 16
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r5 = rindex
x0 = xindex % 4
r3 = (rindex // 16)
x1 = (xindex // 4)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (r5 + (64*x0)), xmask, eviction_policy='evict_last', other=0.0)
tmp3 = tl.load(in_ptr1 + (r3 + (4*x1)), xmask, eviction_policy='evict_last', other=0.0)
tmp1 = 0.125
tmp2 = tmp0 * tmp1
tmp4 = tmp2 * tmp3
tmp5 = tmp4 * tmp4
tmp6 = tl.broadcast_to(tmp5, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = 1e-08
tmp11 = tmp9 + tmp10
tmp12 = libdevice.rsqrt(tmp11)
tmp13 = tmp4 * tmp12
tl.debug_barrier()
tl.store(in_out_ptr0 + (x4), tmp12, xmask)
tl.store(out_ptr0 + (r5 + (64*x4)), tmp13, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, ), (1, ))
assert_size_stride(primals_5, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_0.run(primals_3, buf0, 16, grid=grid(16), stream=stream0)
del primals_3
buf1 = empty_strided_cuda((4, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
triton_poi_fused_mul_1.run(primals_4, buf1, 4, grid=grid(4), stream=stream0)
del primals_4
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul_1, out], Original ATen: [aten.mul, aten.addmm]
extern_kernels.addmm(buf1, primals_2, reinterpret_tensor(buf0, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del buf1
buf3 = buf0; del buf0 # reuse
buf4 = buf3; del buf3 # reuse
buf5 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul_2, weight, pow_1, sum_1, add, demod, weight_1], Original ATen: [aten.mul, aten.pow, aten.sum, aten.add, aten.rsqrt]
triton_per_fused_add_mul_pow_rsqrt_sum_2.run(buf4, primals_5, buf2, buf5, 16, 64, grid=grid(16), stream=stream0)
# Topologically Sorted Source Nodes: [out_1], Original ATen: [aten.convolution]
buf6 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1, 16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf5, (16, 4, 4, 4), (64, 16, 4, 1), 0), stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=4, bias=None)
assert_size_stride(buf6, (1, 16, 5, 5), (400, 25, 5, 1))
return (reinterpret_tensor(buf6, (4, 4, 5, 5), (100, 25, 5, 1), 0), primals_2, primals_5, buf2, buf4, reinterpret_tensor(buf5, (16, 4, 4, 4), (64, 16, 4, 1), 0), reinterpret_tensor(primals_1, (1, 16, 4, 4), (256, 16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((1, 4, 4, 4, 4), (256, 64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
import torch.utils.data
import torch
import torch.nn as nn
import torch.nn.functional as F
def make_kernel(k):
k = torch.tensor(k, dtype=torch.float32)
if len(k.shape) == 1:
k = k[None, :] * k[:, None]
k /= k.sum()
return k
def upfirdn2d_native(input, kernel, up_x, up_y, down_x, down_y, pad_x0,
pad_x1, pad_y0, pad_y1):
_, minor, in_h, in_w = input.shape
kernel_h, kernel_w = kernel.shape
out = input.view(-1, minor, in_h, 1, in_w, 1)
out = F.pad(out, [0, up_x - 1, 0, 0, 0, up_y - 1, 0, 0])
out = out.view(-1, minor, in_h * up_y, in_w * up_x)
out = F.pad(out, [max(pad_x0, 0), max(pad_x1, 0), max(pad_y0, 0), max(
pad_y1, 0)])
out = out[:, :, max(-pad_y0, 0):out.shape[2] - max(-pad_y1, 0), max(-
pad_x0, 0):out.shape[3] - max(-pad_x1, 0)]
out = out.reshape([-1, 1, in_h * up_y + pad_y0 + pad_y1, in_w * up_x +
pad_x0 + pad_x1])
w = torch.flip(kernel, [0, 1]).view(1, 1, kernel_h, kernel_w)
out = F.conv2d(out, w)
out = out.reshape(-1, minor, in_h * up_y + pad_y0 + pad_y1 - kernel_h +
1, in_w * up_x + pad_x0 + pad_x1 - kernel_w + 1)
return out[:, :, ::down_y, ::down_x]
def upfirdn2d(input, kernel, up=1, down=1, pad=(0, 0)):
return upfirdn2d_native(input, kernel, up, up, down, down, pad[0], pad[
1], pad[0], pad[1])
def fused_leaky_relu(input, bias, negative_slope=0.2, scale=2 ** 0.5):
return F.leaky_relu(input + bias, negative_slope) * scale
class Blur(nn.Module):
def __init__(self, kernel, pad, upsample_factor=1):
super().__init__()
kernel = make_kernel(kernel)
if upsample_factor > 1:
kernel = kernel * upsample_factor ** 2
self.register_buffer('kernel', kernel)
self.pad = pad
def forward(self, input):
out = upfirdn2d(input, self.kernel, pad=self.pad)
return out
class EqualLinear(nn.Module):
def __init__(self, in_dim, out_dim, bias=True, bias_init=0, lr_mul=1,
activation=None):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_dim, in_dim).div_(lr_mul))
if bias:
self.bias = nn.Parameter(torch.zeros(out_dim).fill_(bias_init))
else:
self.bias = None
self.activation = activation
self.scale = math.sqrt(1) / math.sqrt(in_dim) * lr_mul
self.lr_mul = lr_mul
def forward(self, input):
if self.activation:
out = F.linear(input, self.weight * self.scale)
out = fused_leaky_relu(out, self.bias * self.lr_mul)
else:
out = F.linear(input, self.weight * self.scale, bias=self.bias *
self.lr_mul)
return out
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]})'
)
class ModulatedConv2d(nn.Module):
def __init__(self, in_channel, out_channel, kernel_size, style_dim,
demodulate=True, upsample=False, downsample=False, blur_kernel=[1,
3, 3, 1]):
super().__init__()
self.eps = 1e-08
self.kernel_size = kernel_size
self.in_channel = in_channel
self.out_channel = out_channel
self.upsample = upsample
self.downsample = downsample
if upsample:
factor = 2
p = len(blur_kernel) - factor - (kernel_size - 1)
pad0 = (p + 1) // 2 + factor - 1
pad1 = p // 2 + 1
self.blur = Blur(blur_kernel, pad=(pad0, pad1), upsample_factor
=factor)
if downsample:
factor = 2
p = len(blur_kernel) - factor + (kernel_size - 1)
pad0 = (p + 1) // 2
pad1 = p // 2
self.blur = Blur(blur_kernel, pad=(pad0, pad1))
fan_in = in_channel * kernel_size ** 2
self.scale = math.sqrt(1) / math.sqrt(fan_in)
self.padding = kernel_size // 2
self.weight = nn.Parameter(torch.randn(1, out_channel, in_channel,
kernel_size, kernel_size))
if style_dim is not None and style_dim > 0:
self.modulation = EqualLinear(style_dim, in_channel, bias_init=1)
self.demodulate = demodulate
def __repr__(self):
return (
f'{self.__class__.__name__}({self.in_channel}, {self.out_channel}, {self.kernel_size}, upsample={self.upsample}, downsample={self.downsample})'
)
def forward(self, input, style):
batch, in_channel, height, width = input.shape
if style is not None:
style = self.modulation(style).view(batch, 1, in_channel, 1, 1)
else:
style = torch.ones(batch, 1, in_channel, 1, 1)
weight = self.scale * self.weight * style
if self.demodulate:
demod = torch.rsqrt(weight.pow(2).sum([2, 3, 4]) + 1e-08)
weight = weight * demod.view(batch, self.out_channel, 1, 1, 1)
weight = weight.view(batch * self.out_channel, in_channel, self.
kernel_size, self.kernel_size)
if self.upsample:
input = input.view(1, batch * in_channel, height, width)
weight = weight.view(batch, self.out_channel, in_channel, self.
kernel_size, self.kernel_size)
weight = weight.transpose(1, 2).reshape(batch * in_channel,
self.out_channel, self.kernel_size, self.kernel_size)
out = F.conv_transpose2d(input, weight, padding=0, stride=2,
groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
out = self.blur(out)
elif self.downsample:
input = self.blur(input)
_, _, height, width = input.shape
input = input.view(1, batch * in_channel, height, width)
out = F.conv2d(input, weight, padding=0, stride=2, groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
else:
input = input.view(1, batch * in_channel, height, width)
out = F.conv2d(input, weight, padding=self.padding, groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'in_channel': 4, 'out_channel': 4, 'kernel_size': 4,
'style_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import math
import torch.utils.data
import torch
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_per_fused_add_mul_pow_rsqrt_sum_2(in_out_ptr0, in_ptr0, in_ptr1,
out_ptr0, xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 16
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r5 = rindex
x0 = xindex % 4
r3 = rindex // 16
x1 = xindex // 4
x4 = xindex
tmp0 = tl.load(in_ptr0 + (r5 + 64 * x0), xmask, eviction_policy=
'evict_last', other=0.0)
tmp3 = tl.load(in_ptr1 + (r3 + 4 * x1), xmask, eviction_policy=
'evict_last', other=0.0)
tmp1 = 0.125
tmp2 = tmp0 * tmp1
tmp4 = tmp2 * tmp3
tmp5 = tmp4 * tmp4
tmp6 = tl.broadcast_to(tmp5, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = 1e-08
tmp11 = tmp9 + tmp10
tmp12 = libdevice.rsqrt(tmp11)
tmp13 = tmp4 * tmp12
tl.debug_barrier()
tl.store(in_out_ptr0 + x4, tmp12, xmask)
tl.store(out_ptr0 + (r5 + 64 * x4), tmp13, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4,), (1,))
assert_size_stride(primals_5, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_0[grid(16)](primals_3, buf0, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del primals_3
buf1 = empty_strided_cuda((4,), (1,), torch.float32)
triton_poi_fused_mul_1[grid(4)](primals_4, buf1, 4, XBLOCK=4,
num_warps=1, num_stages=1)
del primals_4
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(buf1, primals_2, reinterpret_tensor(buf0, (4,
4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del buf1
buf3 = buf0
del buf0
buf4 = buf3
del buf3
buf5 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.float32)
triton_per_fused_add_mul_pow_rsqrt_sum_2[grid(16)](buf4, primals_5,
buf2, buf5, 16, 64, XBLOCK=8, num_warps=4, num_stages=1)
buf6 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1,
16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf5, (16, 4,
4, 4), (64, 16, 4, 1), 0), stride=(1, 1), padding=(2, 2),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=4, bias=None)
assert_size_stride(buf6, (1, 16, 5, 5), (400, 25, 5, 1))
return reinterpret_tensor(buf6, (4, 4, 5, 5), (100, 25, 5, 1), 0
), primals_2, primals_5, buf2, buf4, reinterpret_tensor(buf5, (16,
4, 4, 4), (64, 16, 4, 1), 0), reinterpret_tensor(primals_1, (1, 16,
4, 4), (256, 16, 4, 1), 0)
def make_kernel(k):
k = torch.tensor(k, dtype=torch.float32)
if len(k.shape) == 1:
k = k[None, :] * k[:, None]
k /= k.sum()
return k
def upfirdn2d_native(input, kernel, up_x, up_y, down_x, down_y, pad_x0,
pad_x1, pad_y0, pad_y1):
_, minor, in_h, in_w = input.shape
kernel_h, kernel_w = kernel.shape
out = input.view(-1, minor, in_h, 1, in_w, 1)
out = F.pad(out, [0, up_x - 1, 0, 0, 0, up_y - 1, 0, 0])
out = out.view(-1, minor, in_h * up_y, in_w * up_x)
out = F.pad(out, [max(pad_x0, 0), max(pad_x1, 0), max(pad_y0, 0), max(
pad_y1, 0)])
out = out[:, :, max(-pad_y0, 0):out.shape[2] - max(-pad_y1, 0), max(-
pad_x0, 0):out.shape[3] - max(-pad_x1, 0)]
out = out.reshape([-1, 1, in_h * up_y + pad_y0 + pad_y1, in_w * up_x +
pad_x0 + pad_x1])
w = torch.flip(kernel, [0, 1]).view(1, 1, kernel_h, kernel_w)
out = F.conv2d(out, w)
out = out.reshape(-1, minor, in_h * up_y + pad_y0 + pad_y1 - kernel_h +
1, in_w * up_x + pad_x0 + pad_x1 - kernel_w + 1)
return out[:, :, ::down_y, ::down_x]
def upfirdn2d(input, kernel, up=1, down=1, pad=(0, 0)):
return upfirdn2d_native(input, kernel, up, up, down, down, pad[0], pad[
1], pad[0], pad[1])
def fused_leaky_relu(input, bias, negative_slope=0.2, scale=2 ** 0.5):
return F.leaky_relu(input + bias, negative_slope) * scale
class Blur(nn.Module):
def __init__(self, kernel, pad, upsample_factor=1):
super().__init__()
kernel = make_kernel(kernel)
if upsample_factor > 1:
kernel = kernel * upsample_factor ** 2
self.register_buffer('kernel', kernel)
self.pad = pad
def forward(self, input):
out = upfirdn2d(input, self.kernel, pad=self.pad)
return out
class EqualLinear(nn.Module):
def __init__(self, in_dim, out_dim, bias=True, bias_init=0, lr_mul=1,
activation=None):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_dim, in_dim).div_(lr_mul))
if bias:
self.bias = nn.Parameter(torch.zeros(out_dim).fill_(bias_init))
else:
self.bias = None
self.activation = activation
self.scale = math.sqrt(1) / math.sqrt(in_dim) * lr_mul
self.lr_mul = lr_mul
def forward(self, input):
if self.activation:
out = F.linear(input, self.weight * self.scale)
out = fused_leaky_relu(out, self.bias * self.lr_mul)
else:
out = F.linear(input, self.weight * self.scale, bias=self.bias *
self.lr_mul)
return out
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]})'
)
class ModulatedConv2dNew(nn.Module):
def __init__(self, in_channel, out_channel, kernel_size, style_dim,
demodulate=True, upsample=False, downsample=False, blur_kernel=[1,
3, 3, 1]):
super().__init__()
self.eps = 1e-08
self.kernel_size = kernel_size
self.in_channel = in_channel
self.out_channel = out_channel
self.upsample = upsample
self.downsample = downsample
if upsample:
factor = 2
p = len(blur_kernel) - factor - (kernel_size - 1)
pad0 = (p + 1) // 2 + factor - 1
pad1 = p // 2 + 1
self.blur = Blur(blur_kernel, pad=(pad0, pad1), upsample_factor
=factor)
if downsample:
factor = 2
p = len(blur_kernel) - factor + (kernel_size - 1)
pad0 = (p + 1) // 2
pad1 = p // 2
self.blur = Blur(blur_kernel, pad=(pad0, pad1))
fan_in = in_channel * kernel_size ** 2
self.scale = math.sqrt(1) / math.sqrt(fan_in)
self.padding = kernel_size // 2
self.weight = nn.Parameter(torch.randn(1, out_channel, in_channel,
kernel_size, kernel_size))
if style_dim is not None and style_dim > 0:
self.modulation = EqualLinear(style_dim, in_channel, bias_init=1)
self.demodulate = demodulate
def __repr__(self):
return (
f'{self.__class__.__name__}({self.in_channel}, {self.out_channel}, {self.kernel_size}, upsample={self.upsample}, downsample={self.downsample})'
)
def forward(self, input_0, input_1):
primals_5 = self.weight
primals_2 = self.modulation.weight
primals_4 = self.modulation.bias
primals_1 = input_0
primals_3 = input_1
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
Theomat/colorization-av-enseirb-2020
|
ModulatedConv2d
| false | 14,474 |
[
"Apache-2.0"
] | 1,422 |
c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
https://github.com/Theomat/colorization-av-enseirb-2020/tree/c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
TokenEmbedding
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/wf/cwfgpodivarq2gzz7nodlok35jybiut5djlrlgyaw23yzzih2tt7.py
# Topologically Sorted Source Nodes: [pad], Original ATen: [aten.copy]
# Source node to ATen node mapping:
# pad => copy
# Graph fragment:
# %copy : [num_users=1] = call_function[target=torch.ops.aten.copy.default](args = (%slice_1, %slice_2), kwargs = {})
# %slice_scatter_default : [num_users=2] = call_function[target=torch.ops.aten.slice_scatter.default](args = (%empty, %copy, 2, 1, 5), kwargs = {})
# %slice_scatter_default_1 : [num_users=2] = call_function[target=torch.ops.aten.slice_scatter.default](args = (%slice_scatter_default, %slice_7, 2, 0, 1), kwargs = {})
# %slice_scatter_default_2 : [num_users=2] = call_function[target=torch.ops.aten.slice_scatter.default](args = (%slice_scatter_default_1, %slice_12, 2, 5, 6), kwargs = {})
triton_poi_fused_copy_0 = async_compile.triton('triton_poi_fused_copy_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_copy_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_copy_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 24
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
y0 = yindex % 6
x2 = xindex
y1 = (yindex // 6)
tmp0 = y0
tmp1 = tl.full([1, 1], 5, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.broadcast_to((-4) + y0, [XBLOCK, YBLOCK])
tmp4 = tl.full([1, 1], 1, tl.int64)
tmp5 = tmp3 < tmp4
tmp6 = tmp5 & tmp2
tmp7 = tl.broadcast_to(y0, [XBLOCK, YBLOCK])
tmp8 = tmp7 >= tmp4
tmp9 = tmp7 < tmp1
tmp10 = tmp8 & tmp9
tmp11 = tmp10 & tmp6
tmp12 = tl.load(in_ptr0 + ((-4) + x2 + (4*y0) + (16*y1)), tmp11 & xmask & ymask, eviction_policy='evict_last', other=0.0)
tmp13 = float("nan")
tmp14 = tl.where(tmp10, tmp12, tmp13)
tmp15 = tl.full(tmp14.shape, 0.0, tmp14.dtype)
tmp16 = tl.where(tmp6, tmp14, tmp15)
tmp17 = tmp3 >= tmp4
tmp18 = tmp3 < tmp1
tmp19 = tmp17 & tmp18
tmp20 = tmp19 & tmp2
tmp21 = tl.load(in_ptr0 + ((-20) + x2 + (4*y0) + (16*y1)), tmp20 & xmask & ymask, eviction_policy='evict_last', other=0.0)
tmp22 = tl.where(tmp19, tmp21, tmp13)
tmp23 = tl.where(tmp5, tmp16, tmp22)
tmp24 = tl.full(tmp23.shape, 0.0, tmp23.dtype)
tmp25 = tl.where(tmp2, tmp23, tmp24)
tmp26 = tmp0 < tmp4
tmp27 = tl.broadcast_to(4 + y0, [XBLOCK, YBLOCK])
tmp28 = tmp27 >= tmp4
tmp29 = tmp27 < tmp1
tmp30 = tmp28 & tmp29
tmp31 = tmp30 & tmp26
tmp32 = tl.load(in_ptr0 + (12 + x2 + (4*y0) + (16*y1)), tmp31 & xmask & ymask, eviction_policy='evict_last', other=0.0)
tmp33 = tl.where(tmp30, tmp32, tmp13)
tmp34 = tl.full(tmp33.shape, 0.0, tmp33.dtype)
tmp35 = tl.where(tmp26, tmp33, tmp34)
tmp36 = tmp0 >= tmp4
tmp37 = tmp0 < tmp1
tmp38 = tmp36 & tmp37
tmp39 = tl.load(in_ptr0 + ((-4) + x2 + (4*y0) + (16*y1)), tmp38 & xmask & ymask, eviction_policy='evict_last', other=0.0)
tmp40 = tl.where(tmp38, tmp39, tmp13)
tmp41 = tl.where(tmp26, tmp35, tmp40)
tmp42 = tl.where(tmp2, tmp25, tmp41)
tl.store(out_ptr0 + (y0 + (6*x2) + (24*y1)), tmp42, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/au/cau4pihcaptiev5y2ewn2o2nvrwhk7hogc72cofmmtbyv4rxc2oy.py
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv1d => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%slice_scatter_default_2, %primals_2, %primals_3, [1], [0], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3), (12, 3, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((4, 4, 6), (24, 6, 1), torch.float32)
# Topologically Sorted Source Nodes: [pad], Original ATen: [aten.copy]
stream0 = get_raw_stream(0)
triton_poi_fused_copy_0.run(primals_1, buf1, 24, 4, grid=grid(24, 4), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_2, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf3, primals_3, 64, grid=grid(64), stream=stream0)
del primals_3
return (reinterpret_tensor(buf3, (4, 4, 4), (16, 1, 4), 0), primals_2, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 3), (12, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class TokenEmbedding(nn.Module):
def __init__(self, c_in, d_model):
super(TokenEmbedding, self).__init__()
padding = 1 if torch.__version__ >= '1.5.0' else 2
self.tokenConv = nn.Conv1d(in_channels=c_in, out_channels=d_model,
kernel_size=3, padding=padding, padding_mode='circular')
for m in self.modules():
if isinstance(m, nn.Conv1d):
nn.init.kaiming_normal_(m.weight, mode='fan_in',
nonlinearity='leaky_relu')
def forward(self, x):
x = self.tokenConv(x.permute(0, 2, 1)).transpose(1, 2)
return x
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'c_in': 4, 'd_model': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_copy_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 24
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
y0 = yindex % 6
x2 = xindex
y1 = yindex // 6
tmp0 = y0
tmp1 = tl.full([1, 1], 5, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.broadcast_to(-4 + y0, [XBLOCK, YBLOCK])
tmp4 = tl.full([1, 1], 1, tl.int64)
tmp5 = tmp3 < tmp4
tmp6 = tmp5 & tmp2
tmp7 = tl.broadcast_to(y0, [XBLOCK, YBLOCK])
tmp8 = tmp7 >= tmp4
tmp9 = tmp7 < tmp1
tmp10 = tmp8 & tmp9
tmp11 = tmp10 & tmp6
tmp12 = tl.load(in_ptr0 + (-4 + x2 + 4 * y0 + 16 * y1), tmp11 & xmask &
ymask, eviction_policy='evict_last', other=0.0)
tmp13 = float('nan')
tmp14 = tl.where(tmp10, tmp12, tmp13)
tmp15 = tl.full(tmp14.shape, 0.0, tmp14.dtype)
tmp16 = tl.where(tmp6, tmp14, tmp15)
tmp17 = tmp3 >= tmp4
tmp18 = tmp3 < tmp1
tmp19 = tmp17 & tmp18
tmp20 = tmp19 & tmp2
tmp21 = tl.load(in_ptr0 + (-20 + x2 + 4 * y0 + 16 * y1), tmp20 & xmask &
ymask, eviction_policy='evict_last', other=0.0)
tmp22 = tl.where(tmp19, tmp21, tmp13)
tmp23 = tl.where(tmp5, tmp16, tmp22)
tmp24 = tl.full(tmp23.shape, 0.0, tmp23.dtype)
tmp25 = tl.where(tmp2, tmp23, tmp24)
tmp26 = tmp0 < tmp4
tmp27 = tl.broadcast_to(4 + y0, [XBLOCK, YBLOCK])
tmp28 = tmp27 >= tmp4
tmp29 = tmp27 < tmp1
tmp30 = tmp28 & tmp29
tmp31 = tmp30 & tmp26
tmp32 = tl.load(in_ptr0 + (12 + x2 + 4 * y0 + 16 * y1), tmp31 & xmask &
ymask, eviction_policy='evict_last', other=0.0)
tmp33 = tl.where(tmp30, tmp32, tmp13)
tmp34 = tl.full(tmp33.shape, 0.0, tmp33.dtype)
tmp35 = tl.where(tmp26, tmp33, tmp34)
tmp36 = tmp0 >= tmp4
tmp37 = tmp0 < tmp1
tmp38 = tmp36 & tmp37
tmp39 = tl.load(in_ptr0 + (-4 + x2 + 4 * y0 + 16 * y1), tmp38 & xmask &
ymask, eviction_policy='evict_last', other=0.0)
tmp40 = tl.where(tmp38, tmp39, tmp13)
tmp41 = tl.where(tmp26, tmp35, tmp40)
tmp42 = tl.where(tmp2, tmp25, tmp41)
tl.store(out_ptr0 + (y0 + 6 * x2 + 24 * y1), tmp42, xmask & ymask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3), (12, 3, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((4, 4, 6), (24, 6, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_copy_0[grid(24, 4)](primals_1, buf1, 24, 4, XBLOCK
=4, YBLOCK=32, num_warps=4, num_stages=1)
del primals_1
buf2 = extern_kernels.convolution(buf1, primals_2, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
buf3 = buf2
del buf2
triton_poi_fused_convolution_1[grid(64)](buf3, primals_3, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_3
return reinterpret_tensor(buf3, (4, 4, 4), (16, 1, 4), 0), primals_2, buf1
class TokenEmbeddingNew(nn.Module):
def __init__(self, c_in, d_model):
super(TokenEmbeddingNew, self).__init__()
padding = 1 if torch.__version__ >= '1.5.0' else 2
self.tokenConv = nn.Conv1d(in_channels=c_in, out_channels=d_model,
kernel_size=3, padding=padding, padding_mode='circular')
for m in self.modules():
if isinstance(m, nn.Conv1d):
nn.init.kaiming_normal_(m.weight, mode='fan_in',
nonlinearity='leaky_relu')
def forward(self, input_0):
primals_2 = self.tokenConv.weight
primals_3 = self.tokenConv.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
TheaperDeng/Informer2020
|
TokenEmbedding
| false | 14,475 |
[
"Apache-2.0"
] | 2,296 |
90e080593e9c345f5f9676359bb3d1618e9aa735
|
https://github.com/TheaperDeng/Informer2020/tree/90e080593e9c345f5f9676359bb3d1618e9aa735
|
Normalize
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/6o/c6ojhuuvyz43ogs3olib4lkalnymxhhnoqfyye667gq5hhhh5aqy.py
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, add, out], Original ATen: [aten.pow, aten.sum, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# norm => pow_2
# out => div
# pow_1 => pow_1
# sum_1 => sum_1
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%arg0_1, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [1], True), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sum_1, 0.5), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%pow_2, 1e-07), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, %add), kwargs = {})
triton_poi_fused_add_div_pow_sum_0 = async_compile.triton('triton_poi_fused_add_div_pow_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_pow_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_pow_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x3), tmp15, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, add, out], Original ATen: [aten.pow, aten.sum, aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_pow_sum_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch
import torch.nn as nn
class Normalize(nn.Module):
def __init__(self, power=2):
super(Normalize, self).__init__()
self.power = power
def forward(self, x):
norm = x.pow(self.power).sum(1, keepdim=True).pow(1.0 / self.power)
out = x.div(norm + 1e-07)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.utils.data
import torch
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_pow_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + x3, tmp15, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_pow_sum_0[grid(256)](arg0_1, buf0, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class NormalizeNew(nn.Module):
def __init__(self, power=2):
super(NormalizeNew, self).__init__()
self.power = power
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
Theomat/colorization-av-enseirb-2020
|
Normalize
| false | 14,476 |
[
"Apache-2.0"
] | 1,422 |
c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
https://github.com/Theomat/colorization-av-enseirb-2020/tree/c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
Linear3D
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/64/c64ahxnpt5ixqrlolbug3qf6y4u2zqmcjekif2yu4ba4hcze2fom.py
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# output => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = (xindex // 16) % 4
x2 = (xindex // 64)
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (16*x2) + (64*x1)), xmask)
tl.store(out_ptr0 + (x3), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/2g/c2gow746iojnl6yugujjn3non5klwrqsxgmhc4ib5irxlfwbv7ap.py
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# output => clone_1
# Graph fragment:
# %clone_1 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_1,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_1 = async_compile.triton('triton_poi_fused_clone_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 64
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ri/cripmqky3lo3skozwggzdw7u7ybokgg2w2o3vcesam3ywnqu73dg.py
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# output_1 => add
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_2, %unsqueeze), kwargs = {})
triton_poi_fused_add_2 = async_compile.triton('triton_poi_fused_add_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 4
x2 = (xindex // 16) % 4
tmp0 = tl.load(in_out_ptr0 + (x4), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (4*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x4), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(primals_1, buf0, 256, grid=grid(256), stream=stream0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
triton_poi_fused_clone_1.run(primals_2, buf1, 256, grid=grid(256), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf1, (16, 4, 4), (16, 4, 1), 0), out=buf2)
del buf1
buf3 = reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.add]
triton_poi_fused_add_2.run(buf3, primals_3, 256, grid=grid(256), stream=stream0)
del primals_3
return (reinterpret_tensor(buf3, (4, 4, 4, 4), (16, 64, 4, 1), 0), reinterpret_tensor(buf0, (16, 4, 4), (16, 1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
import torch as th
from torch.nn import Parameter
def functional_linear3d(input, weight, bias=None):
"""
Apply a linear transformation to the incoming data: :math:`y = xA^T + b`.
Shape:
- Input: :math:`(N, *, in\\_features)` where `*` means any number of
additional dimensions
- Weight: :math:`(out\\_features, in\\_features)`
- Bias: :math:`(out\\_features)`
- Output: :math:`(N, *, out\\_features)`
"""
output = input.transpose(0, 1).matmul(weight)
if bias is not None:
output += bias.unsqueeze(1)
return output.transpose(0, 1)
class Linear3D(th.nn.Module):
"""Applies a linear transformation to the incoming data: :math:`y = Ax + b`.
Args:
in_features: size of each input sample
out_features: size of each output sample
bias: If set to False, the layer will not learn an additive bias.
Default: ``True``
Shape:
- Input: :math:`(N, *, in\\_features)` where :math:`*` means any number of
additional dimensions
- Output: :math:`(N, *, out\\_features)` where all but the last dimension
are the same shape as the input.
Attributes:
weight: the learnable weights of the module of shape
`(out_features x in_features)`
bias: the learnable bias of the module of shape `(out_features)`
Examples::
>>> m = nn.Linear(3, 20, 30)
>>> input = torch.randn(128, 20)
>>> output = m(input)
>>> print(output.size())
"""
def __init__(self, channels, in_features, out_features, batch_size=-1,
bias=True, noise=False):
super(Linear3D, self).__init__()
self.in_features = in_features
self.out_features = out_features
self.channels = channels
if noise:
self.in_features += 1
self.weight = Parameter(th.Tensor(channels, self.in_features,
out_features))
if bias:
self.bias = Parameter(th.Tensor(channels, out_features))
else:
self.register_parameter('bias', None)
if noise:
self.register_buffer('noise', th.Tensor(batch_size, channels, 1))
self.reset_parameters()
def reset_parameters(self):
stdv = 1.0 / math.sqrt(self.weight.size(1))
self.weight.data.uniform_(-stdv, stdv)
if self.bias is not None:
self.bias.data.uniform_(-stdv, stdv)
def forward(self, input, adj_matrix=None, permutation_matrix=None):
input_ = [input]
if input.dim() == 2:
if permutation_matrix is not None:
input_.append(input.unsqueeze(1).expand([input.shape[0],
self.channels, permutation_matrix.shape[1]]))
elif hasattr(self, 'noise'):
input_.append(input.unsqueeze(1).expand([input.shape[0],
self.channels, self.in_features - 1]))
else:
input_.append(input.unsqueeze(1).expand([input.shape[0],
self.channels, self.in_features]))
if adj_matrix is not None and permutation_matrix is not None:
input_.append((input_[-1].transpose(0, 1) @ (adj_matrix.t().
unsqueeze(2) * permutation_matrix)).transpose(0, 1))
elif adj_matrix is not None:
input_.append(input_[-1] * adj_matrix.t().unsqueeze(0))
elif permutation_matrix is not None:
input_.append((input_[-1].transpose(0, 1) @ permutation_matrix).t()
)
if hasattr(self, 'noise'):
self.noise.normal_()
input_.append(th.cat([input_[-1], self.noise], 2))
return functional_linear3d(input_[-1], self.weight, self.bias)
def extra_repr(self):
return 'in_features={}, out_features={}, bias={}'.format(self.
in_features, self.out_features, self.bias is not None)
def apply_filter(self, permutation_matrix):
transpose_weight = self.weight.transpose(1, 2) @ permutation_matrix
self.weight = Parameter(transpose_weight.transpose(1, 2))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channels': 4, 'in_features': 4, 'out_features': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import math
import torch as th
from torch.nn import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = xindex // 16 % 4
x2 = xindex // 64
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 16 * x2 + 64 * x1), xmask)
tl.store(out_ptr0 + x3, tmp0, xmask)
@triton.jit
def triton_poi_fused_clone_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 64
x2 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + x2, tmp0, xmask)
@triton.jit
def triton_poi_fused_add_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 4
x2 = xindex // 16 % 4
tmp0 = tl.load(in_out_ptr0 + x4, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x4, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(256)](primals_1, buf0, 256, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_clone_1[grid(256)](primals_2, buf1, 256, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf1, (16, 4, 4), (16, 4, 1), 0), out=buf2)
del buf1
buf3 = reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf2
triton_poi_fused_add_2[grid(256)](buf3, primals_3, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del primals_3
return reinterpret_tensor(buf3, (4, 4, 4, 4), (16, 64, 4, 1), 0
), reinterpret_tensor(buf0, (16, 4, 4), (16, 1, 4), 0)
def functional_linear3d(input, weight, bias=None):
"""
Apply a linear transformation to the incoming data: :math:`y = xA^T + b`.
Shape:
- Input: :math:`(N, *, in\\_features)` where `*` means any number of
additional dimensions
- Weight: :math:`(out\\_features, in\\_features)`
- Bias: :math:`(out\\_features)`
- Output: :math:`(N, *, out\\_features)`
"""
output = input.transpose(0, 1).matmul(weight)
if bias is not None:
output += bias.unsqueeze(1)
return output.transpose(0, 1)
class Linear3DNew(th.nn.Module):
"""Applies a linear transformation to the incoming data: :math:`y = Ax + b`.
Args:
in_features: size of each input sample
out_features: size of each output sample
bias: If set to False, the layer will not learn an additive bias.
Default: ``True``
Shape:
- Input: :math:`(N, *, in\\_features)` where :math:`*` means any number of
additional dimensions
- Output: :math:`(N, *, out\\_features)` where all but the last dimension
are the same shape as the input.
Attributes:
weight: the learnable weights of the module of shape
`(out_features x in_features)`
bias: the learnable bias of the module of shape `(out_features)`
Examples::
>>> m = nn.Linear(3, 20, 30)
>>> input = torch.randn(128, 20)
>>> output = m(input)
>>> print(output.size())
"""
def __init__(self, channels, in_features, out_features, batch_size=-1,
bias=True, noise=False):
super(Linear3DNew, self).__init__()
self.in_features = in_features
self.out_features = out_features
self.channels = channels
if noise:
self.in_features += 1
self.weight = Parameter(th.Tensor(channels, self.in_features,
out_features))
if bias:
self.bias = Parameter(th.Tensor(channels, out_features))
else:
self.register_parameter('bias', None)
if noise:
self.register_buffer('noise', th.Tensor(batch_size, channels, 1))
self.reset_parameters()
def reset_parameters(self):
stdv = 1.0 / math.sqrt(self.weight.size(1))
self.weight.data.uniform_(-stdv, stdv)
if self.bias is not None:
self.bias.data.uniform_(-stdv, stdv)
def extra_repr(self):
return 'in_features={}, out_features={}, bias={}'.format(self.
in_features, self.out_features, self.bias is not None)
def apply_filter(self, permutation_matrix):
transpose_weight = self.weight.transpose(1, 2) @ permutation_matrix
self.weight = Parameter(transpose_weight.transpose(1, 2))
def forward(self, input_0):
primals_2 = self.weight
primals_3 = self.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
TheSignPainter/CausalDiscoveryToolbox
|
Linear3D
| false | 14,477 |
[
"MIT"
] | 528 |
33eae18184905e505be978b08003b9477bf38e0c
|
https://github.com/TheSignPainter/CausalDiscoveryToolbox/tree/33eae18184905e505be978b08003b9477bf38e0c
|
TemporalEmbedding
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/aj/cajffqesecvkyqpb4xsm3itr6xvdpqs6wg4viw32uiltwyyv44is.py
# Topologically Sorted Source Nodes: [embedding, embedding_1, add, embedding_2, add_1, embedding_3, add_2, add_3], Original ATen: [aten.embedding, aten.add]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# add_2 => add_2
# add_3 => add_3
# embedding => embedding
# embedding_1 => embedding_1
# embedding_2 => embedding_2
# embedding_3 => embedding_3
# Graph fragment:
# %embedding : [num_users=1] = call_function[target=torch.ops.aten.embedding.default](args = (%arg1_1, %select), kwargs = {})
# %embedding_1 : [num_users=1] = call_function[target=torch.ops.aten.embedding.default](args = (%arg2_1, %select_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%embedding, %embedding_1), kwargs = {})
# %embedding_2 : [num_users=1] = call_function[target=torch.ops.aten.embedding.default](args = (%arg3_1, %select_2), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add, %embedding_2), kwargs = {})
# %embedding_3 : [num_users=1] = call_function[target=torch.ops.aten.embedding.default](args = (%arg4_1, %select_3), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_1, %embedding_3), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_2, 0.0), kwargs = {})
triton_poi_fused_add_embedding_0 = async_compile.triton('triton_poi_fused_add_embedding_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_embedding_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_embedding_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x2 = (xindex // 16)
x0 = xindex % 4
x4 = xindex
tmp0 = tl.load(in_ptr0 + (12 + x1 + (16*x2)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (8 + x1 + (16*x2)), xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr0 + (4 + x1 + (16*x2)), xmask, eviction_policy='evict_last')
tmp26 = tl.load(in_ptr0 + (x1 + (16*x2)), xmask, eviction_policy='evict_last')
tmp1 = tmp0.to(tl.int64)
tmp2 = tl.full([XBLOCK], 24, tl.int32)
tmp3 = tmp1 + tmp2
tmp4 = tmp1 < 0
tmp5 = tl.where(tmp4, tmp3, tmp1)
tl.device_assert(((0 <= tmp5) & (tmp5 < 24)) | ~(xmask), "index out of bounds: 0 <= tmp5 < 24")
tmp7 = tl.load(in_ptr1 + (x0 + (4*tmp5)), xmask)
tmp9 = tmp8.to(tl.int64)
tmp10 = tl.full([XBLOCK], 7, tl.int32)
tmp11 = tmp9 + tmp10
tmp12 = tmp9 < 0
tmp13 = tl.where(tmp12, tmp11, tmp9)
tl.device_assert(((0 <= tmp13) & (tmp13 < 7)) | ~(xmask), "index out of bounds: 0 <= tmp13 < 7")
tmp15 = tl.load(in_ptr2 + (x0 + (4*tmp13)), xmask)
tmp16 = tmp7 + tmp15
tmp18 = tmp17.to(tl.int64)
tmp19 = tl.full([XBLOCK], 32, tl.int32)
tmp20 = tmp18 + tmp19
tmp21 = tmp18 < 0
tmp22 = tl.where(tmp21, tmp20, tmp18)
tl.device_assert(((0 <= tmp22) & (tmp22 < 32)) | ~(xmask), "index out of bounds: 0 <= tmp22 < 32")
tmp24 = tl.load(in_ptr3 + (x0 + (4*tmp22)), xmask)
tmp25 = tmp16 + tmp24
tmp27 = tmp26.to(tl.int64)
tmp28 = tl.full([XBLOCK], 13, tl.int32)
tmp29 = tmp27 + tmp28
tmp30 = tmp27 < 0
tmp31 = tl.where(tmp30, tmp29, tmp27)
tl.device_assert(((0 <= tmp31) & (tmp31 < 13)) | ~(xmask), "index out of bounds: 0 <= tmp31 < 13")
tmp33 = tl.load(in_ptr4 + (x0 + (4*tmp31)), xmask)
tmp34 = tmp25 + tmp33
tmp35 = 0.0
tmp36 = tmp34 + tmp35
tl.store(out_ptr0 + (x4), tmp36, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1, arg4_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (24, 4), (4, 1))
assert_size_stride(arg2_1, (7, 4), (4, 1))
assert_size_stride(arg3_1, (32, 4), (4, 1))
assert_size_stride(arg4_1, (13, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [embedding, embedding_1, add, embedding_2, add_1, embedding_3, add_2, add_3], Original ATen: [aten.embedding, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_embedding_0.run(arg0_1, arg1_1, arg2_1, arg3_1, arg4_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
del arg4_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((24, 4), (4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((7, 4), (4, 1), device='cuda:0', dtype=torch.float32)
arg3_1 = rand_strided((32, 4), (4, 1), device='cuda:0', dtype=torch.float32)
arg4_1 = rand_strided((13, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1, arg3_1, arg4_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
import torch.nn as nn
class FixedEmbedding(nn.Module):
def __init__(self, c_in, d_model):
super(FixedEmbedding, self).__init__()
w = torch.zeros(c_in, d_model).float()
w.require_grad = False
position = torch.arange(0, c_in).float().unsqueeze(1)
div_term = (torch.arange(0, d_model, 2).float() * -(math.log(
10000.0) / d_model)).exp()
w[:, 0::2] = torch.sin(position * div_term)
w[:, 1::2] = torch.cos(position * div_term)
self.emb = nn.Embedding(c_in, d_model)
self.emb.weight = nn.Parameter(w, requires_grad=False)
def forward(self, x):
return self.emb(x).detach()
class TemporalEmbedding(nn.Module):
def __init__(self, d_model, embed_type='fixed', freq='h'):
super(TemporalEmbedding, self).__init__()
minute_size = 4
hour_size = 24
weekday_size = 7
day_size = 32
month_size = 13
Embed = FixedEmbedding if embed_type == 'fixed' else nn.Embedding
if freq == 't':
self.minute_embed = Embed(minute_size, d_model)
self.hour_embed = Embed(hour_size, d_model)
self.weekday_embed = Embed(weekday_size, d_model)
self.day_embed = Embed(day_size, d_model)
self.month_embed = Embed(month_size, d_model)
def forward(self, x):
x = x.long()
minute_x = self.minute_embed(x[:, :, 4]) if hasattr(self,
'minute_embed') else 0.0
hour_x = self.hour_embed(x[:, :, 3])
weekday_x = self.weekday_embed(x[:, :, 2])
day_x = self.day_embed(x[:, :, 1])
month_x = self.month_embed(x[:, :, 0])
return hour_x + weekday_x + day_x + month_x + minute_x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'d_model': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_embedding_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x2 = xindex // 16
x0 = xindex % 4
x4 = xindex
tmp0 = tl.load(in_ptr0 + (12 + x1 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp8 = tl.load(in_ptr0 + (8 + x1 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp17 = tl.load(in_ptr0 + (4 + x1 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp26 = tl.load(in_ptr0 + (x1 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp1 = tmp0.to(tl.int64)
tmp2 = tl.full([XBLOCK], 24, tl.int32)
tmp3 = tmp1 + tmp2
tmp4 = tmp1 < 0
tmp5 = tl.where(tmp4, tmp3, tmp1)
tl.device_assert((0 <= tmp5) & (tmp5 < 24) | ~xmask,
'index out of bounds: 0 <= tmp5 < 24')
tmp7 = tl.load(in_ptr1 + (x0 + 4 * tmp5), xmask)
tmp9 = tmp8.to(tl.int64)
tmp10 = tl.full([XBLOCK], 7, tl.int32)
tmp11 = tmp9 + tmp10
tmp12 = tmp9 < 0
tmp13 = tl.where(tmp12, tmp11, tmp9)
tl.device_assert((0 <= tmp13) & (tmp13 < 7) | ~xmask,
'index out of bounds: 0 <= tmp13 < 7')
tmp15 = tl.load(in_ptr2 + (x0 + 4 * tmp13), xmask)
tmp16 = tmp7 + tmp15
tmp18 = tmp17.to(tl.int64)
tmp19 = tl.full([XBLOCK], 32, tl.int32)
tmp20 = tmp18 + tmp19
tmp21 = tmp18 < 0
tmp22 = tl.where(tmp21, tmp20, tmp18)
tl.device_assert((0 <= tmp22) & (tmp22 < 32) | ~xmask,
'index out of bounds: 0 <= tmp22 < 32')
tmp24 = tl.load(in_ptr3 + (x0 + 4 * tmp22), xmask)
tmp25 = tmp16 + tmp24
tmp27 = tmp26.to(tl.int64)
tmp28 = tl.full([XBLOCK], 13, tl.int32)
tmp29 = tmp27 + tmp28
tmp30 = tmp27 < 0
tmp31 = tl.where(tmp30, tmp29, tmp27)
tl.device_assert((0 <= tmp31) & (tmp31 < 13) | ~xmask,
'index out of bounds: 0 <= tmp31 < 13')
tmp33 = tl.load(in_ptr4 + (x0 + 4 * tmp31), xmask)
tmp34 = tmp25 + tmp33
tmp35 = 0.0
tmp36 = tmp34 + tmp35
tl.store(out_ptr0 + x4, tmp36, xmask)
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1, arg4_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (24, 4), (4, 1))
assert_size_stride(arg2_1, (7, 4), (4, 1))
assert_size_stride(arg3_1, (32, 4), (4, 1))
assert_size_stride(arg4_1, (13, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_embedding_0[grid(256)](arg0_1, arg1_1, arg2_1,
arg3_1, arg4_1, buf0, 256, XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
del arg4_1
return buf0,
class FixedEmbedding(nn.Module):
def __init__(self, c_in, d_model):
super(FixedEmbedding, self).__init__()
w = torch.zeros(c_in, d_model).float()
w.require_grad = False
position = torch.arange(0, c_in).float().unsqueeze(1)
div_term = (torch.arange(0, d_model, 2).float() * -(math.log(
10000.0) / d_model)).exp()
w[:, 0::2] = torch.sin(position * div_term)
w[:, 1::2] = torch.cos(position * div_term)
self.emb = nn.Embedding(c_in, d_model)
self.emb.weight = nn.Parameter(w, requires_grad=False)
def forward(self, x):
return self.emb(x).detach()
class TemporalEmbeddingNew(nn.Module):
def __init__(self, d_model, embed_type='fixed', freq='h'):
super(TemporalEmbeddingNew, self).__init__()
minute_size = 4
hour_size = 24
weekday_size = 7
day_size = 32
month_size = 13
Embed = FixedEmbedding if embed_type == 'fixed' else nn.Embedding
if freq == 't':
self.minute_embed = Embed(minute_size, d_model)
self.hour_embed = Embed(hour_size, d_model)
self.weekday_embed = Embed(weekday_size, d_model)
self.day_embed = Embed(day_size, d_model)
self.month_embed = Embed(month_size, d_model)
def forward(self, input_0):
arg1_1 = self.hour_embed.emb.weight
arg2_1 = self.weekday_embed.emb.weight
arg3_1 = self.day_embed.emb.weight
arg4_1 = self.month_embed.emb.weight
arg0_1 = input_0
output = call([arg0_1, arg1_1, arg2_1, arg3_1, arg4_1])
return output[0]
|
TheaperDeng/Informer2020
|
TemporalEmbedding
| false | 14,478 |
[
"Apache-2.0"
] | 2,296 |
90e080593e9c345f5f9676359bb3d1618e9aa735
|
https://github.com/TheaperDeng/Informer2020/tree/90e080593e9c345f5f9676359bb3d1618e9aa735
|
EqualConv2d
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/zs/czsbkexmu6ywpra7jqion5n6drhfl2liw6og7nt2lnvf5ix7ikrs.py
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul_1 => mul_1
# Graph fragment:
# %mul_1 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_2, 0.125), kwargs = {})
triton_poi_fused_mul_0 = async_compile.triton('triton_poi_fused_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.125
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/tc/ctcagp37ljugm52zu6ckorigrppqo67voefe2f2odg5r6hyllhyu.py
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# out => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %mul_1, %primals_1, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, ), (1, ))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_0.run(primals_2, buf0, 256, grid=grid(256), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(primals_3, buf0, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 1, 1), (4, 1, 1, 1))
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf2, primals_1, 16, grid=grid(16), stream=stream0)
del primals_1
return (buf2, primals_3, buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
from torch import nn
import torch.nn.functional as F
class EqualConv2d(nn.Module):
def __init__(self, in_channel, out_channel, kernel_size, groups=1,
stride=1, padding=0, bias=True, lr_mul=1):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_channel, in_channel //
groups, kernel_size, kernel_size).div_(lr_mul))
self.scale = lr_mul / math.sqrt(in_channel // groups * kernel_size ** 2
)
self.stride = stride
self.padding = padding
self.groups = groups
self.lr_mul = lr_mul
if bias:
self.bias = nn.Parameter(torch.zeros(out_channel))
else:
self.bias = None
def forward(self, input):
self.bias * self.lr_mul if self.bias is not None else None
out = F.conv2d(input, self.weight * self.scale, bias=self.bias,
stride=self.stride, padding=self.padding, groups=self.groups)
return out
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]}, {self.weight.shape[2]}, stride={self.stride}, padding={self.padding})'
)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channel': 4, 'out_channel': 4, 'kernel_size': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import math
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.125
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4,), (1,))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_0[grid(256)](primals_2, buf0, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del primals_2
buf1 = extern_kernels.convolution(primals_3, buf0, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 1, 1), (4, 1, 1, 1))
buf2 = buf1
del buf1
triton_poi_fused_convolution_1[grid(16)](buf2, primals_1, 16,
XBLOCK=16, num_warps=1, num_stages=1)
del primals_1
return buf2, primals_3, buf0
class EqualConv2dNew(nn.Module):
def __init__(self, in_channel, out_channel, kernel_size, groups=1,
stride=1, padding=0, bias=True, lr_mul=1):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_channel, in_channel //
groups, kernel_size, kernel_size).div_(lr_mul))
self.scale = lr_mul / math.sqrt(in_channel // groups * kernel_size ** 2
)
self.stride = stride
self.padding = padding
self.groups = groups
self.lr_mul = lr_mul
if bias:
self.bias = nn.Parameter(torch.zeros(out_channel))
else:
self.bias = None
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]}, {self.weight.shape[2]}, stride={self.stride}, padding={self.padding})'
)
def forward(self, input_0):
primals_2 = self.weight
primals_1 = self.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Tiamat-Tech/RetrieveInStyle
|
EqualConv2d
| false | 14,479 |
[
"MIT"
] | 53 |
c5714b9c3c219c9ba463f3e162083458702038c1
|
https://github.com/Tiamat-Tech/RetrieveInStyle/tree/c5714b9c3c219c9ba463f3e162083458702038c1
|
HuberLoss
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/h3/ch3qidii4qfm5ghapin7tnms27ccjbq3fj2gekdrfdy7mbjfc6tb.py
# Topologically Sorted Source Nodes: [truediv, truediv_1, loss, mul, mul_1], Original ATen: [aten.div, aten.smooth_l1_loss, aten.mul]
# Source node to ATen node mapping:
# loss => abs_1, div_2, lt, mean, mul, pow_1, sub, sub_1, where
# mul => mul_1
# mul_1 => mul_2
# truediv => div
# truediv_1 => div_1
# Graph fragment:
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, 1), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg1_1, 1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%div, %div_1), kwargs = {})
# %abs_1 : [num_users=3] = call_function[target=torch.ops.aten.abs.default](args = (%sub,), kwargs = {})
# %lt : [num_users=1] = call_function[target=torch.ops.aten.lt.Scalar](args = (%abs_1, 1.0), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%abs_1, 2), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%pow_1, 0.5), kwargs = {})
# %div_2 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul, 1.0), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%abs_1, 0.5), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%lt, %div_2, %sub_1), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%where,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mean, 1), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_1, 1), kwargs = {})
triton_per_fused_div_mul_smooth_l1_loss_0 = async_compile.triton('triton_per_fused_div_mul_smooth_l1_loss_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_div_mul_smooth_l1_loss_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_div_mul_smooth_l1_loss_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp3 = tl.load(in_ptr1 + (r0), None)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp5 = tmp2 - tmp4
tmp6 = tl_math.abs(tmp5)
tmp7 = tmp6 < tmp1
tmp8 = tmp6 * tmp6
tmp9 = 0.5
tmp10 = tmp8 * tmp9
tmp11 = tmp10 * tmp1
tmp12 = tmp6 - tmp9
tmp13 = tl.where(tmp7, tmp11, tmp12)
tmp14 = tl.broadcast_to(tmp13, [RBLOCK])
tmp16 = triton_helpers.promote_to_tensor(tl.sum(tmp14, 0))
tmp17 = 256.0
tmp18 = tmp16 / tmp17
tmp19 = tmp18 * tmp1
tmp20 = tmp19 * tmp1
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp20, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [truediv, truediv_1, loss, mul, mul_1], Original ATen: [aten.div, aten.smooth_l1_loss, aten.mul]
stream0 = get_raw_stream(0)
triton_per_fused_div_mul_smooth_l1_loss_0.run(buf1, arg0_1, arg1_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.utils.data
class HuberLoss(nn.Module):
def __init__(self, delta=1):
super().__init__()
self.huber_loss_delta1 = nn.SmoothL1Loss()
self.delta = delta
def forward(self, x, x_hat):
loss = self.huber_loss_delta1(x / self.delta, x_hat / self.delta)
return loss * self.delta * self.delta
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_div_mul_smooth_l1_loss_0(in_out_ptr0, in_ptr0, in_ptr1,
xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp3 = tl.load(in_ptr1 + r0, None)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp5 = tmp2 - tmp4
tmp6 = tl_math.abs(tmp5)
tmp7 = tmp6 < tmp1
tmp8 = tmp6 * tmp6
tmp9 = 0.5
tmp10 = tmp8 * tmp9
tmp11 = tmp10 * tmp1
tmp12 = tmp6 - tmp9
tmp13 = tl.where(tmp7, tmp11, tmp12)
tmp14 = tl.broadcast_to(tmp13, [RBLOCK])
tmp16 = triton_helpers.promote_to_tensor(tl.sum(tmp14, 0))
tmp17 = 256.0
tmp18 = tmp16 / tmp17
tmp19 = tmp18 * tmp1
tmp20 = tmp19 * tmp1
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp20, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_div_mul_smooth_l1_loss_0[grid(1)](buf1, arg0_1,
arg1_1, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf1,
class HuberLossNew(nn.Module):
def __init__(self, delta=1):
super().__init__()
self.huber_loss_delta1 = nn.SmoothL1Loss()
self.delta = delta
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
Thibaud-Ardoin/d4rl_evaluations
|
HuberLoss
| false | 14,480 |
[
"Apache-2.0"
] | 123 |
135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
https://github.com/Thibaud-Ardoin/d4rl_evaluations/tree/135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
CNormalized_Linear
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ef/cefxo2fn2nc6kz3xqsftatxgyvfkkm7bbs7raicfjscwteq4iixh.py
# Topologically Sorted Source Nodes: [pow_1, sum_1, sqrt, div], Original ATen: [aten.pow, aten.sum, aten.sqrt, aten.div]
# Source node to ATen node mapping:
# div => div
# pow_1 => pow_1
# sqrt => sqrt
# sum_1 => sum_1
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%primals_1, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [0]), kwargs = {})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%sum_1,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%primals_1, %sqrt), kwargs = {})
triton_poi_fused_div_pow_sqrt_sum_0 = async_compile.triton('triton_poi_fused_div_pow_sqrt_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_pow_sqrt_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_pow_sqrt_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (4 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (8 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (12 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = tmp0 / tmp12
tl.store(out_ptr0 + (x2), tmp13, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, sum_1, sqrt, div], Original ATen: [aten.pow, aten.sum, aten.sqrt, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_div_pow_sqrt_sum_0.run(primals_1, buf0, 16, grid=grid(16), stream=stream0)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (64, 4), (4, 1), 0), reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf1)
del buf0
return (reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0), primals_1, reinterpret_tensor(primals_2, (64, 4), (4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
import torch as th
class CNormalized_Linear(th.nn.Module):
"""Linear layer with column-wise normalized input matrix."""
def __init__(self, in_features, out_features, bias=False):
"""Initialize the layer."""
super(CNormalized_Linear, self).__init__()
self.in_features = in_features
self.out_features = out_features
self.weight = th.nn.Parameter(th.Tensor(out_features, in_features))
if bias:
self.bias = th.nn.Parameter(th.Tensor(out_features))
else:
self.register_parameter('bias', None)
self.reset_parameters()
def reset_parameters(self):
"""Reset the parameters."""
stdv = 1.0 / math.sqrt(self.weight.size(1))
self.weight.data.uniform_(-stdv, stdv)
if self.bias is not None:
self.bias.data.uniform_(-stdv, stdv)
def forward(self, input):
"""Feed-forward through the network."""
return th.nn.functional.linear(input, self.weight.div(self.weight.
pow(2).sum(0).sqrt()))
def __repr__(self):
"""For print purposes."""
return self.__class__.__name__ + '(' + 'in_features=' + str(self.
in_features) + ', out_features=' + str(self.out_features
) + ', bias=' + str(self.bias is not None) + ')'
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_features': 4, 'out_features': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import math
import torch as th
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_div_pow_sqrt_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (4 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (8 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (12 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = tmp0 / tmp12
tl.store(out_ptr0 + x2, tmp13, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_div_pow_sqrt_sum_0[grid(16)](primals_1, buf0, 16,
XBLOCK=16, num_warps=1, num_stages=1)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (64, 4), (4, 1), 0),
reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf1)
del buf0
return reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0
), primals_1, reinterpret_tensor(primals_2, (64, 4), (4, 1), 0)
class CNormalized_LinearNew(th.nn.Module):
"""Linear layer with column-wise normalized input matrix."""
def __init__(self, in_features, out_features, bias=False):
"""Initialize the layer."""
super(CNormalized_LinearNew, self).__init__()
self.in_features = in_features
self.out_features = out_features
self.weight = th.nn.Parameter(th.Tensor(out_features, in_features))
if bias:
self.bias = th.nn.Parameter(th.Tensor(out_features))
else:
self.register_parameter('bias', None)
self.reset_parameters()
def reset_parameters(self):
"""Reset the parameters."""
stdv = 1.0 / math.sqrt(self.weight.size(1))
self.weight.data.uniform_(-stdv, stdv)
if self.bias is not None:
self.bias.data.uniform_(-stdv, stdv)
def __repr__(self):
"""For print purposes."""
return self.__class__.__name__ + '(' + 'in_features=' + str(self.
in_features) + ', out_features=' + str(self.out_features
) + ', bias=' + str(self.bias is not None) + ')'
def forward(self, input_0):
primals_1 = self.weight
primals_2 = input_0
output = call([primals_1, primals_2])
return output[0]
|
TheSignPainter/CausalDiscoveryToolbox
|
CNormalized_Linear
| false | 14,481 |
[
"MIT"
] | 528 |
33eae18184905e505be978b08003b9477bf38e0c
|
https://github.com/TheSignPainter/CausalDiscoveryToolbox/tree/33eae18184905e505be978b08003b9477bf38e0c
|
MultiHead
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/fz/cfzmg4qtw6jgry4nhlwopodzjz62ll3n3ykfox77hwd2crdnlh2w.py
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# softmax => exp
# Graph fragment:
# %mul_tensor_3 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%bmm, 1), kwargs = {})
# %amax_default_3 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_tensor_3, [-1], True), kwargs = {})
# %sub_tensor_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_tensor_3, %amax_default_3), kwargs = {})
# %div_tensor_3 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_tensor_3, 2.0), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%div_tensor_3,), kwargs = {})
triton_poi_fused__softmax_0 = async_compile.triton('triton_poi_fused__softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp3 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = 0.5
tmp16 = tmp14 * tmp15
tmp17 = tl_math.exp(tmp16)
tl.store(out_ptr0 + (x2), tmp17, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/kj/ckjtlefzavjukjsytvkak6ek26zmzexpcbnlwelx4k5kascjxlf3.py
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# softmax => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/mk/cmkim2hc4ksxhatli3y5cu7hoqofxcbzqrrxvnlhmswdt4cgww25.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%bmm_1, %bmm_3, %bmm_5, %bmm_7], -1), kwargs = {})
triton_poi_fused_cat_2 = async_compile.triton('triton_poi_fused_cat_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x1), tmp9 & xmask, eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x1), tmp14 & xmask, eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tmp17 = tl.full([1], 4, tl.int64)
tmp18 = tmp0 < tmp17
tmp19 = tl.load(in_ptr3 + (x1), tmp16 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + (x2), tmp22, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [query], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [key], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_4, (16, 4), (4, 1), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [value], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_6, (16, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [dot_products], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 0), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__softmax_0.run(buf3, buf4, 64, grid=grid(64), stream=stream0)
buf5 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf4, buf5, 64, grid=grid(64), stream=stream0)
buf6 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_1], Original ATen: [aten.bmm]
extern_kernels.bmm(buf5, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 0), out=buf6)
buf7 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [dot_products_1], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 1), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 1), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax_1], Original ATen: [aten._softmax]
triton_poi_fused__softmax_0.run(buf7, buf8, 64, grid=grid(64), stream=stream0)
buf9 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [softmax_1], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf8, buf9, 64, grid=grid(64), stream=stream0)
buf10 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_3], Original ATen: [aten.bmm]
extern_kernels.bmm(buf9, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 1), out=buf10)
buf11 = buf8; del buf8 # reuse
# Topologically Sorted Source Nodes: [dot_products_2], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 2), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 2), out=buf11)
buf12 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_0.run(buf11, buf12, 64, grid=grid(64), stream=stream0)
buf13 = buf11; del buf11 # reuse
# Topologically Sorted Source Nodes: [softmax_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf12, buf13, 64, grid=grid(64), stream=stream0)
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_5], Original ATen: [aten.bmm]
extern_kernels.bmm(buf13, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 2), out=buf14)
buf15 = buf12; del buf12 # reuse
# Topologically Sorted Source Nodes: [dot_products_3], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 3), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 3), out=buf15)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_0.run(buf15, buf16, 64, grid=grid(64), stream=stream0)
buf17 = buf15; del buf15 # reuse
# Topologically Sorted Source Nodes: [softmax_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf16, buf17, 64, grid=grid(64), stream=stream0)
buf18 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_7], Original ATen: [aten.bmm]
extern_kernels.bmm(buf17, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 3), out=buf18)
buf19 = buf16; del buf16 # reuse
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
triton_poi_fused_cat_2.run(buf6, buf10, buf14, buf18, buf19, 64, grid=grid(64), stream=stream0)
del buf10
del buf14
del buf18
del buf6
buf20 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf19, (16, 4), (4, 1), 0), reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf20)
return (reinterpret_tensor(buf20, (4, 4, 4), (16, 4, 1), 0), reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_4, (16, 4), (4, 1), 0), reinterpret_tensor(primals_6, (16, 4), (4, 1), 0), buf5, buf9, buf13, buf17, reinterpret_tensor(buf19, (16, 4), (4, 1), 0), primals_7, reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 3), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 3), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 3), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 2), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 2), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 2), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 1), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 1), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 1), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 0), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 0), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
from torch.nn import functional as F
from torch import nn
def matmul(x, y):
if x.dim() == y.dim():
return torch.matmul(x, y)
if x.dim() == y.dim() - 1:
return torch.matmul(x.unsqueeze(-2), y).squeeze(-2)
return torch.matmul(x, y.unsqueeze(-2)).squeeze(-2)
class Attention(nn.Module):
def __init__(self, d_key, drop_ratio, causal):
super(Attention, self).__init__()
self.scale = math.sqrt(d_key)
self.dropout = nn.Dropout(drop_ratio)
self.causal = causal
def forward(self, query, key, value):
dot_products = matmul(query, key.transpose(1, 2))
if query.dim() == 3 and (self is None or self.causal):
tri = torch.ones(key.size(1), key.size(1)).triu(1) * INF
if key.is_cuda:
tri = tri
dot_products.data.sub_(tri.unsqueeze(0))
return matmul(self.dropout(F.softmax(dot_products / self.scale, dim
=-1)), value)
class MultiHead(nn.Module):
def __init__(self, d_key, d_value, n_heads, drop_ratio, causal=False):
super(MultiHead, self).__init__()
self.attention = Attention(d_key, drop_ratio, causal=causal)
self.wq = nn.Linear(d_key, d_key, bias=False)
self.wk = nn.Linear(d_key, d_key, bias=False)
self.wv = nn.Linear(d_value, d_value, bias=False)
self.wo = nn.Linear(d_value, d_key, bias=False)
self.n_heads = n_heads
def forward(self, query, key, value):
query, key, value = self.wq(query), self.wk(key), self.wv(value)
query, key, value = (x.chunk(self.n_heads, -1) for x in (query, key,
value))
return self.wo(torch.cat([self.attention(q, k, v) for q, k, v in
zip(query, key, value)], -1))
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4]), torch.rand([4, 4, 4])
]
def get_init_inputs():
return [[], {'d_key': 4, 'd_value': 4, 'n_heads': 4, 'drop_ratio': 0.5}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import math
from torch.nn import functional as F
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp3 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = 0.5
tmp16 = tmp14 * tmp15
tmp17 = tl_math.exp(tmp16)
tl.store(out_ptr0 + x2, tmp17, xmask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_cat_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = x0
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + x1, tmp4 & xmask, eviction_policy='evict_last',
other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + x1, tmp9 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + x1, tmp14 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tl.full([1], 4, tl.int64)
tmp19 = tl.load(in_ptr3 + x1, tmp16 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + x2, tmp22, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_4, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_6, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
0), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__softmax_0[grid(64)](buf3, buf4, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf5 = buf3
del buf3
triton_poi_fused__softmax_1[grid(64)](buf4, buf5, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf6 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf5, reinterpret_tensor(buf2, (4, 4, 1), (16, 4,
1), 0), out=buf6)
buf7 = buf4
del buf4
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
1), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 1), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_0[grid(64)](buf7, buf8, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf9 = buf7
del buf7
triton_poi_fused__softmax_1[grid(64)](buf8, buf9, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf10 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf9, reinterpret_tensor(buf2, (4, 4, 1), (16, 4,
1), 1), out=buf10)
buf11 = buf8
del buf8
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
2), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 2), out=buf11)
buf12 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_0[grid(64)](buf11, buf12, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf13 = buf11
del buf11
triton_poi_fused__softmax_1[grid(64)](buf12, buf13, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf13, reinterpret_tensor(buf2, (4, 4, 1), (16,
4, 1), 2), out=buf14)
buf15 = buf12
del buf12
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
3), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 3), out=buf15)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_0[grid(64)](buf15, buf16, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf17 = buf15
del buf15
triton_poi_fused__softmax_1[grid(64)](buf16, buf17, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf18 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf17, reinterpret_tensor(buf2, (4, 4, 1), (16,
4, 1), 3), out=buf18)
buf19 = buf16
del buf16
triton_poi_fused_cat_2[grid(64)](buf6, buf10, buf14, buf18, buf19,
64, XBLOCK=64, num_warps=1, num_stages=1)
del buf10
del buf14
del buf18
del buf6
buf20 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf19, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf20)
return reinterpret_tensor(buf20, (4, 4, 4), (16, 4, 1), 0
), reinterpret_tensor(primals_2, (16, 4), (4, 1), 0
), reinterpret_tensor(primals_4, (16, 4), (4, 1), 0
), reinterpret_tensor(primals_6, (16, 4), (4, 1), 0
), buf5, buf9, buf13, buf17, reinterpret_tensor(buf19, (16, 4), (4,
1), 0), primals_7, reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 3
), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 3
), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 3
), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 2
), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 2
), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 2
), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 1
), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 1
), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 1
), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 0
), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 0
), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 0)
def matmul(x, y):
if x.dim() == y.dim():
return torch.matmul(x, y)
if x.dim() == y.dim() - 1:
return torch.matmul(x.unsqueeze(-2), y).squeeze(-2)
return torch.matmul(x, y.unsqueeze(-2)).squeeze(-2)
class Attention(nn.Module):
def __init__(self, d_key, drop_ratio, causal):
super(Attention, self).__init__()
self.scale = math.sqrt(d_key)
self.dropout = nn.Dropout(drop_ratio)
self.causal = causal
def forward(self, query, key, value):
dot_products = matmul(query, key.transpose(1, 2))
if query.dim() == 3 and (self is None or self.causal):
tri = torch.ones(key.size(1), key.size(1)).triu(1) * INF
if key.is_cuda:
tri = tri
dot_products.data.sub_(tri.unsqueeze(0))
return matmul(self.dropout(F.softmax(dot_products / self.scale, dim
=-1)), value)
class MultiHeadNew(nn.Module):
def __init__(self, d_key, d_value, n_heads, drop_ratio, causal=False):
super(MultiHeadNew, self).__init__()
self.attention = Attention(d_key, drop_ratio, causal=causal)
self.wq = nn.Linear(d_key, d_key, bias=False)
self.wk = nn.Linear(d_key, d_key, bias=False)
self.wv = nn.Linear(d_value, d_value, bias=False)
self.wo = nn.Linear(d_value, d_key, bias=False)
self.n_heads = n_heads
def forward(self, input_0, input_1, input_2):
primals_1 = self.wq.weight
primals_3 = self.wk.weight
primals_5 = self.wv.weight
primals_7 = self.wo.weight
primals_2 = input_0
primals_4 = input_1
primals_6 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
|
TheShadow29/vognet-pytorch
|
MultiHead
| false | 14,482 |
[
"MIT"
] | 70 |
238e93c37cf9f03a2fd376a14760bb3d334a113d
|
https://github.com/TheShadow29/vognet-pytorch/tree/238e93c37cf9f03a2fd376a14760bb3d334a113d
|
Value
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ix/cixxyusyg44s2hkoufcgbrv3ix5ookwqjl4ia3xkv7bdqi4yrzus.py
# Topologically Sorted Source Nodes: [v], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# v => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le_1 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32768],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 25600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 400
x2 = xindex % 1600
x3 = (xindex // 1600)
tmp0 = tl.load(in_out_ptr0 + (x4), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x4), tmp4, xmask)
tl.store(out_ptr0 + (x2 + (1664*x3)), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/op/coptu6xep3awc4lajb4xivopppqmjtx3zy7ebtazm45rqvyeknds.py
# Topologically Sorted Source Nodes: [v_1], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# v_1 => relu_1
# Graph fragment:
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_3,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_1, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_1 = async_compile.triton('triton_poi_fused_relu_threshold_backward_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32768],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i1', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 19200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 300
x2 = (xindex // 1200)
x3 = xindex % 1200
tmp0 = tl.load(in_ptr0 + (x4), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x3 + (1216*x2)), tmp4, xmask)
tl.store(out_ptr1 + (x3 + (1280*x2)), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/as/casrc7bf7ghsendgi7tkqxk3hj4ic6aqb4rmkxzuk5dhbidznia7.py
# Topologically Sorted Source Nodes: [v_1, v_2], Original ATen: [aten.relu, aten.view]
# Source node to ATen node mapping:
# v_1 => relu_1
# v_2 => view_4
# Graph fragment:
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_3,), kwargs = {})
# %view_4 : [num_users=2] = call_function[target=torch.ops.aten.reshape.default](args = (%relu_1, [64, 300]), kwargs = {})
triton_poi_fused_relu_view_2 = async_compile.triton('triton_poi_fused_relu_view_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32768],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_view_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_view_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 19200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 300
x1 = (xindex // 300)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (300*(x1 % 4)) + (1216*(x1 // 4))), xmask)
tl.store(out_ptr0 + (x2), tmp0, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (400, 4), (4, 1))
assert_size_stride(primals_2, (400, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (300, 400), (400, 1))
assert_size_stride(primals_5, (300, ), (1, ))
assert_size_stride(primals_6, (1, 300), (300, 1))
assert_size_stride(primals_7, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 400), (400, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 400), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 400), (6400, 1600, 400, 1), 0); del buf0 # reuse
buf8 = empty_strided_cuda((4, 4, 4, 400), (6656, 1664, 400, 1), torch.bool)
# Topologically Sorted Source Nodes: [v], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf8, 25600, grid=grid(25600), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 300), (300, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf1, (64, 400), (400, 1), 0), reinterpret_tensor(primals_4, (400, 300), (1, 400), 0), out=buf2)
buf3 = empty_strided_cuda((4, 4, 4, 300), (4864, 1216, 300, 1), torch.float32)
buf7 = empty_strided_cuda((4, 4, 4, 300), (5120, 1280, 300, 1), torch.bool)
# Topologically Sorted Source Nodes: [v_1], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_1.run(buf2, primals_5, buf3, buf7, 19200, grid=grid(19200), stream=stream0)
del primals_5
buf4 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [v_1, v_2], Original ATen: [aten.relu, aten.view]
triton_poi_fused_relu_view_2.run(buf3, buf4, 19200, grid=grid(19200), stream=stream0)
del buf3
buf6 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [v_2], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_7, buf4, reinterpret_tensor(primals_6, (300, 1), (1, 300), 0), alpha=1, beta=1, out=buf6)
del primals_7
return (reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf1, (64, 400), (400, 1), 0), buf4, primals_6, buf7, primals_4, buf8, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((400, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((400, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((300, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((300, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((1, 300), (300, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.data
class Value(nn.Module):
def __init__(self, state_dim, action_dim):
super(Value, self).__init__()
self.l1 = nn.Linear(state_dim, 400)
self.l2 = nn.Linear(400, 300)
self.l3 = nn.Linear(300, 1)
def forward(self, state):
v = F.relu(self.l1(state))
v = F.relu(self.l2(v))
v = self.l3(v)
return v
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'state_dim': 4, 'action_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 25600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 400
x2 = xindex % 1600
x3 = xindex // 1600
tmp0 = tl.load(in_out_ptr0 + x4, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x4, tmp4, xmask)
tl.store(out_ptr0 + (x2 + 1664 * x3), tmp6, xmask)
@triton.jit
def triton_poi_fused_relu_threshold_backward_1(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 19200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 300
x2 = xindex // 1200
x3 = xindex % 1200
tmp0 = tl.load(in_ptr0 + x4, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x3 + 1216 * x2), tmp4, xmask)
tl.store(out_ptr1 + (x3 + 1280 * x2), tmp6, xmask)
@triton.jit
def triton_poi_fused_relu_view_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 19200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 300
x1 = xindex // 300
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 300 * (x1 % 4) + 1216 * (x1 // 4)), xmask)
tl.store(out_ptr0 + x2, tmp0, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (400, 4), (4, 1))
assert_size_stride(primals_2, (400,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (300, 400), (400, 1))
assert_size_stride(primals_5, (300,), (1,))
assert_size_stride(primals_6, (1, 300), (300, 1))
assert_size_stride(primals_7, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 400), (400, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 400), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 400), (6400, 1600, 400, 1), 0
)
del buf0
buf8 = empty_strided_cuda((4, 4, 4, 400), (6656, 1664, 400, 1),
torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(25600)](buf1,
primals_2, buf8, 25600, XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 300), (300, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf1, (64, 400), (400, 1), 0),
reinterpret_tensor(primals_4, (400, 300), (1, 400), 0), out=buf2)
buf3 = empty_strided_cuda((4, 4, 4, 300), (4864, 1216, 300, 1),
torch.float32)
buf7 = empty_strided_cuda((4, 4, 4, 300), (5120, 1280, 300, 1),
torch.bool)
triton_poi_fused_relu_threshold_backward_1[grid(19200)](buf2,
primals_5, buf3, buf7, 19200, XBLOCK=128, num_warps=4, num_stages=1
)
del primals_5
buf4 = buf2
del buf2
triton_poi_fused_relu_view_2[grid(19200)](buf3, buf4, 19200, XBLOCK
=256, num_warps=4, num_stages=1)
del buf3
buf6 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
extern_kernels.addmm(primals_7, buf4, reinterpret_tensor(primals_6,
(300, 1), (1, 300), 0), alpha=1, beta=1, out=buf6)
del primals_7
return reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), reinterpret_tensor(buf1, (64, 400), (400, 1), 0
), buf4, primals_6, buf7, primals_4, buf8
class ValueNew(nn.Module):
def __init__(self, state_dim, action_dim):
super(ValueNew, self).__init__()
self.l1 = nn.Linear(state_dim, 400)
self.l2 = nn.Linear(400, 300)
self.l3 = nn.Linear(300, 1)
def forward(self, input_0):
primals_1 = self.l1.weight
primals_2 = self.l1.bias
primals_4 = self.l2.weight
primals_5 = self.l2.bias
primals_6 = self.l3.weight
primals_7 = self.l3.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
|
Thibaud-Ardoin/d4rl_evaluations
|
Value
| false | 14,483 |
[
"Apache-2.0"
] | 123 |
135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
https://github.com/Thibaud-Ardoin/d4rl_evaluations/tree/135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
LayerNorm
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/j2/cj2qhnxmzdsgco22jtn4xpfh2ve5irgwuq3b2hzyjdh3jjjjqxpb.py
# Topologically Sorted Source Nodes: [mean, std, sub, add, output, output_1], Original ATen: [aten.mean, aten.std, aten.sub, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# mean => mean
# output => div
# output_1 => add_1
# std => sqrt, var
# sub => sub
# Graph fragment:
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [-1], True), kwargs = {})
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%primals_1, [-1]), kwargs = {correction: 1.0, keepdim: True})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%var,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %mean), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sqrt, 1e-06), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %add), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%div, %primals_2), kwargs = {})
triton_poi_fused_add_div_mean_std_sub_0 = async_compile.triton('triton_poi_fused_add_div_mean_std_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mean_std_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mean_std_sub_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = 4.0
tmp9 = tmp7 / tmp8
tmp10 = tmp0 - tmp9
tmp11 = tmp1 - tmp9
tmp12 = tmp11 * tmp11
tmp13 = tmp2 - tmp9
tmp14 = tmp13 * tmp13
tmp15 = tmp12 + tmp14
tmp16 = tmp4 - tmp9
tmp17 = tmp16 * tmp16
tmp18 = tmp15 + tmp17
tmp19 = tmp6 - tmp9
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = 3.0
tmp23 = tmp21 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-06
tmp26 = tmp24 + tmp25
tmp27 = tmp10 / tmp26
tmp29 = tmp27 + tmp28
tl.store(out_ptr0 + (x2), tmp29, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mean, std, sub, add, output, output_1], Original ATen: [aten.mean, aten.std, aten.sub, aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mean_std_sub_0.run(primals_1, primals_2, buf0, 256, grid=grid(256), stream=stream0)
del primals_1
del primals_2
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.utils.data
class LayerNorm(nn.Module):
"""
Simple 1D LayerNorm.
"""
def __init__(self, features, center=True, scale=False, eps=1e-06):
super().__init__()
self.center = center
self.scale = scale
self.eps = eps
if self.scale:
self.scale_param = nn.Parameter(torch.ones(features))
else:
self.scale_param = None
if self.center:
self.center_param = nn.Parameter(torch.zeros(features))
else:
self.center_param = None
def forward(self, x):
mean = x.mean(-1, keepdim=True)
std = x.std(-1, keepdim=True)
output = (x - mean) / (std + self.eps)
if self.scale:
output = output * self.scale_param
if self.center:
output = output + self.center_param
return output
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'features': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_mean_std_sub_0(in_ptr0, in_ptr1, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = 4.0
tmp9 = tmp7 / tmp8
tmp10 = tmp0 - tmp9
tmp11 = tmp1 - tmp9
tmp12 = tmp11 * tmp11
tmp13 = tmp2 - tmp9
tmp14 = tmp13 * tmp13
tmp15 = tmp12 + tmp14
tmp16 = tmp4 - tmp9
tmp17 = tmp16 * tmp16
tmp18 = tmp15 + tmp17
tmp19 = tmp6 - tmp9
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = 3.0
tmp23 = tmp21 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-06
tmp26 = tmp24 + tmp25
tmp27 = tmp10 / tmp26
tmp29 = tmp27 + tmp28
tl.store(out_ptr0 + x2, tmp29, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_mean_std_sub_0[grid(256)](primals_1,
primals_2, buf0, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_1
del primals_2
return buf0,
class LayerNormNew(nn.Module):
"""
Simple 1D LayerNorm.
"""
def __init__(self, features, center=True, scale=False, eps=1e-06):
super().__init__()
self.center = center
self.scale = scale
self.eps = eps
if self.scale:
self.scale_param = nn.Parameter(torch.ones(features))
else:
self.scale_param = None
if self.center:
self.center_param = nn.Parameter(torch.zeros(features))
else:
self.center_param = None
def forward(self, input_0):
primals_2 = self.center_param
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0]
|
Thibaud-Ardoin/d4rl_evaluations
|
LayerNorm
| false | 14,484 |
[
"Apache-2.0"
] | 123 |
135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
https://github.com/Thibaud-Ardoin/d4rl_evaluations/tree/135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
Downsample
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/cu/ccutvo2v4333pq6xhrg2zryqqwthm7dmmuqprvva2xdwiodpz5jn.py
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv2d => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [2, 2], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(2, 2), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 2, 2), (16, 4, 2, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(buf1, primals_2, 64, grid=grid(64), stream=stream0)
del primals_2
return (buf1, primals_1, primals_3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch
import torch.nn as nn
class Downsample(nn.Module):
def __init__(self, dim):
super().__init__()
self.conv = nn.Conv2d(dim, dim, 3, 2, 1)
def forward(self, x):
return self.conv(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.utils.data
import torch
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(2,
2), padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 2, 2), (16, 4, 2, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(64)](buf1, primals_2, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_2
return buf1, primals_1, primals_3
class DownsampleNew(nn.Module):
def __init__(self, dim):
super().__init__()
self.conv = nn.Conv2d(dim, dim, 3, 2, 1)
def forward(self, input_0):
primals_1 = self.conv.weight
primals_2 = self.conv.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Tiamat-Tech/Image-Super-Resolution-via-Iterative-Refinement
|
Downsample
| false | 14,485 |
[
"Apache-2.0"
] | 1,764 |
ef9b943b573328d7a5ddb1a0c2abd168b91610dc
|
https://github.com/Tiamat-Tech/Image-Super-Resolution-via-Iterative-Refinement/tree/ef9b943b573328d7a5ddb1a0c2abd168b91610dc
|
FusedLeakyReLU
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/cx/ccxnongcgusvhvf5whrpjbz4ddnlsjovjzjngbpjfxvhdw7ihrhu.py
# Topologically Sorted Source Nodes: [add, leaky_relu, out], Original ATen: [aten.add, aten.leaky_relu, aten.mul]
# Source node to ATen node mapping:
# add => add
# leaky_relu => gt, mul, where
# out => mul_1
# Graph fragment:
# %add : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %primals_1), kwargs = {})
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%add, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add, 0.2), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %add, %mul), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%where, 1.4142135623730951), kwargs = {})
triton_poi_fused_add_leaky_relu_mul_0 = async_compile.triton('triton_poi_fused_add_leaky_relu_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_leaky_relu_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_leaky_relu_mul_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 4
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = 1.4142135623730951
tmp9 = tmp7 * tmp8
tl.store(out_ptr0 + (x3), tmp4, xmask)
tl.store(out_ptr1 + (x3), tmp9, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add, leaky_relu, out], Original ATen: [aten.add, aten.leaky_relu, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_add_leaky_relu_mul_0.run(primals_2, primals_1, buf0, buf1, 256, grid=grid(256), stream=stream0)
del primals_1
del primals_2
return (buf1, buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch
import torch.nn as nn
import torch.nn.functional as F
def fused_leaky_relu(input, bias, negative_slope=0.2, scale=2 ** 0.5):
return F.leaky_relu(input + bias, negative_slope) * scale
class FusedLeakyReLU(nn.Module):
def __init__(self, channel, negative_slope=0.2, scale=2 ** 0.5):
super().__init__()
self.bias = nn.Parameter(torch.zeros(1, channel, 1, 1))
self.negative_slope = negative_slope
self.scale = scale
def forward(self, input):
out = fused_leaky_relu(input, self.bias, self.negative_slope, self.
scale)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channel': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.utils.data
import torch
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_leaky_relu_mul_0(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 4
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = 1.4142135623730951
tmp9 = tmp7 * tmp8
tl.store(out_ptr0 + x3, tmp4, xmask)
tl.store(out_ptr1 + x3, tmp9, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_leaky_relu_mul_0[grid(256)](primals_2,
primals_1, buf0, buf1, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_1
del primals_2
return buf1, buf0
def fused_leaky_relu(input, bias, negative_slope=0.2, scale=2 ** 0.5):
return F.leaky_relu(input + bias, negative_slope) * scale
class FusedLeakyReLUNew(nn.Module):
def __init__(self, channel, negative_slope=0.2, scale=2 ** 0.5):
super().__init__()
self.bias = nn.Parameter(torch.zeros(1, channel, 1, 1))
self.negative_slope = negative_slope
self.scale = scale
def forward(self, input_0):
primals_1 = self.bias
primals_2 = input_0
output = call([primals_1, primals_2])
return output[0]
|
Theomat/colorization-av-enseirb-2020
|
FusedLeakyReLU
| false | 14,486 |
[
"Apache-2.0"
] | 1,422 |
c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
https://github.com/Theomat/colorization-av-enseirb-2020/tree/c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
GetGradient
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/y3/cy3cdceq5u5k3edzfeaaractmgt3cya7tx7cb74n6gqapabbgfzk.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# x => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%sqrt, %sqrt_1, %sqrt_2], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 16) % 3
x0 = xindex % 16
x2 = (xindex // 48)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (16*x2)), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp5 * tmp5
tmp7 = tl.load(in_ptr1 + (x0 + (16*x2)), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp8 = tmp7 * tmp7
tmp9 = tmp6 + tmp8
tmp10 = 1e-06
tmp11 = tmp9 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = tl.full(tmp12.shape, 0.0, tmp12.dtype)
tmp14 = tl.where(tmp4, tmp12, tmp13)
tmp15 = tmp0 >= tmp3
tmp16 = tl.full([1], 2, tl.int64)
tmp17 = tmp0 < tmp16
tmp18 = tmp15 & tmp17
tmp19 = tl.load(in_ptr2 + (x0 + (16*x2)), tmp18 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tmp19 * tmp19
tmp21 = tl.load(in_ptr3 + (x0 + (16*x2)), tmp18 & xmask, eviction_policy='evict_last', other=0.0)
tmp22 = tmp21 * tmp21
tmp23 = tmp20 + tmp22
tmp24 = tmp23 + tmp10
tmp25 = libdevice.sqrt(tmp24)
tmp26 = tl.full(tmp25.shape, 0.0, tmp25.dtype)
tmp27 = tl.where(tmp18, tmp25, tmp26)
tmp28 = tmp0 >= tmp16
tmp29 = tl.full([1], 3, tl.int64)
tmp30 = tmp0 < tmp29
tmp31 = tl.load(in_ptr4 + (x0 + (16*x2)), tmp28 & xmask, eviction_policy='evict_last', other=0.0)
tmp32 = tmp31 * tmp31
tmp33 = tl.load(in_ptr5 + (x0 + (16*x2)), tmp28 & xmask, eviction_policy='evict_last', other=0.0)
tmp34 = tmp33 * tmp33
tmp35 = tmp32 + tmp34
tmp36 = tmp35 + tmp10
tmp37 = libdevice.sqrt(tmp36)
tmp38 = tl.full(tmp37.shape, 0.0, tmp37.dtype)
tmp39 = tl.where(tmp28, tmp37, tmp38)
tmp40 = tl.where(tmp18, tmp27, tmp39)
tmp41 = tl.where(tmp4, tmp14, tmp40)
tl.store(out_ptr0 + (x3), tmp41, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 3, 3), (9, 9, 3, 1))
assert_size_stride(arg2_1, (1, 1, 3, 3), (9, 9, 3, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x0_v], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 0, 4, 1), 0), arg1_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 4, 4), (16, 16, 4, 1))
# Topologically Sorted Source Nodes: [x0_h], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 0, 4, 1), 0), arg2_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 4, 4), (16, 16, 4, 1))
# Topologically Sorted Source Nodes: [x1_v], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 0, 4, 1), 16), arg1_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 4, 4), (16, 16, 4, 1))
# Topologically Sorted Source Nodes: [x1_h], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 0, 4, 1), 16), arg2_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 4, 4), (16, 16, 4, 1))
# Topologically Sorted Source Nodes: [x2_v], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 0, 4, 1), 32), arg1_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 1, 4, 4), (16, 16, 4, 1))
del arg1_1
# Topologically Sorted Source Nodes: [x2_h], Original ATen: [aten.convolution]
buf5 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 0, 4, 1), 32), arg2_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf5, (4, 1, 4, 4), (16, 16, 4, 1))
del arg0_1
del arg2_1
buf6 = empty_strided_cuda((4, 3, 4, 4), (48, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(buf0, buf1, buf2, buf3, buf4, buf5, buf6, 192, grid=grid(192), stream=stream0)
del buf0
del buf1
del buf2
del buf3
del buf4
del buf5
return (buf6, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((1, 1, 3, 3), (9, 9, 3, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((1, 1, 3, 3), (9, 9, 3, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
class GetGradient(nn.Module):
""" generate the gradient map
"""
def __init__(self):
super(GetGradient, self).__init__()
kernel_v = [[0, -1, 0], [0, 0, 0], [0, 1, 0]]
kernel_h = [[0, 0, 0], [-1, 0, 1], [0, 0, 0]]
kernel_h = torch.FloatTensor(kernel_h).unsqueeze(0).unsqueeze(0)
kernel_v = torch.FloatTensor(kernel_v).unsqueeze(0).unsqueeze(0)
self.weight_h = nn.Parameter(data=kernel_h, requires_grad=False)
self.weight_v = nn.Parameter(data=kernel_v, requires_grad=False)
def forward(self, x):
x0 = x[:, 0]
x1 = x[:, 1]
x2 = x[:, 2]
x0_v = F.conv2d(x0.unsqueeze(1), self.weight_v, padding=1)
x0_h = F.conv2d(x0.unsqueeze(1), self.weight_h, padding=1)
x1_v = F.conv2d(x1.unsqueeze(1), self.weight_v, padding=1)
x1_h = F.conv2d(x1.unsqueeze(1), self.weight_h, padding=1)
x2_v = F.conv2d(x2.unsqueeze(1), self.weight_v, padding=1)
x2_h = F.conv2d(x2.unsqueeze(1), self.weight_h, padding=1)
x0 = torch.sqrt(torch.pow(x0_v, 2) + torch.pow(x0_h, 2) + 1e-06)
x1 = torch.sqrt(torch.pow(x1_v, 2) + torch.pow(x1_h, 2) + 1e-06)
x2 = torch.sqrt(torch.pow(x2_v, 2) + torch.pow(x2_h, 2) + 1e-06)
x = torch.cat([x0, x1, x2], dim=1)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4,
in_ptr5, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 16 % 3
x0 = xindex % 16
x2 = xindex // 48
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 16 * x2), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp5 * tmp5
tmp7 = tl.load(in_ptr1 + (x0 + 16 * x2), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp8 = tmp7 * tmp7
tmp9 = tmp6 + tmp8
tmp10 = 1e-06
tmp11 = tmp9 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = tl.full(tmp12.shape, 0.0, tmp12.dtype)
tmp14 = tl.where(tmp4, tmp12, tmp13)
tmp15 = tmp0 >= tmp3
tmp16 = tl.full([1], 2, tl.int64)
tmp17 = tmp0 < tmp16
tmp18 = tmp15 & tmp17
tmp19 = tl.load(in_ptr2 + (x0 + 16 * x2), tmp18 & xmask,
eviction_policy='evict_last', other=0.0)
tmp20 = tmp19 * tmp19
tmp21 = tl.load(in_ptr3 + (x0 + 16 * x2), tmp18 & xmask,
eviction_policy='evict_last', other=0.0)
tmp22 = tmp21 * tmp21
tmp23 = tmp20 + tmp22
tmp24 = tmp23 + tmp10
tmp25 = libdevice.sqrt(tmp24)
tmp26 = tl.full(tmp25.shape, 0.0, tmp25.dtype)
tmp27 = tl.where(tmp18, tmp25, tmp26)
tmp28 = tmp0 >= tmp16
tl.full([1], 3, tl.int64)
tmp31 = tl.load(in_ptr4 + (x0 + 16 * x2), tmp28 & xmask,
eviction_policy='evict_last', other=0.0)
tmp32 = tmp31 * tmp31
tmp33 = tl.load(in_ptr5 + (x0 + 16 * x2), tmp28 & xmask,
eviction_policy='evict_last', other=0.0)
tmp34 = tmp33 * tmp33
tmp35 = tmp32 + tmp34
tmp36 = tmp35 + tmp10
tmp37 = libdevice.sqrt(tmp36)
tmp38 = tl.full(tmp37.shape, 0.0, tmp37.dtype)
tmp39 = tl.where(tmp28, tmp37, tmp38)
tmp40 = tl.where(tmp18, tmp27, tmp39)
tmp41 = tl.where(tmp4, tmp14, tmp40)
tl.store(out_ptr0 + x3, tmp41, xmask)
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 3, 3), (9, 9, 3, 1))
assert_size_stride(arg2_1, (1, 1, 3, 3), (9, 9, 3, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 0, 4, 1), 0), arg1_1, stride=(1, 1), padding=(1, 1),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 4, 4), (16, 16, 4, 1))
buf1 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 0, 4, 1), 0), arg2_1, stride=(1, 1), padding=(1, 1),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 4, 4), (16, 16, 4, 1))
buf2 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 0, 4, 1), 16), arg1_1, stride=(1, 1), padding=(1, 1
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 4, 4), (16, 16, 4, 1))
buf3 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 0, 4, 1), 16), arg2_1, stride=(1, 1), padding=(1, 1
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 4, 4), (16, 16, 4, 1))
buf4 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 0, 4, 1), 32), arg1_1, stride=(1, 1), padding=(1, 1
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf4, (4, 1, 4, 4), (16, 16, 4, 1))
del arg1_1
buf5 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 0, 4, 1), 32), arg2_1, stride=(1, 1), padding=(1, 1
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf5, (4, 1, 4, 4), (16, 16, 4, 1))
del arg0_1
del arg2_1
buf6 = empty_strided_cuda((4, 3, 4, 4), (48, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(192)](buf0, buf1, buf2, buf3, buf4,
buf5, buf6, 192, XBLOCK=128, num_warps=4, num_stages=1)
del buf0
del buf1
del buf2
del buf3
del buf4
del buf5
return buf6,
class GetGradientNew(nn.Module):
""" generate the gradient map
"""
def __init__(self):
super(GetGradientNew, self).__init__()
kernel_v = [[0, -1, 0], [0, 0, 0], [0, 1, 0]]
kernel_h = [[0, 0, 0], [-1, 0, 1], [0, 0, 0]]
kernel_h = torch.FloatTensor(kernel_h).unsqueeze(0).unsqueeze(0)
kernel_v = torch.FloatTensor(kernel_v).unsqueeze(0).unsqueeze(0)
self.weight_h = nn.Parameter(data=kernel_h, requires_grad=False)
self.weight_v = nn.Parameter(data=kernel_v, requires_grad=False)
def forward(self, input_0):
arg1_1 = self.weight_h
arg2_1 = self.weight_v
arg0_1 = input_0
output = call([arg0_1, arg1_1, arg2_1])
return output[0]
|
TencentARC/FAIG
|
GetGradient
| false | 14,487 |
[
"Apache-2.0"
] | 74 |
14f856a87e3696953304029532e2f84997d12278
|
https://github.com/TencentARC/FAIG/tree/14f856a87e3696953304029532e2f84997d12278
|
ToRGB
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/wi/cwiyl3lwwtancorrifw77xt3aqb4lermdintht45zvkj3bg54nbl.py
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_3, 0.5), kwargs = {})
triton_poi_fused_mul_0 = async_compile.triton('triton_poi_fused_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/2o/c2oqkq7zaubqmw7vuixxlseb2ff5jzqqbyczicxlmsahuxwdpdyp.py
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul_1 => mul_1
# Graph fragment:
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_4, 1), kwargs = {})
triton_poi_fused_mul_1 = async_compile.triton('triton_poi_fused_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/on/conl6eemb3vyjzkllydlouehrcxphkzifo5kmslz6fgiz6ixsw5h.py
# Topologically Sorted Source Nodes: [mul_2, weight], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul_2 => mul_2
# weight => mul_3
# Graph fragment:
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_5, 0.5), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, %view), kwargs = {})
triton_poi_fused_mul_2 = async_compile.triton('triton_poi_fused_mul_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 48
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex % 12
x0 = xindex % 4
x2 = (xindex // 12)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x3), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (x0 + (4*x2)), xmask, eviction_policy='evict_last')
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + (x4), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/go/cgoav6av4bzem4wmdmkiowlmjpeiubwc67bqu6es4aivwlfpxzhh.py
# Topologically Sorted Source Nodes: [out_3], Original ATen: [aten.add]
# Source node to ATen node mapping:
# out_3 => add
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_3, %primals_6), kwargs = {})
triton_poi_fused_add_3 = async_compile.triton('triton_poi_fused_add_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 3
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, ), (1, ))
assert_size_stride(primals_5, (1, 3, 4, 1, 1), (12, 4, 1, 1, 1))
assert_size_stride(primals_6, (1, 3, 1, 1), (3, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_0.run(primals_3, buf0, 16, grid=grid(16), stream=stream0)
del primals_3
buf1 = empty_strided_cuda((4, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
triton_poi_fused_mul_1.run(primals_4, buf1, 4, grid=grid(4), stream=stream0)
del primals_4
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul_1, out], Original ATen: [aten.mul, aten.addmm]
extern_kernels.addmm(buf1, primals_2, reinterpret_tensor(buf0, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del buf0
del buf1
buf3 = empty_strided_cuda((4, 3, 4, 1, 1), (12, 4, 1, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul_2, weight], Original ATen: [aten.mul]
triton_poi_fused_mul_2.run(primals_5, buf2, buf3, 48, grid=grid(48), stream=stream0)
# Topologically Sorted Source Nodes: [out_1], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1, 16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf3, (12, 4, 1, 1), (4, 1, 0, 0), 0), stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=4, bias=None)
assert_size_stride(buf4, (1, 12, 4, 4), (192, 16, 4, 1))
buf5 = reinterpret_tensor(buf4, (4, 3, 4, 4), (48, 16, 4, 1), 0); del buf4 # reuse
# Topologically Sorted Source Nodes: [out_3], Original ATen: [aten.add]
triton_poi_fused_add_3.run(buf5, primals_6, 192, grid=grid(192), stream=stream0)
del primals_6
return (buf5, primals_2, primals_5, buf2, reinterpret_tensor(buf3, (12, 4, 1, 1), (4, 1, 1, 1), 0), reinterpret_tensor(primals_1, (1, 16, 4, 4), (256, 16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((1, 3, 4, 1, 1), (12, 4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((1, 3, 1, 1), (3, 1, 1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
import torch.utils.data
import torch
import torch.nn as nn
import torch.nn.functional as F
def make_kernel(k):
k = torch.tensor(k, dtype=torch.float32)
if len(k.shape) == 1:
k = k[None, :] * k[:, None]
k /= k.sum()
return k
def upfirdn2d_native(input, kernel, up_x, up_y, down_x, down_y, pad_x0,
pad_x1, pad_y0, pad_y1):
_, minor, in_h, in_w = input.shape
kernel_h, kernel_w = kernel.shape
out = input.view(-1, minor, in_h, 1, in_w, 1)
out = F.pad(out, [0, up_x - 1, 0, 0, 0, up_y - 1, 0, 0])
out = out.view(-1, minor, in_h * up_y, in_w * up_x)
out = F.pad(out, [max(pad_x0, 0), max(pad_x1, 0), max(pad_y0, 0), max(
pad_y1, 0)])
out = out[:, :, max(-pad_y0, 0):out.shape[2] - max(-pad_y1, 0), max(-
pad_x0, 0):out.shape[3] - max(-pad_x1, 0)]
out = out.reshape([-1, 1, in_h * up_y + pad_y0 + pad_y1, in_w * up_x +
pad_x0 + pad_x1])
w = torch.flip(kernel, [0, 1]).view(1, 1, kernel_h, kernel_w)
out = F.conv2d(out, w)
out = out.reshape(-1, minor, in_h * up_y + pad_y0 + pad_y1 - kernel_h +
1, in_w * up_x + pad_x0 + pad_x1 - kernel_w + 1)
return out[:, :, ::down_y, ::down_x]
def upfirdn2d(input, kernel, up=1, down=1, pad=(0, 0)):
return upfirdn2d_native(input, kernel, up, up, down, down, pad[0], pad[
1], pad[0], pad[1])
def fused_leaky_relu(input, bias, negative_slope=0.2, scale=2 ** 0.5):
return F.leaky_relu(input + bias, negative_slope) * scale
class Upsample(nn.Module):
def __init__(self, kernel, factor=2):
super().__init__()
self.factor = factor
kernel = make_kernel(kernel) * factor ** 2
self.register_buffer('kernel', kernel)
p = kernel.shape[0] - factor
pad0 = (p + 1) // 2 + factor - 1
pad1 = p // 2
self.pad = pad0, pad1
def forward(self, input):
out = upfirdn2d(input, self.kernel, up=self.factor, down=1, pad=
self.pad)
return out
class Blur(nn.Module):
def __init__(self, kernel, pad, upsample_factor=1):
super().__init__()
kernel = make_kernel(kernel)
if upsample_factor > 1:
kernel = kernel * upsample_factor ** 2
self.register_buffer('kernel', kernel)
self.pad = pad
def forward(self, input):
out = upfirdn2d(input, self.kernel, pad=self.pad)
return out
class EqualLinear(nn.Module):
def __init__(self, in_dim, out_dim, bias=True, bias_init=0, lr_mul=1,
activation=None):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_dim, in_dim).div_(lr_mul))
if bias:
self.bias = nn.Parameter(torch.zeros(out_dim).fill_(bias_init))
else:
self.bias = None
self.activation = activation
self.scale = math.sqrt(1) / math.sqrt(in_dim) * lr_mul
self.lr_mul = lr_mul
def forward(self, input):
if self.activation:
out = F.linear(input, self.weight * self.scale)
out = fused_leaky_relu(out, self.bias * self.lr_mul)
else:
out = F.linear(input, self.weight * self.scale, bias=self.bias *
self.lr_mul)
return out
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]})'
)
class ModulatedConv2d(nn.Module):
def __init__(self, in_channel, out_channel, kernel_size, style_dim,
demodulate=True, upsample=False, downsample=False, blur_kernel=[1,
3, 3, 1]):
super().__init__()
self.eps = 1e-08
self.kernel_size = kernel_size
self.in_channel = in_channel
self.out_channel = out_channel
self.upsample = upsample
self.downsample = downsample
if upsample:
factor = 2
p = len(blur_kernel) - factor - (kernel_size - 1)
pad0 = (p + 1) // 2 + factor - 1
pad1 = p // 2 + 1
self.blur = Blur(blur_kernel, pad=(pad0, pad1), upsample_factor
=factor)
if downsample:
factor = 2
p = len(blur_kernel) - factor + (kernel_size - 1)
pad0 = (p + 1) // 2
pad1 = p // 2
self.blur = Blur(blur_kernel, pad=(pad0, pad1))
fan_in = in_channel * kernel_size ** 2
self.scale = math.sqrt(1) / math.sqrt(fan_in)
self.padding = kernel_size // 2
self.weight = nn.Parameter(torch.randn(1, out_channel, in_channel,
kernel_size, kernel_size))
if style_dim is not None and style_dim > 0:
self.modulation = EqualLinear(style_dim, in_channel, bias_init=1)
self.demodulate = demodulate
def __repr__(self):
return (
f'{self.__class__.__name__}({self.in_channel}, {self.out_channel}, {self.kernel_size}, upsample={self.upsample}, downsample={self.downsample})'
)
def forward(self, input, style):
batch, in_channel, height, width = input.shape
if style is not None:
style = self.modulation(style).view(batch, 1, in_channel, 1, 1)
else:
style = torch.ones(batch, 1, in_channel, 1, 1)
weight = self.scale * self.weight * style
if self.demodulate:
demod = torch.rsqrt(weight.pow(2).sum([2, 3, 4]) + 1e-08)
weight = weight * demod.view(batch, self.out_channel, 1, 1, 1)
weight = weight.view(batch * self.out_channel, in_channel, self.
kernel_size, self.kernel_size)
if self.upsample:
input = input.view(1, batch * in_channel, height, width)
weight = weight.view(batch, self.out_channel, in_channel, self.
kernel_size, self.kernel_size)
weight = weight.transpose(1, 2).reshape(batch * in_channel,
self.out_channel, self.kernel_size, self.kernel_size)
out = F.conv_transpose2d(input, weight, padding=0, stride=2,
groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
out = self.blur(out)
elif self.downsample:
input = self.blur(input)
_, _, height, width = input.shape
input = input.view(1, batch * in_channel, height, width)
out = F.conv2d(input, weight, padding=0, stride=2, groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
else:
input = input.view(1, batch * in_channel, height, width)
out = F.conv2d(input, weight, padding=self.padding, groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
return out
class ToRGB(nn.Module):
def __init__(self, in_channel, style_dim, upsample=True, blur_kernel=[1,
3, 3, 1]):
super().__init__()
if upsample:
self.upsample = Upsample(blur_kernel)
self.conv = ModulatedConv2d(in_channel, 3, 1, style_dim, demodulate
=False)
self.bias = nn.Parameter(torch.zeros(1, 3, 1, 1))
def forward(self, input, style, skip=None):
out = self.conv(input, style)
out = out + self.bias
if skip is not None:
skip = self.upsample(skip)
out = out + skip
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'in_channel': 4, 'style_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import math
import torch.utils.data
import torch
import torch.nn as nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_poi_fused_mul_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 48
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex % 12
x0 = xindex % 4
x2 = xindex // 12
x4 = xindex
tmp0 = tl.load(in_ptr0 + x3, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (x0 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + x4, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 3
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4,), (1,))
assert_size_stride(primals_5, (1, 3, 4, 1, 1), (12, 4, 1, 1, 1))
assert_size_stride(primals_6, (1, 3, 1, 1), (3, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_0[grid(16)](primals_3, buf0, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del primals_3
buf1 = empty_strided_cuda((4,), (1,), torch.float32)
triton_poi_fused_mul_1[grid(4)](primals_4, buf1, 4, XBLOCK=4,
num_warps=1, num_stages=1)
del primals_4
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(buf1, primals_2, reinterpret_tensor(buf0, (4,
4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del buf0
del buf1
buf3 = empty_strided_cuda((4, 3, 4, 1, 1), (12, 4, 1, 1, 1), torch.
float32)
triton_poi_fused_mul_2[grid(48)](primals_5, buf2, buf3, 48, XBLOCK=
64, num_warps=1, num_stages=1)
buf4 = extern_kernels.convolution(reinterpret_tensor(primals_1, (1,
16, 4, 4), (256, 16, 4, 1), 0), reinterpret_tensor(buf3, (12, 4,
1, 1), (4, 1, 0, 0), 0), stride=(1, 1), padding=(0, 0),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=4, bias=None)
assert_size_stride(buf4, (1, 12, 4, 4), (192, 16, 4, 1))
buf5 = reinterpret_tensor(buf4, (4, 3, 4, 4), (48, 16, 4, 1), 0)
del buf4
triton_poi_fused_add_3[grid(192)](buf5, primals_6, 192, XBLOCK=256,
num_warps=4, num_stages=1)
del primals_6
return buf5, primals_2, primals_5, buf2, reinterpret_tensor(buf3, (12,
4, 1, 1), (4, 1, 1, 1), 0), reinterpret_tensor(primals_1, (1, 16, 4,
4), (256, 16, 4, 1), 0)
def make_kernel(k):
k = torch.tensor(k, dtype=torch.float32)
if len(k.shape) == 1:
k = k[None, :] * k[:, None]
k /= k.sum()
return k
def upfirdn2d_native(input, kernel, up_x, up_y, down_x, down_y, pad_x0,
pad_x1, pad_y0, pad_y1):
_, minor, in_h, in_w = input.shape
kernel_h, kernel_w = kernel.shape
out = input.view(-1, minor, in_h, 1, in_w, 1)
out = F.pad(out, [0, up_x - 1, 0, 0, 0, up_y - 1, 0, 0])
out = out.view(-1, minor, in_h * up_y, in_w * up_x)
out = F.pad(out, [max(pad_x0, 0), max(pad_x1, 0), max(pad_y0, 0), max(
pad_y1, 0)])
out = out[:, :, max(-pad_y0, 0):out.shape[2] - max(-pad_y1, 0), max(-
pad_x0, 0):out.shape[3] - max(-pad_x1, 0)]
out = out.reshape([-1, 1, in_h * up_y + pad_y0 + pad_y1, in_w * up_x +
pad_x0 + pad_x1])
w = torch.flip(kernel, [0, 1]).view(1, 1, kernel_h, kernel_w)
out = F.conv2d(out, w)
out = out.reshape(-1, minor, in_h * up_y + pad_y0 + pad_y1 - kernel_h +
1, in_w * up_x + pad_x0 + pad_x1 - kernel_w + 1)
return out[:, :, ::down_y, ::down_x]
def upfirdn2d(input, kernel, up=1, down=1, pad=(0, 0)):
return upfirdn2d_native(input, kernel, up, up, down, down, pad[0], pad[
1], pad[0], pad[1])
def fused_leaky_relu(input, bias, negative_slope=0.2, scale=2 ** 0.5):
return F.leaky_relu(input + bias, negative_slope) * scale
class Upsample(nn.Module):
def __init__(self, kernel, factor=2):
super().__init__()
self.factor = factor
kernel = make_kernel(kernel) * factor ** 2
self.register_buffer('kernel', kernel)
p = kernel.shape[0] - factor
pad0 = (p + 1) // 2 + factor - 1
pad1 = p // 2
self.pad = pad0, pad1
def forward(self, input):
out = upfirdn2d(input, self.kernel, up=self.factor, down=1, pad=
self.pad)
return out
class Blur(nn.Module):
def __init__(self, kernel, pad, upsample_factor=1):
super().__init__()
kernel = make_kernel(kernel)
if upsample_factor > 1:
kernel = kernel * upsample_factor ** 2
self.register_buffer('kernel', kernel)
self.pad = pad
def forward(self, input):
out = upfirdn2d(input, self.kernel, pad=self.pad)
return out
class EqualLinear(nn.Module):
def __init__(self, in_dim, out_dim, bias=True, bias_init=0, lr_mul=1,
activation=None):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_dim, in_dim).div_(lr_mul))
if bias:
self.bias = nn.Parameter(torch.zeros(out_dim).fill_(bias_init))
else:
self.bias = None
self.activation = activation
self.scale = math.sqrt(1) / math.sqrt(in_dim) * lr_mul
self.lr_mul = lr_mul
def forward(self, input):
if self.activation:
out = F.linear(input, self.weight * self.scale)
out = fused_leaky_relu(out, self.bias * self.lr_mul)
else:
out = F.linear(input, self.weight * self.scale, bias=self.bias *
self.lr_mul)
return out
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]})'
)
class ModulatedConv2d(nn.Module):
def __init__(self, in_channel, out_channel, kernel_size, style_dim,
demodulate=True, upsample=False, downsample=False, blur_kernel=[1,
3, 3, 1]):
super().__init__()
self.eps = 1e-08
self.kernel_size = kernel_size
self.in_channel = in_channel
self.out_channel = out_channel
self.upsample = upsample
self.downsample = downsample
if upsample:
factor = 2
p = len(blur_kernel) - factor - (kernel_size - 1)
pad0 = (p + 1) // 2 + factor - 1
pad1 = p // 2 + 1
self.blur = Blur(blur_kernel, pad=(pad0, pad1), upsample_factor
=factor)
if downsample:
factor = 2
p = len(blur_kernel) - factor + (kernel_size - 1)
pad0 = (p + 1) // 2
pad1 = p // 2
self.blur = Blur(blur_kernel, pad=(pad0, pad1))
fan_in = in_channel * kernel_size ** 2
self.scale = math.sqrt(1) / math.sqrt(fan_in)
self.padding = kernel_size // 2
self.weight = nn.Parameter(torch.randn(1, out_channel, in_channel,
kernel_size, kernel_size))
if style_dim is not None and style_dim > 0:
self.modulation = EqualLinear(style_dim, in_channel, bias_init=1)
self.demodulate = demodulate
def __repr__(self):
return (
f'{self.__class__.__name__}({self.in_channel}, {self.out_channel}, {self.kernel_size}, upsample={self.upsample}, downsample={self.downsample})'
)
def forward(self, input, style):
batch, in_channel, height, width = input.shape
if style is not None:
style = self.modulation(style).view(batch, 1, in_channel, 1, 1)
else:
style = torch.ones(batch, 1, in_channel, 1, 1)
weight = self.scale * self.weight * style
if self.demodulate:
demod = torch.rsqrt(weight.pow(2).sum([2, 3, 4]) + 1e-08)
weight = weight * demod.view(batch, self.out_channel, 1, 1, 1)
weight = weight.view(batch * self.out_channel, in_channel, self.
kernel_size, self.kernel_size)
if self.upsample:
input = input.view(1, batch * in_channel, height, width)
weight = weight.view(batch, self.out_channel, in_channel, self.
kernel_size, self.kernel_size)
weight = weight.transpose(1, 2).reshape(batch * in_channel,
self.out_channel, self.kernel_size, self.kernel_size)
out = F.conv_transpose2d(input, weight, padding=0, stride=2,
groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
out = self.blur(out)
elif self.downsample:
input = self.blur(input)
_, _, height, width = input.shape
input = input.view(1, batch * in_channel, height, width)
out = F.conv2d(input, weight, padding=0, stride=2, groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
else:
input = input.view(1, batch * in_channel, height, width)
out = F.conv2d(input, weight, padding=self.padding, groups=batch)
_, _, height, width = out.shape
out = out.view(batch, self.out_channel, height, width)
return out
class ToRGBNew(nn.Module):
def __init__(self, in_channel, style_dim, upsample=True, blur_kernel=[1,
3, 3, 1]):
super().__init__()
if upsample:
self.upsample = Upsample(blur_kernel)
self.conv = ModulatedConv2d(in_channel, 3, 1, style_dim, demodulate
=False)
self.bias = nn.Parameter(torch.zeros(1, 3, 1, 1))
def forward(self, input_0, input_1):
primals_6 = self.bias
primals_5 = self.conv.weight
primals_2 = self.conv.modulation.weight
primals_4 = self.conv.modulation.bias
primals_1 = input_0
primals_3 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6])
return output[0]
|
Theomat/colorization-av-enseirb-2020
|
ToRGB
| false | 14,488 |
[
"Apache-2.0"
] | 1,422 |
c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
https://github.com/Theomat/colorization-av-enseirb-2020/tree/c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
SpatialTemporalConv3D
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ly/clywzgcyp4q3mqwgp7xvmujn3q36yc542g7f53gdhoijjawxolig.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
# Source node to ATen node mapping:
# x_1 => gt, mul, where
# Graph fragment:
# %gt : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze, 0.01), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %squeeze, %mul), kwargs = {})
# %gt_7 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_3, 0), kwargs = {})
triton_poi_fused_leaky_relu_leaky_relu_backward_0 = async_compile.triton('triton_poi_fused_leaky_relu_leaky_relu_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_leaky_relu_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4096
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x1 = (xindex // 64)
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + (x2), tmp7, None)
tl.store(out_ptr0 + (x2), tmp8, None)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ut/cutlonknxe4wvkf36xewkosqvtuns4bsolmvucqb3rqlupgobpcw.py
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
# Source node to ATen node mapping:
# x_5 => gt_2, mul_2, where_2
# Graph fragment:
# %gt_2 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_8, 0), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_8, 0.01), kwargs = {})
# %where_2 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_2, %squeeze_8, %mul_2), kwargs = {})
# %gt_5 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_11, 0), kwargs = {})
triton_poi_fused_leaky_relu_leaky_relu_backward_1 = async_compile.triton('triton_poi_fused_leaky_relu_leaky_relu_backward_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_leaky_relu_backward_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_1(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 2048
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x1 = (xindex // 64)
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + (x2), tmp7, None)
tl.store(out_ptr0 + (x2), tmp8, None)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/5t/c5tgeylskhrlo2mjvb7th67lzzjyqz2xzxkgzygebov2ab6hkh76.py
# Topologically Sorted Source Nodes: [x_7], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
# Source node to ATen node mapping:
# x_7 => gt_3, mul_3, where_3
# Graph fragment:
# %gt_3 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_12, 0), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%squeeze_12, 0.01), kwargs = {})
# %where_3 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_3, %squeeze_12, %mul_3), kwargs = {})
# %gt_4 : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%squeeze_17, 0), kwargs = {})
triton_poi_fused_leaky_relu_leaky_relu_backward_2 = async_compile.triton('triton_poi_fused_leaky_relu_leaky_relu_backward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_leaky_relu_backward_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_2(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 64)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + (x2), tmp7, xmask)
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (64, 4, 3, 3, 3), (108, 27, 9, 3, 1))
assert_size_stride(primals_2, (64, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (64, 64, 3, 3, 3), (1728, 27, 9, 3, 1))
assert_size_stride(primals_5, (64, ), (1, ))
assert_size_stride(primals_6, (32, 64, 3, 3, 3), (1728, 27, 9, 3, 1))
assert_size_stride(primals_7, (32, ), (1, ))
assert_size_stride(primals_8, (4, 32, 3, 3, 3), (864, 27, 9, 3, 1))
assert_size_stride(primals_9, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1), 0), primals_1, stride=(1, 1, 1), padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False, output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1))
buf1 = reinterpret_tensor(buf0, (64, 4, 4, 4), (64, 16, 4, 1), 0); del buf0 # reuse
buf11 = empty_strided_cuda((64, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_leaky_relu_backward_0.run(buf1, primals_2, buf11, 4096, grid=grid(4096), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(buf1, (1, 64, 4, 4, 4), (0, 64, 16, 4, 1), 0), primals_4, stride=(1, 1, 1), padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False, output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf2, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1))
buf3 = reinterpret_tensor(buf2, (64, 4, 4, 4), (64, 16, 4, 1), 0); del buf2 # reuse
buf10 = empty_strided_cuda((64, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
triton_poi_fused_leaky_relu_leaky_relu_backward_0.run(buf3, primals_5, buf10, 4096, grid=grid(4096), stream=stream0)
del primals_5
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(reinterpret_tensor(buf3, (1, 64, 4, 4, 4), (0, 64, 16, 4, 1), 0), primals_6, stride=(1, 1, 1), padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False, output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf4, (1, 32, 4, 4, 4), (2048, 64, 16, 4, 1))
buf5 = reinterpret_tensor(buf4, (32, 4, 4, 4), (64, 16, 4, 1), 0); del buf4 # reuse
buf9 = empty_strided_cuda((32, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
triton_poi_fused_leaky_relu_leaky_relu_backward_1.run(buf5, primals_7, buf9, 2048, grid=grid(2048), stream=stream0)
del primals_7
# Topologically Sorted Source Nodes: [x_6], Original ATen: [aten.convolution]
buf6 = extern_kernels.convolution(reinterpret_tensor(buf5, (1, 32, 4, 4, 4), (0, 64, 16, 4, 1), 0), primals_8, stride=(1, 1, 1), padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False, output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf6, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1))
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf6 # reuse
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_7], Original ATen: [aten.leaky_relu, aten.leaky_relu_backward]
triton_poi_fused_leaky_relu_leaky_relu_backward_2.run(buf7, primals_9, buf8, 256, grid=grid(256), stream=stream0)
del primals_9
return (buf7, primals_1, primals_4, primals_6, primals_8, reinterpret_tensor(primals_3, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1), 0), reinterpret_tensor(buf1, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1), 0), reinterpret_tensor(buf3, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1), 0), reinterpret_tensor(buf5, (1, 32, 4, 4, 4), (2048, 64, 16, 4, 1), 0), buf8, buf9, buf10, buf11, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((64, 4, 3, 3, 3), (108, 27, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((64, 64, 3, 3, 3), (1728, 27, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((32, 64, 3, 3, 3), (1728, 27, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, 32, 3, 3, 3), (864, 27, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class SpatialTemporalConv3D(nn.Module):
"""
Apply 3D conv. over an input signal composed of several input planes with distinct spatial and time axes, by performing 3D convolution over the spatiotemporal axes
args:
in_channels (int): number of channels in the input tensor
out_channels (int): number of channels produced by the convolution
kernel_size (int or tuple): size of the convolution kernel
stride (int or tuple): stride
padding (int or tuple): zero-padding
"""
def __init__(self, in_channels, out_channels, kernel_size=3, stride=1,
padding=1):
super(SpatialTemporalConv3D, self).__init__()
self.conv1 = nn.Conv3d(in_channels, 64, kernel_size, stride, padding)
self.relu1 = nn.LeakyReLU(inplace=True)
self.conv2 = nn.Conv3d(64, 64, kernel_size, stride, padding)
self.relu2 = nn.LeakyReLU(inplace=True)
self.conv3 = nn.Conv3d(64, 32, kernel_size, stride, padding)
self.relu3 = nn.LeakyReLU(inplace=True)
self.conv4 = nn.Conv3d(32, out_channels, kernel_size, stride, padding)
self.relu4 = nn.LeakyReLU(inplace=True)
def forward(self, x):
x = self.conv1(x)
x = self.relu1(x)
x = self.conv2(x)
x = self.relu2(x)
x = self.conv3(x)
x = self.relu3(x)
x = self.conv4(x)
x = self.relu4(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x1 = xindex // 64
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + x2, tmp7, None)
tl.store(out_ptr0 + x2, tmp8, None)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_1(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x1 = xindex // 64
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + x2, tmp7, None)
tl.store(out_ptr0 + x2, tmp8, None)
@triton.jit
def triton_poi_fused_leaky_relu_leaky_relu_backward_2(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 64
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tmp8 = tmp7 > tmp3
tl.store(in_out_ptr0 + x2, tmp7, xmask)
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (64, 4, 3, 3, 3), (108, 27, 9, 3, 1))
assert_size_stride(primals_2, (64,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (64, 64, 3, 3, 3), (1728, 27, 9, 3, 1))
assert_size_stride(primals_5, (64,), (1,))
assert_size_stride(primals_6, (32, 64, 3, 3, 3), (1728, 27, 9, 3, 1))
assert_size_stride(primals_7, (32,), (1,))
assert_size_stride(primals_8, (4, 32, 3, 3, 3), (864, 27, 9, 3, 1))
assert_size_stride(primals_9, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_3, (1,
4, 4, 4, 4), (256, 64, 16, 4, 1), 0), primals_1, stride=(1, 1,
1), padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False,
output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf0, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1))
buf1 = reinterpret_tensor(buf0, (64, 4, 4, 4), (64, 16, 4, 1), 0)
del buf0
buf11 = empty_strided_cuda((64, 4, 4, 4), (64, 16, 4, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_leaky_relu_leaky_relu_backward_0[grid(4096)](buf1,
primals_2, buf11, 4096, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(reinterpret_tensor(buf1, (1, 64,
4, 4, 4), (0, 64, 16, 4, 1), 0), primals_4, stride=(1, 1, 1),
padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False,
output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf2, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1))
buf3 = reinterpret_tensor(buf2, (64, 4, 4, 4), (64, 16, 4, 1), 0)
del buf2
buf10 = empty_strided_cuda((64, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_leaky_relu_leaky_relu_backward_0[grid(4096)](buf3,
primals_5, buf10, 4096, XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
buf4 = extern_kernels.convolution(reinterpret_tensor(buf3, (1, 64,
4, 4, 4), (0, 64, 16, 4, 1), 0), primals_6, stride=(1, 1, 1),
padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False,
output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf4, (1, 32, 4, 4, 4), (2048, 64, 16, 4, 1))
buf5 = reinterpret_tensor(buf4, (32, 4, 4, 4), (64, 16, 4, 1), 0)
del buf4
buf9 = empty_strided_cuda((32, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_leaky_relu_leaky_relu_backward_1[grid(2048)](buf5,
primals_7, buf9, 2048, XBLOCK=256, num_warps=4, num_stages=1)
del primals_7
buf6 = extern_kernels.convolution(reinterpret_tensor(buf5, (1, 32,
4, 4, 4), (0, 64, 16, 4, 1), 0), primals_8, stride=(1, 1, 1),
padding=(1, 1, 1), dilation=(1, 1, 1), transposed=False,
output_padding=(0, 0, 0), groups=1, bias=None)
assert_size_stride(buf6, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1))
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf6
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_leaky_relu_leaky_relu_backward_2[grid(256)](buf7,
primals_9, buf8, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_9
return (buf7, primals_1, primals_4, primals_6, primals_8,
reinterpret_tensor(primals_3, (1, 4, 4, 4, 4), (256, 64, 16, 4, 1),
0), reinterpret_tensor(buf1, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1),
0), reinterpret_tensor(buf3, (1, 64, 4, 4, 4), (4096, 64, 16, 4, 1),
0), reinterpret_tensor(buf5, (1, 32, 4, 4, 4), (2048, 64, 16, 4, 1),
0), buf8, buf9, buf10, buf11)
class SpatialTemporalConv3DNew(nn.Module):
"""
Apply 3D conv. over an input signal composed of several input planes with distinct spatial and time axes, by performing 3D convolution over the spatiotemporal axes
args:
in_channels (int): number of channels in the input tensor
out_channels (int): number of channels produced by the convolution
kernel_size (int or tuple): size of the convolution kernel
stride (int or tuple): stride
padding (int or tuple): zero-padding
"""
def __init__(self, in_channels, out_channels, kernel_size=3, stride=1,
padding=1):
super(SpatialTemporalConv3DNew, self).__init__()
self.conv1 = nn.Conv3d(in_channels, 64, kernel_size, stride, padding)
self.relu1 = nn.LeakyReLU(inplace=True)
self.conv2 = nn.Conv3d(64, 64, kernel_size, stride, padding)
self.relu2 = nn.LeakyReLU(inplace=True)
self.conv3 = nn.Conv3d(64, 32, kernel_size, stride, padding)
self.relu3 = nn.LeakyReLU(inplace=True)
self.conv4 = nn.Conv3d(32, out_channels, kernel_size, stride, padding)
self.relu4 = nn.LeakyReLU(inplace=True)
def forward(self, input_0):
primals_1 = self.conv1.weight
primals_2 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_6 = self.conv3.weight
primals_7 = self.conv3.bias
primals_8 = self.conv4.weight
primals_9 = self.conv4.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
|
Tencent/DVQA
|
SpatialTemporalConv3D
| false | 14,489 |
[
"BSD-3-Clause"
] | 408 |
21727333a6b41d54ad1a8beca1fcbe00a69ed347
|
https://github.com/Tencent/DVQA/tree/21727333a6b41d54ad1a8beca1fcbe00a69ed347
|
ReshapeF
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ef/cef2l3ajjzplmrqfzg25ehcs22rwnbbsdt4r6sobviedge5satl2.py
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, add, out], Original ATen: [aten.pow, aten.sum, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# norm => pow_2
# out => div
# pow_1 => pow_1
# sum_1 => sum_1
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%view, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [1], True), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sum_1, 0.5), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%pow_2, 1e-07), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%view, %add), kwargs = {})
triton_poi_fused_add_div_pow_sum_0 = async_compile.triton('triton_poi_fused_add_div_pow_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_pow_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_pow_sum_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + ((16*x1) + (64*(y0 // 16)) + (y0 % 16)), xmask & ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + ((64*(y0 // 16)) + (y0 % 16)), ymask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + (64*(y0 // 16)) + (y0 % 16)), ymask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + (64*(y0 // 16)) + (y0 % 16)), ymask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + (64*(y0 // 16)) + (y0 % 16)), ymask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x1 + (4*y0)), tmp15, xmask & ymask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, add, out], Original ATen: [aten.pow, aten.sum, aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_pow_sum_0.run(arg0_1, buf0, 64, 4, grid=grid(64, 4), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.utils.data
import torch
import torch.nn as nn
class Normalize(nn.Module):
def __init__(self, power=2):
super(Normalize, self).__init__()
self.power = power
def forward(self, x):
norm = x.pow(self.power).sum(1, keepdim=True).pow(1.0 / self.power)
out = x.div(norm + 1e-07)
return out
class ReshapeF(nn.Module):
def __init__(self):
super(ReshapeF, self).__init__()
model = [nn.AdaptiveAvgPool2d(4)]
self.model = nn.Sequential(*model)
self.l2norm = Normalize(2)
def forward(self, x):
x = self.model(x)
x_reshape = x.permute(0, 2, 3, 1).flatten(0, 2)
return self.l2norm(x_reshape)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.utils.data
import torch
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_pow_sum_0(in_ptr0, out_ptr0, ynumel, xnumel,
YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (16 * x1 + 64 * (y0 // 16) + y0 % 16), xmask &
ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (64 * (y0 // 16) + y0 % 16), ymask,
eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + 64 * (y0 // 16) + y0 % 16), ymask,
eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + 64 * (y0 // 16) + y0 % 16), ymask,
eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + 64 * (y0 // 16) + y0 % 16), ymask,
eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x1 + 4 * y0), tmp15, xmask & ymask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_pow_sum_0[grid(64, 4)](arg0_1, buf0, 64, 4,
XBLOCK=4, YBLOCK=32, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class Normalize(nn.Module):
def __init__(self, power=2):
super(Normalize, self).__init__()
self.power = power
def forward(self, x):
norm = x.pow(self.power).sum(1, keepdim=True).pow(1.0 / self.power)
out = x.div(norm + 1e-07)
return out
class ReshapeFNew(nn.Module):
def __init__(self):
super(ReshapeFNew, self).__init__()
model = [nn.AdaptiveAvgPool2d(4)]
self.model = nn.Sequential(*model)
self.l2norm = Normalize(2)
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
Theomat/colorization-av-enseirb-2020
|
ReshapeF
| false | 14,490 |
[
"Apache-2.0"
] | 1,422 |
c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
https://github.com/Theomat/colorization-av-enseirb-2020/tree/c54c2388ea39a62289fa2f1c51b4757bf55d3c4f
|
Swish
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/2u/c2uzxklzsbiemtrcl7via4yz7hqy7l5noylwtycrzffjpd453ape.py
# Topologically Sorted Source Nodes: [mul, sigmoid, mul_1], Original ATen: [aten.mul, aten.sigmoid]
# Source node to ATen node mapping:
# mul => mul
# mul_1 => mul_1
# sigmoid => sigmoid
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %sigmoid : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%mul,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, %sigmoid), kwargs = {})
triton_poi_fused_mul_sigmoid_0 = async_compile.triton('triton_poi_fused_mul_sigmoid_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_sigmoid_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_sigmoid_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp3 = tl.sigmoid(tmp2)
tmp4 = tmp0 * tmp3
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, sigmoid, mul_1], Original ATen: [aten.mul, aten.sigmoid]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_sigmoid_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class Swish(nn.Module):
"""The swish activation function: :math:`\\mathrm{swish}(x)=x\\sigma(\\beta x)=\\frac{x}{1+e^{-\\beta x}}`.
:param beta: The :math:`\\beta` parameter in the swish activation.
:type beta: float
:param trainable: Whether scalar :math:`\\beta` can be trained
:type trainable: bool
"""
def __init__(self, beta=1.0, trainable=False):
super(Swish, self).__init__()
beta = float(beta)
self.trainable = trainable
if trainable:
self.beta = nn.Parameter(torch.tensor(beta))
else:
self.beta = beta
def forward(self, x):
return x * torch.sigmoid(self.beta * x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_mul_sigmoid_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp3 = tl.sigmoid(tmp2)
tmp4 = tmp0 * tmp3
tl.store(out_ptr0 + x0, tmp4, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_sigmoid_0[grid(256)](arg0_1, buf0, 256, XBLOCK
=256, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class SwishNew(nn.Module):
"""The swish activation function: :math:`\\mathrm{swish}(x)=x\\sigma(\\beta x)=\\frac{x}{1+e^{-\\beta x}}`.
:param beta: The :math:`\\beta` parameter in the swish activation.
:type beta: float
:param trainable: Whether scalar :math:`\\beta` can be trained
:type trainable: bool
"""
def __init__(self, beta=1.0, trainable=False):
super(SwishNew, self).__init__()
beta = float(beta)
self.trainable = trainable
if trainable:
self.beta = nn.Parameter(torch.tensor(beta))
else:
self.beta = beta
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
Tiamat-Tech/neurodiffeq
|
Swish
| false | 14,491 |
[
"MIT"
] | 202 |
622827e5b9b65d285ebe36614fbdae68ba07f4dc
|
https://github.com/Tiamat-Tech/neurodiffeq/tree/622827e5b9b65d285ebe36614fbdae68ba07f4dc
|
SelfGating
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/32/c32gpnu7y6kwawwiknabqcyafcipv27fjg22cpx6wzdxmd52bm4o.py
# Topologically Sorted Source Nodes: [spatiotemporal_average], Original ATen: [aten.mean]
# Source node to ATen node mapping:
# spatiotemporal_average => mean
# Graph fragment:
# %mean : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [2, 3, 4]), kwargs = {})
triton_per_fused_mean_0 = async_compile.triton('triton_per_fused_mean_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[16, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mean_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mean_0(in_out_ptr0, in_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 16
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp5 = 64.0
tmp6 = tmp4 / tmp5
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/c2/cc26hd4oq32vrxgrqqakqqnzwxv6d6ovzdl26r33utpsmw442x3p.py
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%unsqueeze_2, %primals_1), kwargs = {})
triton_poi_fused_mul_1 = async_compile.triton('triton_poi_fused_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 64)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr1 + (x2), xmask)
tmp1 = tl.sigmoid(tmp0)
tmp3 = tmp1 * tmp2
tl.store(out_ptr0 + (x2), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [spatiotemporal_average], Original ATen: [aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_mean_0.run(buf1, primals_1, 16, 64, grid=grid(16), stream=stream0)
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [weights], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_3, buf1, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del primals_2
del primals_3
buf3 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
triton_poi_fused_mul_1.run(buf2, primals_1, buf3, 1024, grid=grid(1024), stream=stream0)
return (buf3, primals_1, buf1, buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch as th
import torch.nn as nn
class SelfGating(nn.Module):
def __init__(self, input_dim):
super(SelfGating, self).__init__()
self.fc = nn.Linear(input_dim, input_dim)
def forward(self, input_tensor):
"""Feature gating as used in S3D-G."""
spatiotemporal_average = th.mean(input_tensor, dim=[2, 3, 4])
weights = self.fc(spatiotemporal_average)
weights = th.sigmoid(weights)
return weights[:, :, None, None, None] * input_tensor
def get_inputs():
return [torch.rand([4, 4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_mean_0(in_out_ptr0, in_ptr0, xnumel, rnumel, XBLOCK:
tl.constexpr):
xnumel = 16
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp5 = 64.0
tmp6 = tmp4 / tmp5
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp6, xmask)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 64
x2 = xindex
tmp0 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr1 + x2, xmask)
tmp1 = tl.sigmoid(tmp0)
tmp3 = tmp1 * tmp2
tl.store(out_ptr0 + x2, tmp3, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_mean_0[grid(16)](buf1, primals_1, 16, 64, XBLOCK=8,
num_warps=4, num_stages=1)
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_3, buf1, reinterpret_tensor(primals_2,
(4, 4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del primals_2
del primals_3
buf3 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.float32)
triton_poi_fused_mul_1[grid(1024)](buf2, primals_1, buf3, 1024,
XBLOCK=256, num_warps=4, num_stages=1)
return buf3, primals_1, buf1, buf2
class SelfGatingNew(nn.Module):
def __init__(self, input_dim):
super(SelfGatingNew, self).__init__()
self.fc = nn.Linear(input_dim, input_dim)
def forward(self, input_0):
primals_2 = self.fc.weight
primals_3 = self.fc.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Tiamat-Tech/just-ask
|
SelfGating
| false | 14,492 |
[
"Apache-2.0"
] | 59 |
80725161e12ad0682b4c2091f61a5889a335ba21
|
https://github.com/Tiamat-Tech/just-ask/tree/80725161e12ad0682b4c2091f61a5889a335ba21
|
InvertibleLinearFlow
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/o7/co7osyxs2cvh2l5heuqokn4jvtjd7nknmgqdhgapquyibl2quplf.py
# Topologically Sorted Source Nodes: [double], Original ATen: [aten._to_copy]
# Source node to ATen node mapping:
# double => convert_element_type
# Graph fragment:
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%primals_2, torch.float64), kwargs = {})
triton_poi_fused__to_copy_0 = async_compile.triton('triton_poi_fused__to_copy_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp64', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tmp0.to(tl.float64)
tl.store(out_ptr0 + (x0), tmp1, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/j3/cj3ohnrpnbmgzo767eyf4rs7a6fce645oyyhhxyyda5yjb2obqnw.py
# Topologically Sorted Source Nodes: [logdet, mul, logdet_1], Original ATen: [aten._to_copy, aten.mul]
# Source node to ATen node mapping:
# logdet => convert_element_type_1
# logdet_1 => mul_1
# mul => full_default
# Graph fragment:
# %convert_element_type_1 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%getitem_1, torch.float32), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([4], 4.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%full_default, %convert_element_type_1), kwargs = {})
triton_poi_fused__to_copy_mul_1 = async_compile.triton('triton_poi_fused__to_copy_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp64', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (0))
tmp1 = tl.broadcast_to(tmp0, [XBLOCK])
tmp2 = tmp1.to(tl.float32)
tmp3 = 4.0
tmp4 = tmp3 * tmp2
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [outputs], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0), primals_2, out=buf0)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float64)
# Topologically Sorted Source Nodes: [double], Original ATen: [aten._to_copy]
stream0 = get_raw_stream(0)
triton_poi_fused__to_copy_0.run(primals_2, buf1, 16, grid=grid(16), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [double, linalg_slogdet], Original ATen: [aten._to_copy, aten._linalg_slogdet]
buf2 = torch.ops.aten._linalg_slogdet.default(buf1)
del buf1
buf4 = buf2[1]
buf5 = buf2[2]
buf6 = buf2[3]
del buf2
buf7 = empty_strided_cuda((4, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [logdet, mul, logdet_1], Original ATen: [aten._to_copy, aten.mul]
triton_poi_fused__to_copy_mul_1.run(buf4, buf7, 4, grid=grid(4), stream=stream0)
del buf4
return (reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0), buf7, buf5, buf6, reinterpret_tensor(primals_1, (4, 64), (1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import numpy as np
import torch.nn as nn
from typing import Tuple
class Flow(nn.Module):
def __init__(self):
super(Flow, self).__init__()
def forward(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*inputs: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def inverse(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*input: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def init(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Initiate the weights according to the initial input data
:param inputs:
:param kwargs:
:return:
"""
raise NotImplementedError
class InvertibleLinearFlow(Flow):
def __init__(self, channels):
super(InvertibleLinearFlow, self).__init__()
self.channels = channels
w_init = np.linalg.qr(np.random.randn(channels, channels))[0].astype(np
.float32)
self.register_parameter('weight', nn.Parameter(torch.from_numpy(
w_init)))
def forward(self, inputs: 'torch.Tensor', inputs_lengths=None) ->Tuple[
torch.Tensor, torch.Tensor]:
input_shape = inputs.shape
outputs = torch.matmul(inputs, self.weight)
logdet = torch.linalg.slogdet(self.weight.double())[1].float()
if inputs_lengths is None:
logdet = torch.ones(input_shape[0], device=inputs.device) * float(
input_shape[1]) * logdet
else:
logdet = inputs_lengths.float() * logdet
return outputs, logdet
def inverse(self, inputs: 'torch.Tensor', inputs_lengths=None) ->Tuple[
torch.Tensor, torch.Tensor]:
input_shape = inputs.shape
outputs = torch.matmul(inputs, torch.linalg.inv(self.weight))
logdet = torch.linalg.slogdet(torch.linalg.inv(self.weight.double()))[1
].float()
if inputs_lengths is None:
logdet = torch.ones(input_shape[0], device=inputs.device) * float(
input_shape[1]) * logdet
else:
logdet = inputs_lengths.float() * logdet
return outputs, logdet
def init(self, inputs: 'torch.Tensor', inputs_lengths=None):
return self.forward(inputs, inputs_lengths)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import numpy as np
import torch.nn as nn
from typing import Tuple
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__to_copy_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tmp0.to(tl.float64)
tl.store(out_ptr0 + x0, tmp1, xmask)
@triton.jit
def triton_poi_fused__to_copy_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK])
tmp2 = tmp1.to(tl.float32)
tmp3 = 4.0
tmp4 = tmp3 * tmp2
tl.store(out_ptr0 + x0, tmp4, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0),
primals_2, out=buf0)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float64)
get_raw_stream(0)
triton_poi_fused__to_copy_0[grid(16)](primals_2, buf1, 16, XBLOCK=
16, num_warps=1, num_stages=1)
del primals_2
buf2 = torch.ops.aten._linalg_slogdet.default(buf1)
del buf1
buf4 = buf2[1]
buf5 = buf2[2]
buf6 = buf2[3]
del buf2
buf7 = empty_strided_cuda((4,), (1,), torch.float32)
triton_poi_fused__to_copy_mul_1[grid(4)](buf4, buf7, 4, XBLOCK=4,
num_warps=1, num_stages=1)
del buf4
return reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0
), buf7, buf5, buf6, reinterpret_tensor(primals_1, (4, 64), (1, 4), 0)
class Flow(nn.Module):
def __init__(self):
super(Flow, self).__init__()
def forward(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*inputs: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def inverse(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*input: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def init(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Initiate the weights according to the initial input data
:param inputs:
:param kwargs:
:return:
"""
raise NotImplementedError
class InvertibleLinearFlowNew(Flow):
def __init__(self, channels):
super(InvertibleLinearFlowNew, self).__init__()
self.channels = channels
w_init = np.linalg.qr(np.random.randn(channels, channels))[0].astype(np
.float32)
self.register_parameter('weight', nn.Parameter(torch.from_numpy(
w_init)))
def inverse(self, inputs: 'torch.Tensor', inputs_lengths=None) ->Tuple[
torch.Tensor, torch.Tensor]:
input_shape = inputs.shape
outputs = torch.matmul(inputs, torch.linalg.inv(self.weight))
logdet = torch.linalg.slogdet(torch.linalg.inv(self.weight.double()))[1
].float()
if inputs_lengths is None:
logdet = torch.ones(input_shape[0], device=inputs.device) * float(
input_shape[1]) * logdet
else:
logdet = inputs_lengths.float() * logdet
return outputs, logdet
def init(self, inputs: 'torch.Tensor', inputs_lengths=None):
return self.forward(inputs, inputs_lengths)
def forward(self, input_0):
primals_2 = self.weight
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0], output[1]
|
Tiamat-Tech/VAENAR-TTS
|
InvertibleLinearFlow
| false | 14,493 |
[
"MIT"
] | 62 |
69b6b5be1ab5168cfd3c6ab902075638e76a3b8d
|
https://github.com/Tiamat-Tech/VAENAR-TTS/tree/69b6b5be1ab5168cfd3c6ab902075638e76a3b8d
|
EqualLinear
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/wi/cwiyl3lwwtancorrifw77xt3aqb4lermdintht45zvkj3bg54nbl.py
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul_1 => mul_1
# Graph fragment:
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_2, 0.5), kwargs = {})
triton_poi_fused_mul_0 = async_compile.triton('triton_poi_fused_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/2o/c2oqkq7zaubqmw7vuixxlseb2ff5jzqqbyczicxlmsahuxwdpdyp.py
# Topologically Sorted Source Nodes: [bias], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# bias => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_1, 1), kwargs = {})
triton_poi_fused_mul_1 = async_compile.triton('triton_poi_fused_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, ), (1, ))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul_1], Original ATen: [aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_0.run(primals_2, buf0, 16, grid=grid(16), stream=stream0)
del primals_2
buf1 = empty_strided_cuda((4, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [bias], Original ATen: [aten.mul]
triton_poi_fused_mul_1.run(primals_1, buf1, 4, grid=grid(4), stream=stream0)
del primals_1
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [bias, out], Original ATen: [aten.mul, aten.addmm]
extern_kernels.addmm(buf1, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf0, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del buf0
del buf1
return (reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
from torch.autograd import Function
import math
import torch
from torch import nn
from torch.nn import functional as F
def fused_leaky_relu(input, bias=None, negative_slope=0.2, scale=2 ** 0.5):
if input.device.type == 'cpu':
if bias is not None:
rest_dim = [1] * (input.ndim - bias.ndim - 1)
return F.leaky_relu(input + bias.view(1, bias.shape[0], *
rest_dim), negative_slope=0.2) * scale
else:
return F.leaky_relu(input, negative_slope=0.2) * scale
else:
return FusedLeakyReLUFunction.apply(input.contiguous(), bias,
negative_slope, scale)
class FusedLeakyReLUFunctionBackward(Function):
@staticmethod
def forward(ctx, grad_output, out, bias, negative_slope, scale):
ctx.save_for_backward(out)
ctx.negative_slope = negative_slope
ctx.scale = scale
empty = grad_output.new_empty(0)
grad_input = fused.fused_bias_act(grad_output.contiguous(), empty,
out, 3, 1, negative_slope, scale)
dim = [0]
if grad_input.ndim > 2:
dim += list(range(2, grad_input.ndim))
if bias:
grad_bias = grad_input.sum(dim).detach()
else:
grad_bias = empty
return grad_input, grad_bias
@staticmethod
def backward(ctx, gradgrad_input, gradgrad_bias):
out, = ctx.saved_tensors
gradgrad_out = fused.fused_bias_act(gradgrad_input.contiguous(),
gradgrad_bias, out, 3, 1, ctx.negative_slope, ctx.scale)
return gradgrad_out, None, None, None, None
class FusedLeakyReLUFunction(Function):
@staticmethod
def forward(ctx, input, bias, negative_slope, scale):
empty = input.new_empty(0)
ctx.bias = bias is not None
if bias is None:
bias = empty
out = fused.fused_bias_act(input, bias, empty, 3, 0, negative_slope,
scale)
ctx.save_for_backward(out)
ctx.negative_slope = negative_slope
ctx.scale = scale
return out
@staticmethod
def backward(ctx, grad_output):
out, = ctx.saved_tensors
grad_input, grad_bias = FusedLeakyReLUFunctionBackward.apply(
grad_output, out, ctx.bias, ctx.negative_slope, ctx.scale)
if not ctx.bias:
grad_bias = None
return grad_input, grad_bias, None, None
class EqualLinear(nn.Module):
def __init__(self, in_dim, out_dim, bias=True, bias_init=0, lr_mul=1,
activation=None):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_dim, in_dim).div_(lr_mul))
if bias:
self.bias = nn.Parameter(torch.zeros(out_dim).fill_(bias_init))
else:
self.bias = None
self.activation = activation
self.scale = 1 / math.sqrt(in_dim) * lr_mul
self.lr_mul = lr_mul
def forward(self, input):
bias = self.bias
if bias is not None:
bias = self.bias * self.lr_mul
if self.activation:
out = F.linear(input, self.weight * self.scale)
out = fused_leaky_relu(out, bias)
else:
out = F.linear(input, self.weight * self.scale, bias=bias)
return out
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]})'
)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_dim': 4, 'out_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.autograd import Function
import math
from torch import nn
from torch.nn import functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4,), (1,))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_0[grid(16)](primals_2, buf0, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del primals_2
buf1 = empty_strided_cuda((4,), (1,), torch.float32)
triton_poi_fused_mul_1[grid(4)](primals_1, buf1, 4, XBLOCK=4,
num_warps=1, num_stages=1)
del primals_1
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(buf1, reinterpret_tensor(primals_3, (64, 4), (
4, 1), 0), reinterpret_tensor(buf0, (4, 4), (1, 4), 0), alpha=1,
beta=1, out=buf2)
del buf0
del buf1
return reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0)
def fused_leaky_relu(input, bias=None, negative_slope=0.2, scale=2 ** 0.5):
if input.device.type == 'cpu':
if bias is not None:
rest_dim = [1] * (input.ndim - bias.ndim - 1)
return F.leaky_relu(input + bias.view(1, bias.shape[0], *
rest_dim), negative_slope=0.2) * scale
else:
return F.leaky_relu(input, negative_slope=0.2) * scale
else:
return FusedLeakyReLUFunction.apply(input.contiguous(), bias,
negative_slope, scale)
class FusedLeakyReLUFunctionBackward(Function):
@staticmethod
def forward(ctx, grad_output, out, bias, negative_slope, scale):
ctx.save_for_backward(out)
ctx.negative_slope = negative_slope
ctx.scale = scale
empty = grad_output.new_empty(0)
grad_input = fused.fused_bias_act(grad_output.contiguous(), empty,
out, 3, 1, negative_slope, scale)
dim = [0]
if grad_input.ndim > 2:
dim += list(range(2, grad_input.ndim))
if bias:
grad_bias = grad_input.sum(dim).detach()
else:
grad_bias = empty
return grad_input, grad_bias
@staticmethod
def backward(ctx, gradgrad_input, gradgrad_bias):
out, = ctx.saved_tensors
gradgrad_out = fused.fused_bias_act(gradgrad_input.contiguous(),
gradgrad_bias, out, 3, 1, ctx.negative_slope, ctx.scale)
return gradgrad_out, None, None, None, None
class FusedLeakyReLUFunction(Function):
@staticmethod
def forward(ctx, input, bias, negative_slope, scale):
empty = input.new_empty(0)
ctx.bias = bias is not None
if bias is None:
bias = empty
out = fused.fused_bias_act(input, bias, empty, 3, 0, negative_slope,
scale)
ctx.save_for_backward(out)
ctx.negative_slope = negative_slope
ctx.scale = scale
return out
@staticmethod
def backward(ctx, grad_output):
out, = ctx.saved_tensors
grad_input, grad_bias = FusedLeakyReLUFunctionBackward.apply(
grad_output, out, ctx.bias, ctx.negative_slope, ctx.scale)
if not ctx.bias:
grad_bias = None
return grad_input, grad_bias, None, None
class EqualLinearNew(nn.Module):
def __init__(self, in_dim, out_dim, bias=True, bias_init=0, lr_mul=1,
activation=None):
super().__init__()
self.weight = nn.Parameter(torch.randn(out_dim, in_dim).div_(lr_mul))
if bias:
self.bias = nn.Parameter(torch.zeros(out_dim).fill_(bias_init))
else:
self.bias = None
self.activation = activation
self.scale = 1 / math.sqrt(in_dim) * lr_mul
self.lr_mul = lr_mul
def __repr__(self):
return (
f'{self.__class__.__name__}({self.weight.shape[1]}, {self.weight.shape[0]})'
)
def forward(self, input_0):
primals_2 = self.weight
primals_1 = self.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Tiamat-Tech/alias-free-gan-pytorch
|
EqualLinear
| false | 14,494 |
[
"MIT"
] | 485 |
f14d54ce2d973880b0c352614b2d63088c9026ae
|
https://github.com/Tiamat-Tech/alias-free-gan-pytorch/tree/f14d54ce2d973880b0c352614b2d63088c9026ae
|
MonomialNN
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ng/cngqivfvvl2dq67u2ui7jcj3osrhwji3l6tjo2vukmtvgnfcg6nv.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%arg0_1, %pow_2, %pow_3, %pow_4], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 16) % 16
x0 = xindex % 16
x2 = (xindex // 256)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (16*x1) + (64*x2)), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr0 + (x0 + (16*((-4) + x1)) + (64*x2)), tmp9 & xmask, other=0.0)
tmp11 = tmp10 * tmp10
tmp12 = tl.full(tmp11.shape, 0.0, tmp11.dtype)
tmp13 = tl.where(tmp9, tmp11, tmp12)
tmp14 = tmp0 >= tmp7
tmp15 = tl.full([1], 12, tl.int64)
tmp16 = tmp0 < tmp15
tmp17 = tmp14 & tmp16
tmp18 = tl.load(in_ptr0 + (x0 + (16*((-8) + x1)) + (64*x2)), tmp17 & xmask, other=0.0)
tmp19 = tmp18 * tmp18
tmp20 = tmp19 * tmp18
tmp21 = tl.full(tmp20.shape, 0.0, tmp20.dtype)
tmp22 = tl.where(tmp17, tmp20, tmp21)
tmp23 = tmp0 >= tmp15
tmp24 = tl.full([1], 16, tl.int64)
tmp25 = tmp0 < tmp24
tmp26 = tl.load(in_ptr0 + (x0 + (16*((-12) + x1)) + (64*x2)), tmp23 & xmask, other=0.0)
tmp27 = tmp26 * tmp26
tmp28 = tmp27 * tmp27
tmp29 = tl.full(tmp28.shape, 0.0, tmp28.dtype)
tmp30 = tl.where(tmp23, tmp28, tmp29)
tmp31 = tl.where(tmp17, tmp22, tmp30)
tmp32 = tl.where(tmp9, tmp13, tmp31)
tmp33 = tl.where(tmp4, tmp5, tmp32)
tl.store(out_ptr0 + (x3), tmp33, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 16, 4, 4), (256, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(arg0_1, buf0, 1024, grid=grid(1024), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
from warnings import warn
class MonomialNN(nn.Module):
"""A network that expands its input to a given list of monomials.
Its output shape will be (n_samples, n_input_units * n_degrees)
:param degrees: max degree to be included, or a list of degrees that will be used
:type degrees: int or list[int] or tuple[int]
"""
def __init__(self, degrees):
super(MonomialNN, self).__init__()
if isinstance(degrees, int):
degrees = [d for d in range(1, degrees + 1)]
self.degrees = tuple(degrees)
if len(self.degrees) == 0:
raise ValueError('No degrees used, check `degrees` argument again')
if 0 in degrees:
warn(
'One of the degrees is 0 which might introduce redundant features'
)
if len(set(self.degrees)) < len(self.degrees):
warn(f'Duplicate degrees found: {self.degrees}')
def forward(self, x):
return torch.cat([(x ** d) for d in self.degrees], dim=1)
def __repr__(self):
return f'{self.__class__.__name__}(degrees={self.degrees})'
def __str__(self):
return self.__repr__()
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'degrees': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
from warnings import warn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 16 % 16
x0 = xindex % 16
x2 = xindex // 256
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 16 * x1 + 64 * x2), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr0 + (x0 + 16 * (-4 + x1) + 64 * x2), tmp9 & xmask,
other=0.0)
tmp11 = tmp10 * tmp10
tmp12 = tl.full(tmp11.shape, 0.0, tmp11.dtype)
tmp13 = tl.where(tmp9, tmp11, tmp12)
tmp14 = tmp0 >= tmp7
tmp15 = tl.full([1], 12, tl.int64)
tmp16 = tmp0 < tmp15
tmp17 = tmp14 & tmp16
tmp18 = tl.load(in_ptr0 + (x0 + 16 * (-8 + x1) + 64 * x2), tmp17 &
xmask, other=0.0)
tmp19 = tmp18 * tmp18
tmp20 = tmp19 * tmp18
tmp21 = tl.full(tmp20.shape, 0.0, tmp20.dtype)
tmp22 = tl.where(tmp17, tmp20, tmp21)
tmp23 = tmp0 >= tmp15
tl.full([1], 16, tl.int64)
tmp26 = tl.load(in_ptr0 + (x0 + 16 * (-12 + x1) + 64 * x2), tmp23 &
xmask, other=0.0)
tmp27 = tmp26 * tmp26
tmp28 = tmp27 * tmp27
tmp29 = tl.full(tmp28.shape, 0.0, tmp28.dtype)
tmp30 = tl.where(tmp23, tmp28, tmp29)
tmp31 = tl.where(tmp17, tmp22, tmp30)
tmp32 = tl.where(tmp9, tmp13, tmp31)
tmp33 = tl.where(tmp4, tmp5, tmp32)
tl.store(out_ptr0 + x3, tmp33, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 16, 4, 4), (256, 16, 4, 1), torch.float32
)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(1024)](arg0_1, buf0, 1024, XBLOCK=256,
num_warps=4, num_stages=1)
del arg0_1
return buf0,
class MonomialNNNew(nn.Module):
"""A network that expands its input to a given list of monomials.
Its output shape will be (n_samples, n_input_units * n_degrees)
:param degrees: max degree to be included, or a list of degrees that will be used
:type degrees: int or list[int] or tuple[int]
"""
def __init__(self, degrees):
super(MonomialNNNew, self).__init__()
if isinstance(degrees, int):
degrees = [d for d in range(1, degrees + 1)]
self.degrees = tuple(degrees)
if len(self.degrees) == 0:
raise ValueError('No degrees used, check `degrees` argument again')
if 0 in degrees:
warn(
'One of the degrees is 0 which might introduce redundant features'
)
if len(set(self.degrees)) < len(self.degrees):
warn(f'Duplicate degrees found: {self.degrees}')
def __repr__(self):
return f'{self.__class__.__name__}(degrees={self.degrees})'
def __str__(self):
return self.__repr__()
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
Tiamat-Tech/neurodiffeq
|
MonomialNN
| false | 14,495 |
[
"MIT"
] | 202 |
622827e5b9b65d285ebe36614fbdae68ba07f4dc
|
https://github.com/Tiamat-Tech/neurodiffeq/tree/622827e5b9b65d285ebe36614fbdae68ba07f4dc
|
Actor
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ms/cmsuzohbg5nq52jnvirovzkvykrzzko5xomu7zyu5e5u2lhegppw.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%primals_1, %primals_2], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 8
x1 = (xindex // 8)
x2 = xindex
tmp0 = x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((4*x1) + x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr1 + ((4*x1) + ((-4) + x0)), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + (x2), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ac/cacdwifxdru2eihx3n66wqfym5hjpdo6yxk3gsol5t54xroplkwv.py
# Topologically Sorted Source Nodes: [a], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# a => relu
# Graph fragment:
# %add_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default_1, %primals_4), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor_1,), kwargs = {})
triton_poi_fused_relu_1 = async_compile.triton('triton_poi_fused_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 400
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/vp/cvpyw2bjgo55x2ne47dmpi2vmqu5t4eb3wcvjgjcnove3xyj7bcr.py
# Topologically Sorted Source Nodes: [a_1], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# a_1 => relu_1
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_6), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor,), kwargs = {})
triton_poi_fused_relu_2 = async_compile.triton('triton_poi_fused_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 300
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/tw/ctwaaaylf5igc5ibl6fkd3mledcooplynpspw36onh63ebqcbbwv.py
# Topologically Sorted Source Nodes: [tanh, a_2, add, clamp], Original ATen: [aten.tanh, aten.mul, aten.add, aten.clamp, aten.ge, aten.le, aten.logical_and]
# Source node to ATen node mapping:
# a_2 => mul
# add => add
# clamp => clamp_max, clamp_min
# tanh => tanh
# Graph fragment:
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%addmm_2,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%tanh, 0.2), kwargs = {})
# %add : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %primals_2), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add, -4), kwargs = {})
# %clamp_max : [num_users=1] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min, 4), kwargs = {})
# %ge : [num_users=1] = call_function[target=torch.ops.aten.ge.Scalar](args = (%add, -4), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%add, 4), kwargs = {})
# %logical_and : [num_users=1] = call_function[target=torch.ops.aten.logical_and.default](args = (%ge, %le), kwargs = {})
triton_poi_fused_add_clamp_ge_le_logical_and_mul_tanh_3 = async_compile.triton('triton_poi_fused_add_clamp_ge_le_logical_and_mul_tanh_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i1', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_ge_le_logical_and_mul_tanh_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_ge_le_logical_and_mul_tanh_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp4 = tl.load(in_ptr1 + (x0), xmask)
tmp1 = libdevice.tanh(tmp0)
tmp2 = 0.2
tmp3 = tmp1 * tmp2
tmp5 = tmp3 + tmp4
tmp6 = -4.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = 4.0
tmp9 = triton_helpers.minimum(tmp7, tmp8)
tmp10 = tmp5 >= tmp6
tmp11 = tmp5 <= tmp8
tmp12 = tmp10 & tmp11
tl.store(out_ptr0 + (x0), tmp9, xmask)
tl.store(out_ptr1 + (x0), tmp12, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (400, 8), (8, 1))
assert_size_stride(primals_4, (400, ), (1, ))
assert_size_stride(primals_5, (300, 400), (400, 1))
assert_size_stride(primals_6, (300, ), (1, ))
assert_size_stride(primals_7, (4, 300), (300, 1))
assert_size_stride(primals_8, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(primals_1, primals_2, buf0, 32, grid=grid(32), stream=stream0)
del primals_1
buf1 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf0, reinterpret_tensor(primals_3, (8, 400), (1, 8), 0), out=buf1)
del primals_3
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [a], Original ATen: [aten.relu]
triton_poi_fused_relu_1.run(buf2, primals_4, 1600, grid=grid(1600), stream=stream0)
del primals_4
buf3 = empty_strided_cuda((4, 300), (300, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf2, reinterpret_tensor(primals_5, (400, 300), (1, 400), 0), out=buf3)
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [a_1], Original ATen: [aten.relu]
triton_poi_fused_relu_2.run(buf4, primals_6, 1200, grid=grid(1200), stream=stream0)
del primals_6
buf5 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_2], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_8, buf4, reinterpret_tensor(primals_7, (300, 4), (1, 300), 0), alpha=1, beta=1, out=buf5)
del primals_8
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
buf7 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [tanh, a_2, add, clamp], Original ATen: [aten.tanh, aten.mul, aten.add, aten.clamp, aten.ge, aten.le, aten.logical_and]
triton_poi_fused_add_clamp_ge_le_logical_and_mul_tanh_3.run(buf5, primals_2, buf6, buf7, 16, grid=grid(16), stream=stream0)
del primals_2
return (buf6, buf0, buf2, buf4, buf5, buf7, primals_7, primals_5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((400, 8), (8, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((400, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((300, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((300, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 300), (300, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.data
class Actor(nn.Module):
def __init__(self, state_dim, action_dim, max_action):
super(Actor, self).__init__()
self.l1 = nn.Linear(state_dim + action_dim, 400)
self.l2 = nn.Linear(400, 300)
self.l3 = nn.Linear(300, action_dim)
self.max_action = max_action
def forward(self, state, action):
a = F.relu(self.l1(torch.cat([state, action], 1)))
a = F.relu(self.l2(a))
a = 0.05 * self.max_action * torch.tanh(self.l3(a))
return (a + action).clamp(-self.max_action, self.max_action)
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'state_dim': 4, 'action_dim': 4, 'max_action': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 8
x1 = xindex // 8
x2 = xindex
tmp0 = x0
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (4 * x1 + x0), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp9 = tl.load(in_ptr1 + (4 * x1 + (-4 + x0)), tmp6 & xmask,
eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + x2, tmp10, xmask)
@triton.jit
def triton_poi_fused_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 1600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 400
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_relu_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 1200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 300
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_clamp_ge_le_logical_and_mul_tanh_3(in_ptr0,
in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp4 = tl.load(in_ptr1 + x0, xmask)
tmp1 = libdevice.tanh(tmp0)
tmp2 = 0.2
tmp3 = tmp1 * tmp2
tmp5 = tmp3 + tmp4
tmp6 = -4.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = 4.0
tmp9 = triton_helpers.minimum(tmp7, tmp8)
tmp10 = tmp5 >= tmp6
tmp11 = tmp5 <= tmp8
tmp12 = tmp10 & tmp11
tl.store(out_ptr0 + x0, tmp9, xmask)
tl.store(out_ptr1 + x0, tmp12, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (400, 8), (8, 1))
assert_size_stride(primals_4, (400,), (1,))
assert_size_stride(primals_5, (300, 400), (400, 1))
assert_size_stride(primals_6, (300,), (1,))
assert_size_stride(primals_7, (4, 300), (300, 1))
assert_size_stride(primals_8, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(32)](primals_1, primals_2, buf0, 32,
XBLOCK=32, num_warps=1, num_stages=1)
del primals_1
buf1 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
extern_kernels.mm(buf0, reinterpret_tensor(primals_3, (8, 400), (1,
8), 0), out=buf1)
del primals_3
buf2 = buf1
del buf1
triton_poi_fused_relu_1[grid(1600)](buf2, primals_4, 1600, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_4
buf3 = empty_strided_cuda((4, 300), (300, 1), torch.float32)
extern_kernels.mm(buf2, reinterpret_tensor(primals_5, (400, 300), (
1, 400), 0), out=buf3)
buf4 = buf3
del buf3
triton_poi_fused_relu_2[grid(1200)](buf4, primals_6, 1200, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_6
buf5 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_8, buf4, reinterpret_tensor(primals_7,
(300, 4), (1, 300), 0), alpha=1, beta=1, out=buf5)
del primals_8
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
buf7 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_add_clamp_ge_le_logical_and_mul_tanh_3[grid(16)](buf5,
primals_2, buf6, buf7, 16, XBLOCK=16, num_warps=1, num_stages=1)
del primals_2
return buf6, buf0, buf2, buf4, buf5, buf7, primals_7, primals_5
class ActorNew(nn.Module):
def __init__(self, state_dim, action_dim, max_action):
super(ActorNew, self).__init__()
self.l1 = nn.Linear(state_dim + action_dim, 400)
self.l2 = nn.Linear(400, 300)
self.l3 = nn.Linear(300, action_dim)
self.max_action = max_action
def forward(self, input_0, input_1):
primals_3 = self.l1.weight
primals_4 = self.l1.bias
primals_5 = self.l2.weight
primals_6 = self.l2.bias
primals_7 = self.l3.weight
primals_8 = self.l3.bias
primals_1 = input_0
primals_2 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8])
return output[0]
|
Thibaud-Ardoin/d4rl_evaluations
|
Actor
| false | 14,496 |
[
"Apache-2.0"
] | 123 |
135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
https://github.com/Thibaud-Ardoin/d4rl_evaluations/tree/135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
InstanceSimilarity
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ut/cut5f4hltl6d7ynm46tiuq5eqxyuetbydusf3gimebkwxzdtb2bl.py
# Topologically Sorted Source Nodes: [norm_G_s, norm_G_t, loss, add, norm_G_s_1, norm_G_t_1, loss_1, add_1, norm_G_s_2, norm_G_t_2, loss_2, add_2, norm_G_s_3, norm_G_t_3, loss_3, add_3], Original ATen: [aten.div, aten.mse_loss, aten.add]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# add_2 => add_2
# add_3 => add_3
# loss => mean, pow_5, sub
# loss_1 => mean_1, pow_10, sub_1
# loss_2 => mean_2, pow_15, sub_2
# loss_3 => mean_3, pow_20, sub_3
# norm_G_s => div
# norm_G_s_1 => div_2
# norm_G_s_2 => div_4
# norm_G_s_3 => div_6
# norm_G_t => div_1
# norm_G_t_1 => div_3
# norm_G_t_2 => div_5
# norm_G_t_3 => div_7
# Graph fragment:
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm, %expand), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm_1, %expand_1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%div, %div_1), kwargs = {})
# %pow_5 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub, 2), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%pow_5,), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean, 0), kwargs = {})
# %div_2 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm_2, %expand_2), kwargs = {})
# %div_3 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm_3, %expand_3), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%div_2, %div_3), kwargs = {})
# %pow_10 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_1, 2), kwargs = {})
# %mean_1 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%pow_10,), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add, %mean_1), kwargs = {})
# %div_4 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm_4, %expand_4), kwargs = {})
# %div_5 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm_5, %expand_5), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%div_4, %div_5), kwargs = {})
# %pow_15 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_2, 2), kwargs = {})
# %mean_2 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%pow_15,), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_1, %mean_2), kwargs = {})
# %div_6 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm_6, %expand_6), kwargs = {})
# %div_7 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mm_7, %expand_7), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%div_6, %div_7), kwargs = {})
# %pow_20 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_3, 2), kwargs = {})
# %mean_3 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%pow_20,), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_2, %mean_3), kwargs = {})
triton_per_fused_add_div_mse_loss_0 = async_compile.triton('triton_per_fused_add_div_mse_loss_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 16],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: '*fp32', 8: '*fp32', 9: 'i32', 10: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {9: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 10), equal_to_1=(9,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_mse_loss_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 40, 'num_reduction': 4, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_mse_loss_0(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r2 = rindex
r1 = (rindex // 4)
tmp0 = tl.load(in_ptr0 + (r2), None)
tmp1 = tl.load(in_ptr0 + (4*r1), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr1 + (r2), None)
tmp17 = tl.load(in_ptr1 + (4*r1), None, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr1 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr1 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr1 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp36 = tl.load(in_ptr2 + (r2), None)
tmp37 = tl.load(in_ptr2 + (4*r1), None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr2 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp42 = tl.load(in_ptr2 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp45 = tl.load(in_ptr2 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp51 = tl.load(in_ptr3 + (r2), None)
tmp52 = tl.load(in_ptr3 + (4*r1), None, eviction_policy='evict_last')
tmp54 = tl.load(in_ptr3 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr3 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp60 = tl.load(in_ptr3 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp71 = tl.load(in_ptr4 + (r2), None)
tmp72 = tl.load(in_ptr4 + (4*r1), None, eviction_policy='evict_last')
tmp74 = tl.load(in_ptr4 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp77 = tl.load(in_ptr4 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp80 = tl.load(in_ptr4 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp86 = tl.load(in_ptr5 + (r2), None)
tmp87 = tl.load(in_ptr5 + (4*r1), None, eviction_policy='evict_last')
tmp89 = tl.load(in_ptr5 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp92 = tl.load(in_ptr5 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp95 = tl.load(in_ptr5 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp106 = tl.load(in_ptr6 + (r2), None)
tmp107 = tl.load(in_ptr6 + (4*r1), None, eviction_policy='evict_last')
tmp109 = tl.load(in_ptr6 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp112 = tl.load(in_ptr6 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp115 = tl.load(in_ptr6 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp121 = tl.load(in_ptr7 + (r2), None)
tmp122 = tl.load(in_ptr7 + (4*r1), None, eviction_policy='evict_last')
tmp124 = tl.load(in_ptr7 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp127 = tl.load(in_ptr7 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp130 = tl.load(in_ptr7 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tmp18 = tmp17 * tmp17
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = libdevice.sqrt(tmp27)
tmp29 = triton_helpers.maximum(tmp28, tmp13)
tmp30 = tmp16 / tmp29
tmp31 = tmp15 - tmp30
tmp32 = tmp31 * tmp31
tmp33 = tl.broadcast_to(tmp32, [XBLOCK, RBLOCK])
tmp35 = tl.sum(tmp33, 1)[:, None]
tmp38 = tmp37 * tmp37
tmp40 = tmp39 * tmp39
tmp41 = tmp38 + tmp40
tmp43 = tmp42 * tmp42
tmp44 = tmp41 + tmp43
tmp46 = tmp45 * tmp45
tmp47 = tmp44 + tmp46
tmp48 = libdevice.sqrt(tmp47)
tmp49 = triton_helpers.maximum(tmp48, tmp13)
tmp50 = tmp36 / tmp49
tmp53 = tmp52 * tmp52
tmp55 = tmp54 * tmp54
tmp56 = tmp53 + tmp55
tmp58 = tmp57 * tmp57
tmp59 = tmp56 + tmp58
tmp61 = tmp60 * tmp60
tmp62 = tmp59 + tmp61
tmp63 = libdevice.sqrt(tmp62)
tmp64 = triton_helpers.maximum(tmp63, tmp13)
tmp65 = tmp51 / tmp64
tmp66 = tmp50 - tmp65
tmp67 = tmp66 * tmp66
tmp68 = tl.broadcast_to(tmp67, [XBLOCK, RBLOCK])
tmp70 = tl.sum(tmp68, 1)[:, None]
tmp73 = tmp72 * tmp72
tmp75 = tmp74 * tmp74
tmp76 = tmp73 + tmp75
tmp78 = tmp77 * tmp77
tmp79 = tmp76 + tmp78
tmp81 = tmp80 * tmp80
tmp82 = tmp79 + tmp81
tmp83 = libdevice.sqrt(tmp82)
tmp84 = triton_helpers.maximum(tmp83, tmp13)
tmp85 = tmp71 / tmp84
tmp88 = tmp87 * tmp87
tmp90 = tmp89 * tmp89
tmp91 = tmp88 + tmp90
tmp93 = tmp92 * tmp92
tmp94 = tmp91 + tmp93
tmp96 = tmp95 * tmp95
tmp97 = tmp94 + tmp96
tmp98 = libdevice.sqrt(tmp97)
tmp99 = triton_helpers.maximum(tmp98, tmp13)
tmp100 = tmp86 / tmp99
tmp101 = tmp85 - tmp100
tmp102 = tmp101 * tmp101
tmp103 = tl.broadcast_to(tmp102, [XBLOCK, RBLOCK])
tmp105 = tl.sum(tmp103, 1)[:, None]
tmp108 = tmp107 * tmp107
tmp110 = tmp109 * tmp109
tmp111 = tmp108 + tmp110
tmp113 = tmp112 * tmp112
tmp114 = tmp111 + tmp113
tmp116 = tmp115 * tmp115
tmp117 = tmp114 + tmp116
tmp118 = libdevice.sqrt(tmp117)
tmp119 = triton_helpers.maximum(tmp118, tmp13)
tmp120 = tmp106 / tmp119
tmp123 = tmp122 * tmp122
tmp125 = tmp124 * tmp124
tmp126 = tmp123 + tmp125
tmp128 = tmp127 * tmp127
tmp129 = tmp126 + tmp128
tmp131 = tmp130 * tmp130
tmp132 = tmp129 + tmp131
tmp133 = libdevice.sqrt(tmp132)
tmp134 = triton_helpers.maximum(tmp133, tmp13)
tmp135 = tmp121 / tmp134
tmp136 = tmp120 - tmp135
tmp137 = tmp136 * tmp136
tmp138 = tl.broadcast_to(tmp137, [XBLOCK, RBLOCK])
tmp140 = tl.sum(tmp138, 1)[:, None]
tmp141 = 16.0
tmp142 = tmp35 / tmp141
tmp143 = 0.0
tmp144 = tmp142 + tmp143
tmp145 = tmp70 / tmp141
tmp146 = tmp144 + tmp145
tmp147 = tmp105 / tmp141
tmp148 = tmp146 + tmp147
tmp149 = tmp140 / tmp141
tmp150 = tmp148 + tmp149
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp150, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_s], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 0), reinterpret_tensor(arg0_1, (16, 4), (1, 16), 0), out=buf0)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_t], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 0), reinterpret_tensor(arg1_1, (16, 4), (1, 16), 0), out=buf1)
buf12 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_s_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 192), reinterpret_tensor(arg0_1, (16, 4), (1, 16), 192), out=buf12)
buf13 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_t_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 192), reinterpret_tensor(arg1_1, (16, 4), (1, 16), 192), out=buf13)
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_s_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 64), reinterpret_tensor(arg0_1, (16, 4), (1, 16), 64), out=buf4)
buf5 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_t_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 64), reinterpret_tensor(arg1_1, (16, 4), (1, 16), 64), out=buf5)
buf8 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_s_2], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 128), reinterpret_tensor(arg0_1, (16, 4), (1, 16), 128), out=buf8)
del arg0_1
buf9 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [G_t_2], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 128), reinterpret_tensor(arg1_1, (16, 4), (1, 16), 128), out=buf9)
del arg1_1
buf11 = empty_strided_cuda((), (), torch.float32)
buf16 = buf11; del buf11 # reuse
# Topologically Sorted Source Nodes: [norm_G_s, norm_G_t, loss, add, norm_G_s_1, norm_G_t_1, loss_1, add_1, norm_G_s_2, norm_G_t_2, loss_2, add_2, norm_G_s_3, norm_G_t_3, loss_3, add_3], Original ATen: [aten.div, aten.mse_loss, aten.add]
stream0 = get_raw_stream(0)
triton_per_fused_add_div_mse_loss_0.run(buf16, buf0, buf1, buf4, buf5, buf8, buf9, buf12, buf13, 1, 16, grid=grid(1), stream=stream0)
del buf0
del buf1
del buf12
del buf13
del buf4
del buf5
del buf8
del buf9
return (buf16, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn.functional as F
import torch.nn as nn
class InstanceSimilarity(nn.Module):
"""
Instance Similarity based loss
"""
def __init__(self, mse=True):
super(InstanceSimilarity, self).__init__()
self.mse = mse
def _loss(self, fm_s, fm_t):
fm_s = fm_s.view(fm_s.size(0), -1)
G_s = torch.mm(fm_s, fm_s.t())
norm_G_s = F.normalize(G_s, p=2, dim=1)
fm_t = fm_t.view(fm_t.size(0), -1)
G_t = torch.mm(fm_t, fm_t.t())
norm_G_t = F.normalize(G_t, p=2, dim=1)
loss = F.mse_loss(norm_G_s, norm_G_t) if self.mse else F.l1_loss(
norm_G_s, norm_G_t)
return loss
def forward(self, g_s, g_t):
return sum(self._loss(f_s, f_t) for f_s, f_t in zip(g_s, g_t))
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn.functional as F
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_add_div_mse_loss_0(in_out_ptr0, in_ptr0, in_ptr1,
in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, xnumel, rnumel,
XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r2 = rindex
r1 = rindex // 4
tmp0 = tl.load(in_ptr0 + r2, None)
tmp1 = tl.load(in_ptr0 + 4 * r1, None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr1 + r2, None)
tmp17 = tl.load(in_ptr1 + 4 * r1, None, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr1 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr1 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr1 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp36 = tl.load(in_ptr2 + r2, None)
tmp37 = tl.load(in_ptr2 + 4 * r1, None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr2 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp42 = tl.load(in_ptr2 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp45 = tl.load(in_ptr2 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp51 = tl.load(in_ptr3 + r2, None)
tmp52 = tl.load(in_ptr3 + 4 * r1, None, eviction_policy='evict_last')
tmp54 = tl.load(in_ptr3 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr3 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp60 = tl.load(in_ptr3 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp71 = tl.load(in_ptr4 + r2, None)
tmp72 = tl.load(in_ptr4 + 4 * r1, None, eviction_policy='evict_last')
tmp74 = tl.load(in_ptr4 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp77 = tl.load(in_ptr4 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp80 = tl.load(in_ptr4 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp86 = tl.load(in_ptr5 + r2, None)
tmp87 = tl.load(in_ptr5 + 4 * r1, None, eviction_policy='evict_last')
tmp89 = tl.load(in_ptr5 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp92 = tl.load(in_ptr5 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp95 = tl.load(in_ptr5 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp106 = tl.load(in_ptr6 + r2, None)
tmp107 = tl.load(in_ptr6 + 4 * r1, None, eviction_policy='evict_last')
tmp109 = tl.load(in_ptr6 + (1 + 4 * r1), None, eviction_policy='evict_last'
)
tmp112 = tl.load(in_ptr6 + (2 + 4 * r1), None, eviction_policy='evict_last'
)
tmp115 = tl.load(in_ptr6 + (3 + 4 * r1), None, eviction_policy='evict_last'
)
tmp121 = tl.load(in_ptr7 + r2, None)
tmp122 = tl.load(in_ptr7 + 4 * r1, None, eviction_policy='evict_last')
tmp124 = tl.load(in_ptr7 + (1 + 4 * r1), None, eviction_policy='evict_last'
)
tmp127 = tl.load(in_ptr7 + (2 + 4 * r1), None, eviction_policy='evict_last'
)
tmp130 = tl.load(in_ptr7 + (3 + 4 * r1), None, eviction_policy='evict_last'
)
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tmp18 = tmp17 * tmp17
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = libdevice.sqrt(tmp27)
tmp29 = triton_helpers.maximum(tmp28, tmp13)
tmp30 = tmp16 / tmp29
tmp31 = tmp15 - tmp30
tmp32 = tmp31 * tmp31
tmp33 = tl.broadcast_to(tmp32, [XBLOCK, RBLOCK])
tmp35 = tl.sum(tmp33, 1)[:, None]
tmp38 = tmp37 * tmp37
tmp40 = tmp39 * tmp39
tmp41 = tmp38 + tmp40
tmp43 = tmp42 * tmp42
tmp44 = tmp41 + tmp43
tmp46 = tmp45 * tmp45
tmp47 = tmp44 + tmp46
tmp48 = libdevice.sqrt(tmp47)
tmp49 = triton_helpers.maximum(tmp48, tmp13)
tmp50 = tmp36 / tmp49
tmp53 = tmp52 * tmp52
tmp55 = tmp54 * tmp54
tmp56 = tmp53 + tmp55
tmp58 = tmp57 * tmp57
tmp59 = tmp56 + tmp58
tmp61 = tmp60 * tmp60
tmp62 = tmp59 + tmp61
tmp63 = libdevice.sqrt(tmp62)
tmp64 = triton_helpers.maximum(tmp63, tmp13)
tmp65 = tmp51 / tmp64
tmp66 = tmp50 - tmp65
tmp67 = tmp66 * tmp66
tmp68 = tl.broadcast_to(tmp67, [XBLOCK, RBLOCK])
tmp70 = tl.sum(tmp68, 1)[:, None]
tmp73 = tmp72 * tmp72
tmp75 = tmp74 * tmp74
tmp76 = tmp73 + tmp75
tmp78 = tmp77 * tmp77
tmp79 = tmp76 + tmp78
tmp81 = tmp80 * tmp80
tmp82 = tmp79 + tmp81
tmp83 = libdevice.sqrt(tmp82)
tmp84 = triton_helpers.maximum(tmp83, tmp13)
tmp85 = tmp71 / tmp84
tmp88 = tmp87 * tmp87
tmp90 = tmp89 * tmp89
tmp91 = tmp88 + tmp90
tmp93 = tmp92 * tmp92
tmp94 = tmp91 + tmp93
tmp96 = tmp95 * tmp95
tmp97 = tmp94 + tmp96
tmp98 = libdevice.sqrt(tmp97)
tmp99 = triton_helpers.maximum(tmp98, tmp13)
tmp100 = tmp86 / tmp99
tmp101 = tmp85 - tmp100
tmp102 = tmp101 * tmp101
tmp103 = tl.broadcast_to(tmp102, [XBLOCK, RBLOCK])
tmp105 = tl.sum(tmp103, 1)[:, None]
tmp108 = tmp107 * tmp107
tmp110 = tmp109 * tmp109
tmp111 = tmp108 + tmp110
tmp113 = tmp112 * tmp112
tmp114 = tmp111 + tmp113
tmp116 = tmp115 * tmp115
tmp117 = tmp114 + tmp116
tmp118 = libdevice.sqrt(tmp117)
tmp119 = triton_helpers.maximum(tmp118, tmp13)
tmp120 = tmp106 / tmp119
tmp123 = tmp122 * tmp122
tmp125 = tmp124 * tmp124
tmp126 = tmp123 + tmp125
tmp128 = tmp127 * tmp127
tmp129 = tmp126 + tmp128
tmp131 = tmp130 * tmp130
tmp132 = tmp129 + tmp131
tmp133 = libdevice.sqrt(tmp132)
tmp134 = triton_helpers.maximum(tmp133, tmp13)
tmp135 = tmp121 / tmp134
tmp136 = tmp120 - tmp135
tmp137 = tmp136 * tmp136
tmp138 = tl.broadcast_to(tmp137, [XBLOCK, RBLOCK])
tmp140 = tl.sum(tmp138, 1)[:, None]
tmp141 = 16.0
tmp142 = tmp35 / tmp141
tmp143 = 0.0
tmp144 = tmp142 + tmp143
tmp145 = tmp70 / tmp141
tmp146 = tmp144 + tmp145
tmp147 = tmp105 / tmp141
tmp148 = tmp146 + tmp147
tmp149 = tmp140 / tmp141
tmp150 = tmp148 + tmp149
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp150, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 0),
reinterpret_tensor(arg0_1, (16, 4), (1, 16), 0), out=buf0)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 0),
reinterpret_tensor(arg1_1, (16, 4), (1, 16), 0), out=buf1)
buf12 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 192),
reinterpret_tensor(arg0_1, (16, 4), (1, 16), 192), out=buf12)
buf13 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 192),
reinterpret_tensor(arg1_1, (16, 4), (1, 16), 192), out=buf13)
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 64),
reinterpret_tensor(arg0_1, (16, 4), (1, 16), 64), out=buf4)
buf5 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 64),
reinterpret_tensor(arg1_1, (16, 4), (1, 16), 64), out=buf5)
buf8 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg0_1, (4, 16), (16, 1), 128),
reinterpret_tensor(arg0_1, (16, 4), (1, 16), 128), out=buf8)
del arg0_1
buf9 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 16), (16, 1), 128),
reinterpret_tensor(arg1_1, (16, 4), (1, 16), 128), out=buf9)
del arg1_1
buf11 = empty_strided_cuda((), (), torch.float32)
buf16 = buf11
del buf11
get_raw_stream(0)
triton_per_fused_add_div_mse_loss_0[grid(1)](buf16, buf0, buf1,
buf4, buf5, buf8, buf9, buf12, buf13, 1, 16, XBLOCK=1,
num_warps=2, num_stages=1)
del buf0
del buf1
del buf12
del buf13
del buf4
del buf5
del buf8
del buf9
return buf16,
class InstanceSimilarityNew(nn.Module):
"""
Instance Similarity based loss
"""
def __init__(self, mse=True):
super(InstanceSimilarityNew, self).__init__()
self.mse = mse
def _loss(self, fm_s, fm_t):
fm_s = fm_s.view(fm_s.size(0), -1)
G_s = torch.mm(fm_s, fm_s.t())
norm_G_s = F.normalize(G_s, p=2, dim=1)
fm_t = fm_t.view(fm_t.size(0), -1)
G_t = torch.mm(fm_t, fm_t.t())
norm_G_t = F.normalize(G_t, p=2, dim=1)
loss = F.mse_loss(norm_G_s, norm_G_t) if self.mse else F.l1_loss(
norm_G_s, norm_G_t)
return loss
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
Tiamat-Tech/ZAQ-code
|
InstanceSimilarity
| false | 14,497 |
[
"MIT"
] | 55 |
e7e9f55791e36c6784d58c356d3ced76a7583369
|
https://github.com/Tiamat-Tech/ZAQ-code/tree/e7e9f55791e36c6784d58c356d3ced76a7583369
|
VNLinear
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/4b/c4bin6v3nraanry5cu4t7z3qycurd6acdwc4l5azpezbv2erqt7w.py
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# linear => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%permute,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex % 4
x3 = (xindex // 4)
y0 = yindex % 4
y1 = (yindex // 4)
x5 = xindex
y4 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x3) + (16*x2) + (64*y1)), xmask & ymask)
tl.store(out_ptr0 + (x5 + (16*y4)), tmp0, xmask & ymask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(primals_1, buf0, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_1
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (64, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf1)
del primals_2
return (reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 1, 4, 16), 0), reinterpret_tensor(buf0, (64, 4), (4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.utils.data
import torch
import torch.nn.parallel
class VNLinear(nn.Module):
def __init__(self, in_channels, out_channels):
super(VNLinear, self).__init__()
self.map_to_feat = nn.Linear(in_channels, out_channels, bias=False)
def forward(self, x):
"""
x: point features of shape [B, N_feat, 3, N_samples, ...]
"""
x_out = self.map_to_feat(x.transpose(1, -1)).transpose(1, -1)
return x_out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
import torch.utils.data
import torch
import torch.nn.parallel
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex % 4
x3 = xindex // 4
y0 = yindex % 4
y1 = yindex // 4
x5 = xindex
y4 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x3 + 16 * x2 + 64 * y1), xmask & ymask)
tl.store(out_ptr0 + (x5 + 16 * y4), tmp0, xmask & ymask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(16, 16)](primals_1, buf0, 16, 16,
XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
del primals_1
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf1)
del primals_2
return reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 1, 4, 16), 0
), reinterpret_tensor(buf0, (64, 4), (4, 1), 0)
class VNLinearNew(nn.Module):
def __init__(self, in_channels, out_channels):
super(VNLinearNew, self).__init__()
self.map_to_feat = nn.Linear(in_channels, out_channels, bias=False)
def forward(self, input_0):
primals_2 = self.map_to_feat.weight
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0]
|
Tiamat-Tech/vnn
|
VNLinear
| false | 14,498 |
[
"MIT"
] | 280 |
f3197e210022b5f0015e0da6456adf66bd0cd73e
|
https://github.com/Tiamat-Tech/vnn/tree/f3197e210022b5f0015e0da6456adf66bd0cd73e
|
FC_Q
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/6o/c6o7ainbzocsswla76yvmdsc5donraaar3dzlx2icwrueb7fc46u.py
# Topologically Sorted Source Nodes: [q], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# q => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le_4 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16384],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16384
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 256
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, None)
tl.store(out_ptr0 + (x2), tmp6, None)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/sb/csbqfhl3tbhobxxibww6rnv4q33jyajqsvetse4kiun22xct43oo.py
# Topologically Sorted Source Nodes: [i_2], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# i_2 => relu_4
# Graph fragment:
# %relu_4 : [num_users=4] = call_function[target=torch.ops.aten.relu.default](args = (%view_9,), kwargs = {})
triton_poi_fused_relu_1 = async_compile.triton('triton_poi_fused_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/h4/ch4hu6nfniyge6mjev7hxoom67hh6g5wt742eougy6iak3mso6oo.py
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax, aten.threshold_backward]
# Source node to ATen node mapping:
# log_softmax => amax, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%relu_4, [1], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%relu_4, %amax), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_4, 0), kwargs = {})
triton_poi_fused__log_softmax_threshold_backward_2 = async_compile.triton('triton_poi_fused__log_softmax_threshold_backward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_threshold_backward_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_threshold_backward_2(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = 0.0
tmp10 = tmp0 <= tmp9
tl.store(out_ptr0 + (x3), tmp8, xmask)
tl.store(out_ptr1 + (x3), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/pu/cpuo34qggtttrulwolbrgtcihegg5pnmq65f46vdbaenzdat4oka.py
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# log_softmax => exp, log, sub_1, sum_1
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
triton_poi_fused__log_softmax_3 = async_compile.triton('triton_poi_fused__log_softmax_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tl.store(out_ptr0 + (x3), tmp13, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13 = args
args.clear()
assert_size_stride(primals_1, (256, 4), (4, 1))
assert_size_stride(primals_2, (256, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (256, 256), (256, 1))
assert_size_stride(primals_5, (256, ), (1, ))
assert_size_stride(primals_6, (256, 4), (4, 1))
assert_size_stride(primals_7, (256, ), (1, ))
assert_size_stride(primals_8, (256, 256), (256, 1))
assert_size_stride(primals_9, (256, ), (1, ))
assert_size_stride(primals_10, (4, 256), (256, 1))
assert_size_stride(primals_11, (4, ), (1, ))
assert_size_stride(primals_12, (4, 256), (256, 1))
assert_size_stride(primals_13, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 256), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 256), (4096, 1024, 256, 1), 0); del buf0 # reuse
buf17 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1), torch.bool)
# Topologically Sorted Source Nodes: [q], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf17, 16384, grid=grid(16384), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf1, (64, 256), (256, 1), 0), reinterpret_tensor(primals_4, (256, 256), (1, 256), 0), out=buf2)
buf3 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 256), (1, 4), 0), out=buf3)
del primals_6
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 256), (4096, 1024, 256, 1), 0); del buf3 # reuse
buf15 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1), torch.bool)
# Topologically Sorted Source Nodes: [i], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_0.run(buf4, primals_7, buf15, 16384, grid=grid(16384), stream=stream0)
del primals_7
buf5 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf4, (64, 256), (256, 1), 0), reinterpret_tensor(primals_8, (256, 256), (1, 256), 0), out=buf5)
buf6 = reinterpret_tensor(buf5, (4, 4, 4, 256), (4096, 1024, 256, 1), 0); del buf5 # reuse
buf14 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1), torch.bool)
# Topologically Sorted Source Nodes: [i_1], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_0.run(buf6, primals_9, buf14, 16384, grid=grid(16384), stream=stream0)
del primals_9
buf7 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf6, (64, 256), (256, 1), 0), reinterpret_tensor(primals_10, (256, 4), (1, 256), 0), out=buf7)
buf8 = reinterpret_tensor(buf7, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf7 # reuse
# Topologically Sorted Source Nodes: [i_2], Original ATen: [aten.relu]
triton_poi_fused_relu_1.run(buf8, primals_11, 256, grid=grid(256), stream=stream0)
del primals_11
buf9 = reinterpret_tensor(buf2, (4, 4, 4, 256), (4096, 1024, 256, 1), 0); del buf2 # reuse
buf16 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1), torch.bool)
# Topologically Sorted Source Nodes: [q_1], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_0.run(buf9, primals_5, buf16, 16384, grid=grid(16384), stream=stream0)
del primals_5
buf10 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_5], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_13, reinterpret_tensor(buf9, (64, 256), (256, 1), 0), reinterpret_tensor(primals_12, (256, 4), (1, 256), 0), alpha=1, beta=1, out=buf10)
del primals_13
buf11 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf13 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax, aten.threshold_backward]
triton_poi_fused__log_softmax_threshold_backward_2.run(buf8, buf11, buf13, 256, grid=grid(256), stream=stream0)
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
triton_poi_fused__log_softmax_3.run(buf11, buf12, 256, grid=grid(256), stream=stream0)
del buf11
return (reinterpret_tensor(buf10, (4, 4, 4, 4), (64, 16, 4, 1), 0), buf12, buf8, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf1, (64, 256), (256, 1), 0), reinterpret_tensor(buf4, (64, 256), (256, 1), 0), reinterpret_tensor(buf6, (64, 256), (256, 1), 0), reinterpret_tensor(buf9, (64, 256), (256, 1), 0), buf12, primals_12, buf13, primals_10, buf14, primals_8, buf15, buf16, primals_4, buf17, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((256, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((256, 256), (256, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((256, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((256, 256), (256, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, 256), (256, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, 256), (256, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.data
class FC_Q(nn.Module):
def __init__(self, state_dim, num_actions):
super(FC_Q, self).__init__()
self.q1 = nn.Linear(state_dim, 256)
self.q2 = nn.Linear(256, 256)
self.q3 = nn.Linear(256, num_actions)
self.i1 = nn.Linear(state_dim, 256)
self.i2 = nn.Linear(256, 256)
self.i3 = nn.Linear(256, num_actions)
def forward(self, state):
q = F.relu(self.q1(state))
q = F.relu(self.q2(q))
i = F.relu(self.i1(state))
i = F.relu(self.i2(i))
i = F.relu(self.i3(i))
return self.q3(q), F.log_softmax(i, dim=1), i
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'state_dim': 4, 'num_actions': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 256
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, None)
tl.store(out_ptr0 + x2, tmp6, None)
@triton.jit
def triton_poi_fused_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused__log_softmax_threshold_backward_2(in_ptr0, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = 0.0
tmp10 = tmp0 <= tmp9
tl.store(out_ptr0 + x3, tmp8, xmask)
tl.store(out_ptr1 + x3, tmp10, xmask)
@triton.jit
def triton_poi_fused__log_softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tl.store(out_ptr0 + x3, tmp13, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13) = args
args.clear()
assert_size_stride(primals_1, (256, 4), (4, 1))
assert_size_stride(primals_2, (256,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (256, 256), (256, 1))
assert_size_stride(primals_5, (256,), (1,))
assert_size_stride(primals_6, (256, 4), (4, 1))
assert_size_stride(primals_7, (256,), (1,))
assert_size_stride(primals_8, (256, 256), (256, 1))
assert_size_stride(primals_9, (256,), (1,))
assert_size_stride(primals_10, (4, 256), (256, 1))
assert_size_stride(primals_11, (4,), (1,))
assert_size_stride(primals_12, (4, 256), (256, 1))
assert_size_stride(primals_13, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 256), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 256), (4096, 1024, 256, 1), 0
)
del buf0
buf17 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1),
torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(16384)](buf1,
primals_2, buf17, 16384, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf1, (64, 256), (256, 1), 0),
reinterpret_tensor(primals_4, (256, 256), (1, 256), 0), out=buf2)
buf3 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 256), (1, 4), 0), out=buf3)
del primals_6
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 256), (4096, 1024, 256, 1), 0
)
del buf3
buf15 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1),
torch.bool)
triton_poi_fused_relu_threshold_backward_0[grid(16384)](buf4,
primals_7, buf15, 16384, XBLOCK=128, num_warps=4, num_stages=1)
del primals_7
buf5 = empty_strided_cuda((64, 256), (256, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf4, (64, 256), (256, 1), 0),
reinterpret_tensor(primals_8, (256, 256), (1, 256), 0), out=buf5)
buf6 = reinterpret_tensor(buf5, (4, 4, 4, 256), (4096, 1024, 256, 1), 0
)
del buf5
buf14 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1),
torch.bool)
triton_poi_fused_relu_threshold_backward_0[grid(16384)](buf6,
primals_9, buf14, 16384, XBLOCK=128, num_warps=4, num_stages=1)
del primals_9
buf7 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf6, (64, 256), (256, 1), 0),
reinterpret_tensor(primals_10, (256, 4), (1, 256), 0), out=buf7)
buf8 = reinterpret_tensor(buf7, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf7
triton_poi_fused_relu_1[grid(256)](buf8, primals_11, 256, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_11
buf9 = reinterpret_tensor(buf2, (4, 4, 4, 256), (4096, 1024, 256, 1), 0
)
del buf2
buf16 = empty_strided_cuda((4, 4, 4, 256), (4096, 1024, 256, 1),
torch.bool)
triton_poi_fused_relu_threshold_backward_0[grid(16384)](buf9,
primals_5, buf16, 16384, XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
buf10 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_13, reinterpret_tensor(buf9, (64, 256),
(256, 1), 0), reinterpret_tensor(primals_12, (256, 4), (1, 256),
0), alpha=1, beta=1, out=buf10)
del primals_13
buf11 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf13 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused__log_softmax_threshold_backward_2[grid(256)](buf8,
buf11, buf13, 256, XBLOCK=256, num_warps=4, num_stages=1)
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__log_softmax_3[grid(256)](buf11, buf12, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del buf11
return (reinterpret_tensor(buf10, (4, 4, 4, 4), (64, 16, 4, 1), 0),
buf12, buf8, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(buf1, (64, 256), (256, 1), 0),
reinterpret_tensor(buf4, (64, 256), (256, 1), 0),
reinterpret_tensor(buf6, (64, 256), (256, 1), 0),
reinterpret_tensor(buf9, (64, 256), (256, 1), 0), buf12, primals_12,
buf13, primals_10, buf14, primals_8, buf15, buf16, primals_4, buf17)
class FC_QNew(nn.Module):
def __init__(self, state_dim, num_actions):
super(FC_QNew, self).__init__()
self.q1 = nn.Linear(state_dim, 256)
self.q2 = nn.Linear(256, 256)
self.q3 = nn.Linear(256, num_actions)
self.i1 = nn.Linear(state_dim, 256)
self.i2 = nn.Linear(256, 256)
self.i3 = nn.Linear(256, num_actions)
def forward(self, input_0):
primals_1 = self.q1.weight
primals_2 = self.q1.bias
primals_4 = self.q2.weight
primals_5 = self.q2.bias
primals_10 = self.q3.weight
primals_11 = self.q3.bias
primals_6 = self.i1.weight
primals_7 = self.i1.bias
primals_8 = self.i2.weight
primals_9 = self.i2.bias
primals_12 = self.i3.weight
primals_13 = self.i3.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13])
return output[0], output[1], output[2]
|
Thibaud-Ardoin/d4rl_evaluations
|
FC_Q
| false | 14,499 |
[
"Apache-2.0"
] | 123 |
135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
https://github.com/Thibaud-Ardoin/d4rl_evaluations/tree/135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
MultiNonLinearClassifier
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/z5/cz5bgdo2gmhnnmtf6w7lrjkvliacxo7nomq7mbmjquxqyxqgt5bj.py
# Topologically Sorted Source Nodes: [features_output1_1], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# features_output1_1 => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 2), (2, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 2), (32, 8, 2, 1), 0); del buf0 # reuse
buf3 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.bool)
# Topologically Sorted Source Nodes: [features_output1_1], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf3, 128, grid=grid(128), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [features_output2], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 2), (2, 1), 0), reinterpret_tensor(primals_4, (2, 4), (1, 2), 0), alpha=1, beta=1, out=buf2)
del primals_5
return (reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf1, (64, 2), (2, 1), 0), primals_4, buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((2, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 2), (2, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class MultiNonLinearClassifier(nn.Module):
def __init__(self, hidden_size, num_label):
super(MultiNonLinearClassifier, self).__init__()
self.num_label = num_label
self.classifier1 = nn.Linear(hidden_size, int(hidden_size / 2))
self.classifier2 = nn.Linear(int(hidden_size / 2), num_label)
def forward(self, input_features):
features_output1 = self.classifier1(input_features)
features_output1 = nn.ReLU()(features_output1)
features_output2 = self.classifier2(features_output1)
return features_output2
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'hidden_size': 4, 'num_label': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 2), (2, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 2), (32, 8, 2, 1), 0)
del buf0
buf3 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(128)](buf1,
primals_2, buf3, 128, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 2), (
2, 1), 0), reinterpret_tensor(primals_4, (2, 4), (1, 2), 0),
alpha=1, beta=1, out=buf2)
del primals_5
return reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), reinterpret_tensor(buf1, (64, 2), (2, 1), 0), primals_4, buf3
class MultiNonLinearClassifierNew(nn.Module):
def __init__(self, hidden_size, num_label):
super(MultiNonLinearClassifierNew, self).__init__()
self.num_label = num_label
self.classifier1 = nn.Linear(hidden_size, int(hidden_size / 2))
self.classifier2 = nn.Linear(int(hidden_size / 2), num_label)
def forward(self, input_0):
primals_1 = self.classifier1.weight
primals_2 = self.classifier1.bias
primals_4 = self.classifier2.weight
primals_5 = self.classifier2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
TimSYQQX/glyce
|
MultiNonLinearClassifier
| false | 14,500 |
[
"Apache-2.0"
] | 396 |
1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
https://github.com/TimSYQQX/glyce/tree/1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
Sentence_Maxpool
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/kd/ckd6j3xpikv6r3ocngw2xi3cx6l5ljory3eztuudmnwbkjnurqi5.py
# Topologically Sorted Source Nodes: [max_1, x_2], Original ATen: [aten.max, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# max_1 => max_1
# x_2 => relu
# Graph fragment:
# %max_1 : [num_users=2] = call_function[target=torch.ops.aten.max.dim](args = (%view_1, 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%getitem,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_max_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_max_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i64', 2: '*fp32', 3: '*i1', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_relu_threshold_backward_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_relu_threshold_backward_0(in_ptr0, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = (xindex // 16)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (64*x1)), xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0 + (64*x1)), xmask)
tmp17 = tl.load(in_ptr0 + (32 + x0 + (64*x1)), xmask)
tmp32 = tl.load(in_ptr0 + (48 + x0 + (64*x1)), xmask)
tmp2 = tmp0 > tmp1
tmp3 = tmp0 == tmp1
tmp4 = tmp0 != tmp0
tmp5 = tmp1 != tmp1
tmp6 = tmp4 > tmp5
tmp7 = tmp2 | tmp6
tmp8 = tmp4 & tmp5
tmp9 = tmp3 | tmp8
tmp10 = tl.full([1], 0, tl.int64)
tmp11 = tl.full([1], 1, tl.int64)
tmp12 = tmp10 < tmp11
tmp13 = tmp9 & tmp12
tmp14 = tmp7 | tmp13
tmp15 = tl.where(tmp14, tmp0, tmp1)
tmp16 = tl.where(tmp14, tmp10, tmp11)
tmp18 = tmp15 > tmp17
tmp19 = tmp15 == tmp17
tmp20 = tmp15 != tmp15
tmp21 = tmp17 != tmp17
tmp22 = tmp20 > tmp21
tmp23 = tmp18 | tmp22
tmp24 = tmp20 & tmp21
tmp25 = tmp19 | tmp24
tmp26 = tl.full([1], 2, tl.int64)
tmp27 = tmp16 < tmp26
tmp28 = tmp25 & tmp27
tmp29 = tmp23 | tmp28
tmp30 = tl.where(tmp29, tmp15, tmp17)
tmp31 = tl.where(tmp29, tmp16, tmp26)
tmp33 = tmp30 > tmp32
tmp34 = tmp30 == tmp32
tmp35 = tmp30 != tmp30
tmp36 = tmp32 != tmp32
tmp37 = tmp35 > tmp36
tmp38 = tmp33 | tmp37
tmp39 = tmp35 & tmp36
tmp40 = tmp34 | tmp39
tmp41 = tl.full([1], 3, tl.int64)
tmp42 = tmp31 < tmp41
tmp43 = tmp40 & tmp42
tmp44 = tmp38 | tmp43
tmp45 = tl.where(tmp44, tmp30, tmp32)
tmp46 = tl.where(tmp44, tmp31, tmp41)
tmp47 = triton_helpers.maximum(tmp0, tmp1)
tmp48 = triton_helpers.maximum(tmp47, tmp17)
tmp49 = triton_helpers.maximum(tmp48, tmp32)
tmp50 = tl.full([1], 0, tl.int32)
tmp51 = triton_helpers.maximum(tmp50, tmp49)
tmp52 = 0.0
tmp53 = tmp51 <= tmp52
tl.store(out_ptr0 + (x2), tmp46, xmask)
tl.store(out_ptr1 + (x2), tmp51, xmask)
tl.store(out_ptr2 + (x2), tmp53, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.int64)
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [max_1, x_2], Original ATen: [aten.max, aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_max_relu_threshold_backward_0.run(buf0, buf1, buf2, buf3, 64, grid=grid(64), stream=stream0)
del buf0
return (buf2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf3, reinterpret_tensor(buf1, (4, 1, 4, 4), (16, 16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
class Sentence_Maxpool(nn.Module):
""" Utilitary for the answer module """
def __init__(self, word_dimension, output_dim, relu=True):
super(Sentence_Maxpool, self).__init__()
self.fc = nn.Linear(word_dimension, output_dim)
self.out_dim = output_dim
self.relu = relu
def forward(self, x_in):
x = self.fc(x_in)
x = torch.max(x, dim=1)[0]
if self.relu:
x = F.relu(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'word_dimension': 4, 'output_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_max_relu_threshold_backward_0(in_ptr0, out_ptr0,
out_ptr1, out_ptr2, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = xindex // 16
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 64 * x1), xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0 + 64 * x1), xmask)
tmp17 = tl.load(in_ptr0 + (32 + x0 + 64 * x1), xmask)
tmp32 = tl.load(in_ptr0 + (48 + x0 + 64 * x1), xmask)
tmp2 = tmp0 > tmp1
tmp3 = tmp0 == tmp1
tmp4 = tmp0 != tmp0
tmp5 = tmp1 != tmp1
tmp6 = tmp4 > tmp5
tmp7 = tmp2 | tmp6
tmp8 = tmp4 & tmp5
tmp9 = tmp3 | tmp8
tmp10 = tl.full([1], 0, tl.int64)
tmp11 = tl.full([1], 1, tl.int64)
tmp12 = tmp10 < tmp11
tmp13 = tmp9 & tmp12
tmp14 = tmp7 | tmp13
tmp15 = tl.where(tmp14, tmp0, tmp1)
tmp16 = tl.where(tmp14, tmp10, tmp11)
tmp18 = tmp15 > tmp17
tmp19 = tmp15 == tmp17
tmp20 = tmp15 != tmp15
tmp21 = tmp17 != tmp17
tmp22 = tmp20 > tmp21
tmp23 = tmp18 | tmp22
tmp24 = tmp20 & tmp21
tmp25 = tmp19 | tmp24
tmp26 = tl.full([1], 2, tl.int64)
tmp27 = tmp16 < tmp26
tmp28 = tmp25 & tmp27
tmp29 = tmp23 | tmp28
tmp30 = tl.where(tmp29, tmp15, tmp17)
tmp31 = tl.where(tmp29, tmp16, tmp26)
tmp33 = tmp30 > tmp32
tmp34 = tmp30 == tmp32
tmp35 = tmp30 != tmp30
tmp36 = tmp32 != tmp32
tmp37 = tmp35 > tmp36
tmp38 = tmp33 | tmp37
tmp39 = tmp35 & tmp36
tmp40 = tmp34 | tmp39
tmp41 = tl.full([1], 3, tl.int64)
tmp42 = tmp31 < tmp41
tmp43 = tmp40 & tmp42
tmp44 = tmp38 | tmp43
tl.where(tmp44, tmp30, tmp32)
tmp46 = tl.where(tmp44, tmp31, tmp41)
tmp47 = triton_helpers.maximum(tmp0, tmp1)
tmp48 = triton_helpers.maximum(tmp47, tmp17)
tmp49 = triton_helpers.maximum(tmp48, tmp32)
tmp50 = tl.full([1], 0, tl.int32)
tmp51 = triton_helpers.maximum(tmp50, tmp49)
tmp52 = 0.0
tmp53 = tmp51 <= tmp52
tl.store(out_ptr0 + x2, tmp46, xmask)
tl.store(out_ptr1 + x2, tmp51, xmask)
tl.store(out_ptr2 + x2, tmp53, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.int64)
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_max_relu_threshold_backward_0[grid(64)](buf0, buf1,
buf2, buf3, 64, XBLOCK=64, num_warps=1, num_stages=1)
del buf0
return buf2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf3, reinterpret_tensor(buf1, (4, 1, 4, 4), (16, 16, 4, 1), 0)
class Sentence_MaxpoolNew(nn.Module):
""" Utilitary for the answer module """
def __init__(self, word_dimension, output_dim, relu=True):
super(Sentence_MaxpoolNew, self).__init__()
self.fc = nn.Linear(word_dimension, output_dim)
self.out_dim = output_dim
self.relu = relu
def forward(self, input_0):
primals_1 = self.fc.weight
primals_2 = self.fc.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Tiamat-Tech/just-ask
|
Sentence_Maxpool
| false | 14,501 |
[
"Apache-2.0"
] | 59 |
80725161e12ad0682b4c2091f61a5889a335ba21
|
https://github.com/Tiamat-Tech/just-ask/tree/80725161e12ad0682b4c2091f61a5889a335ba21
|
AbsLayer
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/gn/cgnakgpc2ihihojtlb466su5scbp6ziuoraworb454o4qbzwpgnf.py
# Topologically Sorted Source Nodes: [abs_1], Original ATen: [aten.abs]
# Source node to ATen node mapping:
# abs_1 => abs_1
# Graph fragment:
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%arg0_1,), kwargs = {})
triton_poi_fused_abs_0 = async_compile.triton('triton_poi_fused_abs_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_abs_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_abs_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl_math.abs(tmp0)
tl.store(out_ptr0 + (x0), tmp1, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [abs_1], Original ATen: [aten.abs]
stream0 = get_raw_stream(0)
triton_poi_fused_abs_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (reinterpret_tensor(buf0, (256, 1), (1, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
from torch.nn import Module
import torch
from torch import Tensor
from torch.nn.modules import Module
import torch.optim.lr_scheduler
class AbsLayer(Module):
def forward(self, x: 'Tensor') ->Tensor:
return torch.abs(x).reshape((-1, 1))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch.nn import Module
from torch.nn.modules import Module
import torch.optim.lr_scheduler
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_abs_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl_math.abs(tmp0)
tl.store(out_ptr0 + x0, tmp1, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_abs_0[grid(256)](arg0_1, buf0, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del arg0_1
return reinterpret_tensor(buf0, (256, 1), (1, 1), 0),
class AbsLayerNew(Module):
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
TomVeniat/avalanche
|
AbsLayer
| false | 14,502 |
[
"MIT"
] | 810 |
6e89f9945cf40c14471406a4cf4830a8d95c5705
|
https://github.com/TomVeniat/avalanche/tree/6e89f9945cf40c14471406a4cf4830a8d95c5705
|
ActNormFlow
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/n3/cn3wlek7bl5jirlp73pieiarvuadnzcb6uy5z3ieztq35hnq6trv.py
# Topologically Sorted Source Nodes: [exp, mul, outputs], Original ATen: [aten.exp, aten.mul, aten.add]
# Source node to ATen node mapping:
# exp => exp
# mul => mul
# outputs => add
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%primals_2,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_1, %exp), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %primals_3), kwargs = {})
triton_poi_fused_add_exp_mul_0 = async_compile.triton('triton_poi_fused_add_exp_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_exp_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_exp_mul_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr2 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp3 = tmp0 * tmp2
tmp5 = tmp3 + tmp4
tl.store(out_ptr0 + (x2), tmp5, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ra/crazihruh2kzmrnyhgx3yj3nw4rsa37jhrfgej2cntpe6cyiufcw.py
# Topologically Sorted Source Nodes: [logdet, mul_1, logdet_1], Original ATen: [aten.sum, aten.mul]
# Source node to ATen node mapping:
# logdet => sum_1
# logdet_1 => mul_2
# mul_1 => full_default
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%primals_2,), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([4], 4.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%full_default, %sum_1), kwargs = {})
triton_per_fused_mul_sum_1 = async_compile.triton('triton_per_fused_mul_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 4],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mul_sum_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mul_sum_1(in_ptr0, out_ptr1, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 4
RBLOCK: tl.constexpr = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.sum(tmp1, 1)[:, None]
tmp4 = 4.0
tmp5 = tmp4 * tmp3
tl.store(out_ptr1 + (tl.broadcast_to(r0, [XBLOCK, RBLOCK])), tmp5, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [exp, mul, outputs], Original ATen: [aten.exp, aten.mul, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_exp_mul_0.run(primals_1, primals_2, primals_3, buf0, 256, grid=grid(256), stream=stream0)
del primals_3
buf2 = empty_strided_cuda((4, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [logdet, mul_1, logdet_1], Original ATen: [aten.sum, aten.mul]
triton_per_fused_mul_sum_1.run(primals_2, buf2, 1, 4, grid=grid(1), stream=stream0)
return (buf0, buf2, primals_1, primals_2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
from typing import Tuple
class Flow(nn.Module):
def __init__(self):
super(Flow, self).__init__()
def forward(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*inputs: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def inverse(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*input: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def init(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Initiate the weights according to the initial input data
:param inputs:
:param kwargs:
:return:
"""
raise NotImplementedError
class ActNormFlow(Flow):
def __init__(self, channels):
super(ActNormFlow, self).__init__()
self.channels = channels
self.register_parameter('log_scale', nn.Parameter(torch.normal(0.0,
0.05, [self.channels])))
self.register_parameter('bias', nn.Parameter(torch.zeros(self.
channels)))
def forward(self, inputs: 'torch.Tensor', input_lengths=None) ->Tuple[
torch.Tensor, torch.Tensor]:
input_shape = inputs.shape
outputs = inputs * torch.exp(self.log_scale) + self.bias
logdet = torch.sum(self.log_scale)
if input_lengths is None:
logdet = torch.ones(input_shape[0], device=inputs.device) * float(
input_shape[1]) * logdet
else:
logdet = input_lengths.float() * logdet
return outputs, logdet
def inverse(self, inputs: 'torch.Tensor', input_lengths=None, epsilon=1e-08
) ->Tuple[torch.Tensor, torch.Tensor]:
input_shape = inputs.shape
outputs = (inputs - self.bias) / (torch.exp(self.log_scale) + epsilon)
logdet = -torch.sum(self.log_scale)
if input_lengths is None:
logdet = torch.ones(input_shape[0], device=inputs.device) * float(
input_shape[1]) * logdet
else:
logdet = input_lengths.float() * logdet
return outputs, logdet
def init(self, inputs: 'torch.Tensor', input_lengths=None, init_scale=
1.0, epsilon=1e-08):
_mean = torch.mean(inputs.view(-1, self.channels), dim=0)
_std = torch.std(inputs.view(-1, self.channels), dim=0)
self.log_scale.copy_(torch.log(init_scale / (_std + epsilon)))
self.bias.copy_(-_mean / (_std + epsilon))
return self.forward(inputs, input_lengths)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channels': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
from typing import Tuple
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_exp_mul_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr2 + x0, xmask, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp3 = tmp0 * tmp2
tmp5 = tmp3 + tmp4
tl.store(out_ptr0 + x2, tmp5, xmask)
@triton.jit
def triton_per_fused_mul_sum_1(in_ptr0, out_ptr1, xnumel, rnumel, XBLOCK:
tl.constexpr):
RBLOCK: tl.constexpr = 4
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.sum(tmp1, 1)[:, None]
tmp4 = 4.0
tmp5 = tmp4 * tmp3
tl.store(out_ptr1 + tl.broadcast_to(r0, [XBLOCK, RBLOCK]), tmp5, None)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_exp_mul_0[grid(256)](primals_1, primals_2,
primals_3, buf0, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_3
buf2 = empty_strided_cuda((4,), (1,), torch.float32)
triton_per_fused_mul_sum_1[grid(1)](primals_2, buf2, 1, 4, XBLOCK=1,
num_warps=2, num_stages=1)
return buf0, buf2, primals_1, primals_2
class Flow(nn.Module):
def __init__(self):
super(Flow, self).__init__()
def forward(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*inputs: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def inverse(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Args:
*input: input [batch, *input_size]
Returns: out: Tensor [batch, *input_size], logdet: Tensor [batch]
out, the output of the flow
logdet, the log determinant of :math:`\\partial output / \\partial input`
"""
raise NotImplementedError
def init(self, *inputs, **kwargs) ->Tuple[torch.Tensor, torch.Tensor]:
"""
Initiate the weights according to the initial input data
:param inputs:
:param kwargs:
:return:
"""
raise NotImplementedError
class ActNormFlowNew(Flow):
def __init__(self, channels):
super(ActNormFlowNew, self).__init__()
self.channels = channels
self.register_parameter('log_scale', nn.Parameter(torch.normal(0.0,
0.05, [self.channels])))
self.register_parameter('bias', nn.Parameter(torch.zeros(self.
channels)))
def inverse(self, inputs: 'torch.Tensor', input_lengths=None, epsilon=1e-08
) ->Tuple[torch.Tensor, torch.Tensor]:
input_shape = inputs.shape
outputs = (inputs - self.bias) / (torch.exp(self.log_scale) + epsilon)
logdet = -torch.sum(self.log_scale)
if input_lengths is None:
logdet = torch.ones(input_shape[0], device=inputs.device) * float(
input_shape[1]) * logdet
else:
logdet = input_lengths.float() * logdet
return outputs, logdet
def init(self, inputs: 'torch.Tensor', input_lengths=None, init_scale=
1.0, epsilon=1e-08):
_mean = torch.mean(inputs.view(-1, self.channels), dim=0)
_std = torch.std(inputs.view(-1, self.channels), dim=0)
self.log_scale.copy_(torch.log(init_scale / (_std + epsilon)))
self.bias.copy_(-_mean / (_std + epsilon))
return self.forward(inputs, input_lengths)
def forward(self, input_0):
primals_2 = self.log_scale
primals_3 = self.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0], output[1]
|
Tiamat-Tech/VAENAR-TTS
|
ActNormFlow
| false | 14,503 |
[
"MIT"
] | 62 |
69b6b5be1ab5168cfd3c6ab902075638e76a3b8d
|
https://github.com/Tiamat-Tech/VAENAR-TTS/tree/69b6b5be1ab5168cfd3c6ab902075638e76a3b8d
|
MetaBilinear
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/v6/cv6odvhmmcyvquog4eo62pdliew53orxzwe2wfzampr64jy3ppa7.py
# Topologically Sorted Source Nodes: [bilinear], Original ATen: [aten.add]
# Source node to ATen node mapping:
# bilinear => add
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_2, %primals_2), kwargs = {})
triton_poi_fused_add_0 = async_compile.triton('triton_poi_fused_add_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [bilinear], Original ATen: [aten._trilinear]
buf0 = torch.ops.aten._trilinear.default(reinterpret_tensor(primals_4, (64, 4), (4, 1), 0), primals_1, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), [1, 3], [0], [1, 2], [2, 3])
del primals_1
buf1 = buf0
del buf0
buf2 = reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf1 # reuse
# Topologically Sorted Source Nodes: [bilinear], Original ATen: [aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_0.run(buf2, primals_2, 256, grid=grid(256), stream=stream0)
del primals_2
return (buf2, reinterpret_tensor(primals_4, (64, 4), (4, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import re
import torch
import warnings
from torch import nn
import torch.nn.functional as F
from collections import OrderedDict
class MetaModule(nn.Module):
"""
Base class for PyTorch meta-learning modules. These modules accept an
additional argument `params` in their `forward` method.
Notes
-----
Objects inherited from `MetaModule` are fully compatible with PyTorch
modules from `torch.nn.Module`. The argument `params` is a dictionary of
tensors, with full support of the computation graph (for differentiation).
"""
def __init__(self):
super(MetaModule, self).__init__()
self._children_modules_parameters_cache = dict()
def meta_named_parameters(self, prefix='', recurse=True):
gen = self._named_members(lambda module: module._parameters.items() if
isinstance(module, MetaModule) else [], prefix=prefix, recurse=
recurse)
for elem in gen:
yield elem
def meta_parameters(self, recurse=True):
for name, param in self.meta_named_parameters(recurse=recurse):
yield param
def get_subdict(self, params, key=None):
if params is None:
return None
all_names = tuple(params.keys())
if (key, all_names) not in self._children_modules_parameters_cache:
if key is None:
self._children_modules_parameters_cache[key, all_names
] = all_names
else:
key_escape = re.escape(key)
key_re = re.compile('^{0}\\.(.+)'.format(key_escape))
self._children_modules_parameters_cache[key, all_names] = [
key_re.sub('\\1', k) for k in all_names if key_re.match
(k) is not None]
names = self._children_modules_parameters_cache[key, all_names]
if not names:
warnings.warn(
'Module `{0}` has no parameter corresponding to the submodule named `{1}` in the dictionary `params` provided as an argument to `forward()`. Using the default parameters for this submodule. The list of the parameters in `params`: [{2}].'
.format(self.__class__.__name__, key, ', '.join(all_names)),
stacklevel=2)
return None
return OrderedDict([(name, params[f'{key}.{name}']) for name in names])
class MetaBilinear(nn.Bilinear, MetaModule):
__doc__ = nn.Bilinear.__doc__
def forward(self, input1, input2, params=None):
if params is None:
params = OrderedDict(self.named_parameters())
bias = params.get('bias', None)
return F.bilinear(input1, input2, params['weight'], bias)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in1_features': 4, 'in2_features': 4, 'out_features': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import re
import warnings
from torch import nn
from collections import OrderedDict
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = torch.ops.aten._trilinear.default(reinterpret_tensor(
primals_4, (64, 4), (4, 1), 0), primals_1, reinterpret_tensor(
primals_3, (64, 4), (4, 1), 0), [1, 3], [0], [1, 2], [2, 3])
del primals_1
buf1 = buf0
del buf0
buf2 = reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf1
get_raw_stream(0)
triton_poi_fused_add_0[grid(256)](buf2, primals_2, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del primals_2
return buf2, reinterpret_tensor(primals_4, (64, 4), (4, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0)
class MetaModule(nn.Module):
"""
Base class for PyTorch meta-learning modules. These modules accept an
additional argument `params` in their `forward` method.
Notes
-----
Objects inherited from `MetaModule` are fully compatible with PyTorch
modules from `torch.nn.Module`. The argument `params` is a dictionary of
tensors, with full support of the computation graph (for differentiation).
"""
def __init__(self):
super(MetaModule, self).__init__()
self._children_modules_parameters_cache = dict()
def meta_named_parameters(self, prefix='', recurse=True):
gen = self._named_members(lambda module: module._parameters.items() if
isinstance(module, MetaModule) else [], prefix=prefix, recurse=
recurse)
for elem in gen:
yield elem
def meta_parameters(self, recurse=True):
for name, param in self.meta_named_parameters(recurse=recurse):
yield param
def get_subdict(self, params, key=None):
if params is None:
return None
all_names = tuple(params.keys())
if (key, all_names) not in self._children_modules_parameters_cache:
if key is None:
self._children_modules_parameters_cache[key, all_names
] = all_names
else:
key_escape = re.escape(key)
key_re = re.compile('^{0}\\.(.+)'.format(key_escape))
self._children_modules_parameters_cache[key, all_names] = [
key_re.sub('\\1', k) for k in all_names if key_re.match
(k) is not None]
names = self._children_modules_parameters_cache[key, all_names]
if not names:
warnings.warn(
'Module `{0}` has no parameter corresponding to the submodule named `{1}` in the dictionary `params` provided as an argument to `forward()`. Using the default parameters for this submodule. The list of the parameters in `params`: [{2}].'
.format(self.__class__.__name__, key, ', '.join(all_names)),
stacklevel=2)
return None
return OrderedDict([(name, params[f'{key}.{name}']) for name in names])
class MetaBilinearNew(nn.Bilinear, MetaModule):
__doc__ = nn.Bilinear.__doc__
def forward(self, input_0, input_1):
primals_1 = self.weight
primals_2 = self.bias
primals_3 = input_0
primals_4 = input_1
output = call([primals_1, primals_2, primals_3, primals_4])
return output[0]
|
Timothy102/light-field-networks
|
MetaBilinear
| false | 14,504 |
[
"MIT"
] | 95 |
0d2d6099ea1df4332b173fab47e5606d579b4293
|
https://github.com/Timothy102/light-field-networks/tree/0d2d6099ea1df4332b173fab47e5606d579b4293
|
EncoderLayer
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/fz/cfzmg4qtw6jgry4nhlwopodzjz62ll3n3ykfox77hwd2crdnlh2w.py
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# softmax => exp
# Graph fragment:
# %mul_tensor_3 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%bmm, 1), kwargs = {})
# %amax_default_3 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_tensor_3, [-1], True), kwargs = {})
# %sub_tensor_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_tensor_3, %amax_default_3), kwargs = {})
# %div_tensor_3 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_tensor_3, 2.0), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%div_tensor_3,), kwargs = {})
triton_poi_fused__softmax_0 = async_compile.triton('triton_poi_fused__softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp3 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = 0.5
tmp16 = tmp14 * tmp15
tmp17 = tl_math.exp(tmp16)
tl.store(out_ptr0 + (x2), tmp17, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/kj/ckjtlefzavjukjsytvkak6ek26zmzexpcbnlwelx4k5kascjxlf3.py
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# softmax => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/mk/cmkim2hc4ksxhatli3y5cu7hoqofxcbzqrrxvnlhmswdt4cgww25.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%bmm_1, %bmm_3, %bmm_5, %bmm_7], -1), kwargs = {})
triton_poi_fused_cat_2 = async_compile.triton('triton_poi_fused_cat_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x1), tmp9 & xmask, eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x1), tmp14 & xmask, eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tmp17 = tl.full([1], 4, tl.int64)
tmp18 = tmp0 < tmp17
tmp19 = tl.load(in_ptr3 + (x1), tmp16 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + (x2), tmp22, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/tc/ctcugqu2nbqlxcf2thnspnnypxifbalbzmclmutd5vaxdes2oyyk.py
# Topologically Sorted Source Nodes: [add, layer_norm], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# add => add
# layer_norm => var_mean
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %view_31), kwargs = {})
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [2]), kwargs = {correction: 0, keepdim: True})
triton_poi_fused_add_native_layer_norm_3 = async_compile.triton('triton_poi_fused_add_native_layer_norm_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + (x0), tmp16, xmask)
tl.store(out_ptr1 + (x0), tmp28, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ua/cuaquah4oaz43nhi25wixnpzlhvf2zfzdiezhmmwkuy5wfhtw6z4.py
# Topologically Sorted Source Nodes: [add, layer_norm], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# add => add
# layer_norm => add_1, add_2, mul, mul_1, rsqrt, sub_4
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %view_31), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_12, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_1,), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %getitem_13), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_4, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_6), kwargs = {})
# %add_2 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_7), kwargs = {})
triton_poi_fused_add_native_layer_norm_4 = async_compile.triton('triton_poi_fused_add_native_layer_norm_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-05
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + (x2), tmp13, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/va/cvayouropyisaprtjrhemadbdvsels72axdjsrgmbayknhu335yc.py
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# relu => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_33,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_5 = async_compile.triton('triton_poi_fused_relu_threshold_backward_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_5', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_5(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/dg/cdg2dxfjk7prchu44e4cgkid2y4524hl5vpyijgt6dwrnsrwzz2k.py
# Topologically Sorted Source Nodes: [add_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# add_1 => add_3
# Graph fragment:
# %add_3 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_2, %view_35), kwargs = {})
triton_poi_fused_add_6 = async_compile.triton('triton_poi_fused_add_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_6', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_6(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_out_ptr0 + (x2), xmask)
tmp2 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ly/clyngaczex3aq23lq4vvbajq6ig7vbayovzffexoltgvkl4mnmwv.py
# Topologically Sorted Source Nodes: [layer_norm_1], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# layer_norm_1 => add_4, rsqrt_1, var_mean_1
# Graph fragment:
# %var_mean_1 : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add_3, [2]), kwargs = {correction: 0, keepdim: True})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_14, 1e-05), kwargs = {})
# %rsqrt_1 : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_4,), kwargs = {})
triton_poi_fused_native_layer_norm_7 = async_compile.triton('triton_poi_fused_native_layer_norm_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_7(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ev/cevwu4ha52ebsczlu4ai4awtkxr6r5mjdoyil2av5baz2emhshoo.py
# Topologically Sorted Source Nodes: [layer_norm_1], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# layer_norm_1 => add_4, add_5, mul_2, mul_3, rsqrt_1, sub_5, var_mean_1
# Graph fragment:
# %var_mean_1 : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add_3, [2]), kwargs = {correction: 0, keepdim: True})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_14, 1e-05), kwargs = {})
# %rsqrt_1 : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_4,), kwargs = {})
# %sub_5 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_3, %getitem_15), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_5, %rsqrt_1), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, %primals_12), kwargs = {})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_3, %primals_13), kwargs = {})
triton_poi_fused_native_layer_norm_8 = async_compile.triton('triton_poi_fused_native_layer_norm_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_8(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, 4), (4, 1))
assert_size_stride(primals_9, (4, ), (1, ))
assert_size_stride(primals_10, (4, 4), (4, 1))
assert_size_stride(primals_11, (4, ), (1, ))
assert_size_stride(primals_12, (4, ), (1, ))
assert_size_stride(primals_13, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [query], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [key], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [value], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
del primals_4
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [dot_products], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 0), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__softmax_0.run(buf3, buf4, 64, grid=grid(64), stream=stream0)
buf5 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf4, buf5, 64, grid=grid(64), stream=stream0)
buf6 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_1], Original ATen: [aten.bmm]
extern_kernels.bmm(buf5, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 0), out=buf6)
buf7 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [dot_products_1], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 1), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 1), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax_1], Original ATen: [aten._softmax]
triton_poi_fused__softmax_0.run(buf7, buf8, 64, grid=grid(64), stream=stream0)
buf9 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [softmax_1], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf8, buf9, 64, grid=grid(64), stream=stream0)
buf10 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_3], Original ATen: [aten.bmm]
extern_kernels.bmm(buf9, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 1), out=buf10)
buf11 = buf8; del buf8 # reuse
# Topologically Sorted Source Nodes: [dot_products_2], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 2), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 2), out=buf11)
buf12 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_0.run(buf11, buf12, 64, grid=grid(64), stream=stream0)
buf13 = buf11; del buf11 # reuse
# Topologically Sorted Source Nodes: [softmax_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf12, buf13, 64, grid=grid(64), stream=stream0)
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_5], Original ATen: [aten.bmm]
extern_kernels.bmm(buf13, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 2), out=buf14)
buf15 = buf12; del buf12 # reuse
# Topologically Sorted Source Nodes: [dot_products_3], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1), 3), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 3), out=buf15)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_0.run(buf15, buf16, 64, grid=grid(64), stream=stream0)
buf17 = buf15; del buf15 # reuse
# Topologically Sorted Source Nodes: [softmax_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf16, buf17, 64, grid=grid(64), stream=stream0)
buf18 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_7], Original ATen: [aten.bmm]
extern_kernels.bmm(buf17, reinterpret_tensor(buf2, (4, 4, 1), (16, 4, 1), 3), out=buf18)
buf19 = buf16; del buf16 # reuse
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
triton_poi_fused_cat_2.run(buf6, buf10, buf14, buf18, buf19, 64, grid=grid(64), stream=stream0)
del buf10
del buf14
buf20 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf19, (16, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf20)
buf21 = reinterpret_tensor(buf6, (4, 4, 1), (4, 1, 16), 0); del buf6 # reuse
buf22 = reinterpret_tensor(buf18, (4, 4, 1), (4, 1, 16), 0); del buf18 # reuse
# Topologically Sorted Source Nodes: [add, layer_norm], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_3.run(primals_2, buf20, buf21, buf22, 16, grid=grid(16), stream=stream0)
buf23 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add, layer_norm], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_4.run(primals_2, buf20, buf21, buf22, primals_6, primals_7, buf23, 64, grid=grid(64), stream=stream0)
del primals_7
buf24 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf23, (16, 4), (4, 1), 0), reinterpret_tensor(primals_8, (4, 4), (1, 4), 0), out=buf24)
buf25 = reinterpret_tensor(buf24, (4, 4, 4), (16, 4, 1), 0); del buf24 # reuse
buf31 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_5.run(buf25, primals_9, buf31, 64, grid=grid(64), stream=stream0)
del primals_9
buf26 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf25, (16, 4), (4, 1), 0), reinterpret_tensor(primals_10, (4, 4), (1, 4), 0), out=buf26)
buf27 = reinterpret_tensor(buf26, (4, 4, 4), (16, 4, 1), 0); del buf26 # reuse
# Topologically Sorted Source Nodes: [add_1], Original ATen: [aten.add]
triton_poi_fused_add_6.run(buf27, buf23, primals_11, 64, grid=grid(64), stream=stream0)
del primals_11
buf28 = buf22; del buf22 # reuse
buf29 = buf21; del buf21 # reuse
# Topologically Sorted Source Nodes: [layer_norm_1], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_7.run(buf27, buf28, buf29, 16, grid=grid(16), stream=stream0)
buf30 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [layer_norm_1], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_8.run(buf27, buf28, buf29, primals_12, primals_13, buf30, 64, grid=grid(64), stream=stream0)
del buf28
del buf29
del primals_13
return (buf30, primals_2, primals_6, primals_12, buf5, buf9, buf13, buf17, reinterpret_tensor(buf19, (16, 4), (4, 1), 0), buf20, reinterpret_tensor(buf23, (16, 4), (4, 1), 0), reinterpret_tensor(buf25, (16, 4), (4, 1), 0), buf27, primals_10, buf31, primals_8, primals_5, reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 3), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 3), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 3), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 2), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 2), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 2), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 1), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 1), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 1), reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 0), reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 0), reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
from torch.nn import functional as F
from torch import nn
def matmul(x, y):
if x.dim() == y.dim():
return torch.matmul(x, y)
if x.dim() == y.dim() - 1:
return torch.matmul(x.unsqueeze(-2), y).squeeze(-2)
return torch.matmul(x, y.unsqueeze(-2)).squeeze(-2)
class FeedForward(nn.Module):
def __init__(self, d_model, d_hidden):
super(FeedForward, self).__init__()
self.linear1 = nn.Linear(d_model, d_hidden)
self.linear2 = nn.Linear(d_hidden, d_model)
def forward(self, x):
return self.linear2(F.relu(self.linear1(x)))
class ResidualBlock(nn.Module):
def __init__(self, layer, d_model, drop_ratio):
super(ResidualBlock, self).__init__()
self.layer = layer
self.dropout = nn.Dropout(drop_ratio)
self.layernorm = nn.LayerNorm(d_model)
def forward(self, *x):
return self.layernorm(x[0] + self.dropout(self.layer(*x)))
class Attention(nn.Module):
def __init__(self, d_key, drop_ratio, causal):
super(Attention, self).__init__()
self.scale = math.sqrt(d_key)
self.dropout = nn.Dropout(drop_ratio)
self.causal = causal
def forward(self, query, key, value):
dot_products = matmul(query, key.transpose(1, 2))
if query.dim() == 3 and (self is None or self.causal):
tri = torch.ones(key.size(1), key.size(1)).triu(1) * INF
if key.is_cuda:
tri = tri
dot_products.data.sub_(tri.unsqueeze(0))
return matmul(self.dropout(F.softmax(dot_products / self.scale, dim
=-1)), value)
class MultiHead(nn.Module):
def __init__(self, d_key, d_value, n_heads, drop_ratio, causal=False):
super(MultiHead, self).__init__()
self.attention = Attention(d_key, drop_ratio, causal=causal)
self.wq = nn.Linear(d_key, d_key, bias=False)
self.wk = nn.Linear(d_key, d_key, bias=False)
self.wv = nn.Linear(d_value, d_value, bias=False)
self.wo = nn.Linear(d_value, d_key, bias=False)
self.n_heads = n_heads
def forward(self, query, key, value):
query, key, value = self.wq(query), self.wk(key), self.wv(value)
query, key, value = (x.chunk(self.n_heads, -1) for x in (query, key,
value))
return self.wo(torch.cat([self.attention(q, k, v) for q, k, v in
zip(query, key, value)], -1))
class EncoderLayer(nn.Module):
def __init__(self, d_model, d_hidden, n_heads, drop_ratio):
super(EncoderLayer, self).__init__()
self.selfattn = ResidualBlock(MultiHead(d_model, d_model, n_heads,
drop_ratio), d_model, drop_ratio)
self.feedforward = ResidualBlock(FeedForward(d_model, d_hidden),
d_model, drop_ratio)
def forward(self, x):
return self.feedforward(self.selfattn(x, x, x))
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'d_model': 4, 'd_hidden': 4, 'n_heads': 4, 'drop_ratio': 0.5}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import math
from torch.nn import functional as F
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp3 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = 0.5
tmp16 = tmp14 * tmp15
tmp17 = tl_math.exp(tmp16)
tl.store(out_ptr0 + x2, tmp17, xmask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_cat_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = x0
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + x1, tmp4 & xmask, eviction_policy='evict_last',
other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + x1, tmp9 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + x1, tmp14 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tl.full([1], 4, tl.int64)
tmp19 = tl.load(in_ptr3 + x1, tmp16 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + x2, tmp22, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_3(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp12 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + x0, tmp16, xmask)
tl.store(out_ptr1 + x0, tmp28, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_4(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-05
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + x2, tmp13, xmask)
@triton.jit
def triton_poi_fused_relu_threshold_backward_5(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_add_6(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK:
tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_out_ptr0 + x2, xmask)
tmp2 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_7(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_8(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4, 4), (4, 1))
assert_size_stride(primals_9, (4,), (1,))
assert_size_stride(primals_10, (4, 4), (4, 1))
assert_size_stride(primals_11, (4,), (1,))
assert_size_stride(primals_12, (4,), (1,))
assert_size_stride(primals_13, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
del primals_4
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
0), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__softmax_0[grid(64)](buf3, buf4, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf5 = buf3
del buf3
triton_poi_fused__softmax_1[grid(64)](buf4, buf5, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf6 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf5, reinterpret_tensor(buf2, (4, 4, 1), (16, 4,
1), 0), out=buf6)
buf7 = buf4
del buf4
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
1), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 1), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_0[grid(64)](buf7, buf8, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf9 = buf7
del buf7
triton_poi_fused__softmax_1[grid(64)](buf8, buf9, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf10 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf9, reinterpret_tensor(buf2, (4, 4, 1), (16, 4,
1), 1), out=buf10)
buf11 = buf8
del buf8
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
2), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 2), out=buf11)
buf12 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_0[grid(64)](buf11, buf12, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf13 = buf11
del buf11
triton_poi_fused__softmax_1[grid(64)](buf12, buf13, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf13, reinterpret_tensor(buf2, (4, 4, 1), (16,
4, 1), 2), out=buf14)
buf15 = buf12
del buf12
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 1), (16, 4, 1),
3), reinterpret_tensor(buf1, (4, 1, 4), (16, 1, 4), 3), out=buf15)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_0[grid(64)](buf15, buf16, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf17 = buf15
del buf15
triton_poi_fused__softmax_1[grid(64)](buf16, buf17, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf18 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(buf17, reinterpret_tensor(buf2, (4, 4, 1), (16,
4, 1), 3), out=buf18)
buf19 = buf16
del buf16
triton_poi_fused_cat_2[grid(64)](buf6, buf10, buf14, buf18, buf19,
64, XBLOCK=64, num_warps=1, num_stages=1)
del buf10
del buf14
buf20 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf19, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf20)
buf21 = reinterpret_tensor(buf6, (4, 4, 1), (4, 1, 16), 0)
del buf6
buf22 = reinterpret_tensor(buf18, (4, 4, 1), (4, 1, 16), 0)
del buf18
triton_poi_fused_add_native_layer_norm_3[grid(16)](primals_2, buf20,
buf21, buf22, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf23 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_add_native_layer_norm_4[grid(64)](primals_2, buf20,
buf21, buf22, primals_6, primals_7, buf23, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_7
buf24 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf23, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_8, (4, 4), (1, 4), 0), out=buf24)
buf25 = reinterpret_tensor(buf24, (4, 4, 4), (16, 4, 1), 0)
del buf24
buf31 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
triton_poi_fused_relu_threshold_backward_5[grid(64)](buf25,
primals_9, buf31, 64, XBLOCK=64, num_warps=1, num_stages=1)
del primals_9
buf26 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf25, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_10, (4, 4), (1, 4), 0), out=buf26)
buf27 = reinterpret_tensor(buf26, (4, 4, 4), (16, 4, 1), 0)
del buf26
triton_poi_fused_add_6[grid(64)](buf27, buf23, primals_11, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_11
buf28 = buf22
del buf22
buf29 = buf21
del buf21
triton_poi_fused_native_layer_norm_7[grid(16)](buf27, buf28, buf29,
16, XBLOCK=16, num_warps=1, num_stages=1)
buf30 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_8[grid(64)](buf27, buf28, buf29,
primals_12, primals_13, buf30, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del buf28
del buf29
del primals_13
return (buf30, primals_2, primals_6, primals_12, buf5, buf9, buf13,
buf17, reinterpret_tensor(buf19, (16, 4), (4, 1), 0), buf20,
reinterpret_tensor(buf23, (16, 4), (4, 1), 0), reinterpret_tensor(
buf25, (16, 4), (4, 1), 0), buf27, primals_10, buf31, primals_8,
primals_5, reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 3),
reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 3),
reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 3),
reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 2),
reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 2),
reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 2),
reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 1),
reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 1),
reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 1),
reinterpret_tensor(buf2, (4, 1, 4), (16, 1, 4), 0),
reinterpret_tensor(buf0, (4, 1, 4), (16, 1, 4), 0),
reinterpret_tensor(buf1, (4, 4, 1), (16, 4, 1), 0))
def matmul(x, y):
if x.dim() == y.dim():
return torch.matmul(x, y)
if x.dim() == y.dim() - 1:
return torch.matmul(x.unsqueeze(-2), y).squeeze(-2)
return torch.matmul(x, y.unsqueeze(-2)).squeeze(-2)
class FeedForward(nn.Module):
def __init__(self, d_model, d_hidden):
super(FeedForward, self).__init__()
self.linear1 = nn.Linear(d_model, d_hidden)
self.linear2 = nn.Linear(d_hidden, d_model)
def forward(self, x):
return self.linear2(F.relu(self.linear1(x)))
class ResidualBlock(nn.Module):
def __init__(self, layer, d_model, drop_ratio):
super(ResidualBlock, self).__init__()
self.layer = layer
self.dropout = nn.Dropout(drop_ratio)
self.layernorm = nn.LayerNorm(d_model)
def forward(self, *x):
return self.layernorm(x[0] + self.dropout(self.layer(*x)))
class Attention(nn.Module):
def __init__(self, d_key, drop_ratio, causal):
super(Attention, self).__init__()
self.scale = math.sqrt(d_key)
self.dropout = nn.Dropout(drop_ratio)
self.causal = causal
def forward(self, query, key, value):
dot_products = matmul(query, key.transpose(1, 2))
if query.dim() == 3 and (self is None or self.causal):
tri = torch.ones(key.size(1), key.size(1)).triu(1) * INF
if key.is_cuda:
tri = tri
dot_products.data.sub_(tri.unsqueeze(0))
return matmul(self.dropout(F.softmax(dot_products / self.scale, dim
=-1)), value)
class MultiHead(nn.Module):
def __init__(self, d_key, d_value, n_heads, drop_ratio, causal=False):
super(MultiHead, self).__init__()
self.attention = Attention(d_key, drop_ratio, causal=causal)
self.wq = nn.Linear(d_key, d_key, bias=False)
self.wk = nn.Linear(d_key, d_key, bias=False)
self.wv = nn.Linear(d_value, d_value, bias=False)
self.wo = nn.Linear(d_value, d_key, bias=False)
self.n_heads = n_heads
def forward(self, query, key, value):
query, key, value = self.wq(query), self.wk(key), self.wv(value)
query, key, value = (x.chunk(self.n_heads, -1) for x in (query, key,
value))
return self.wo(torch.cat([self.attention(q, k, v) for q, k, v in
zip(query, key, value)], -1))
class EncoderLayerNew(nn.Module):
def __init__(self, d_model, d_hidden, n_heads, drop_ratio):
super(EncoderLayerNew, self).__init__()
self.selfattn = ResidualBlock(MultiHead(d_model, d_model, n_heads,
drop_ratio), d_model, drop_ratio)
self.feedforward = ResidualBlock(FeedForward(d_model, d_hidden),
d_model, drop_ratio)
def forward(self, input_0):
primals_1 = self.selfattn.layer.wq.weight
primals_3 = self.selfattn.layer.wk.weight
primals_4 = self.selfattn.layer.wv.weight
primals_5 = self.selfattn.layer.wo.weight
primals_6 = self.selfattn.layernorm.weight
primals_7 = self.selfattn.layernorm.bias
primals_8 = self.feedforward.layer.linear1.weight
primals_9 = self.feedforward.layer.linear1.bias
primals_10 = self.feedforward.layer.linear2.weight
primals_11 = self.feedforward.layer.linear2.bias
primals_12 = self.feedforward.layernorm.weight
primals_13 = self.feedforward.layernorm.bias
primals_2 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13])
return output[0]
|
TheShadow29/vognet-pytorch
|
EncoderLayer
| false | 14,505 |
[
"MIT"
] | 70 |
238e93c37cf9f03a2fd376a14760bb3d334a113d
|
https://github.com/TheShadow29/vognet-pytorch/tree/238e93c37cf9f03a2fd376a14760bb3d334a113d
|
HighwayCNN
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/qz/cqza6p5fjiie2hfiu5dfjqqugrnzziwuwxzlhzy2aa7khopxjbym.py
# Topologically Sorted Source Nodes: [gate_layer_result], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# gate_layer_result => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%view_3, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_3, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_0 = async_compile.triton('triton_poi_fused__softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x3), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/3v/c3vcvf5mcfw3jy7grlk23jx64xlbsyodas33i5qo4yxxd3nicv2m.py
# Topologically Sorted Source Nodes: [normal_layer_result, gate_layer_result, multiplyed_gate_and_normal, sub, multiplyed_gate_and_input, add], Original ATen: [aten.relu, aten._softmax, aten.mul, aten.rsub, aten.add]
# Source node to ATen node mapping:
# add => add
# gate_layer_result => div, sum_1
# multiplyed_gate_and_input => mul_1
# multiplyed_gate_and_normal => mul
# normal_layer_result => relu
# sub => sub_1
# Graph fragment:
# %relu : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%relu, %div), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %div), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %primals_3), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
triton_poi_fused__softmax_add_mul_relu_rsub_1 = async_compile.triton('triton_poi_fused__softmax_add_mul_relu_rsub_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_add_mul_relu_rsub_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 7, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_add_mul_relu_rsub_1(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr1 + (x3), xmask)
tmp15 = tl.load(in_ptr2 + (x3), xmask)
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tmp10 = tl.full([1], 0, tl.int32)
tmp11 = triton_helpers.maximum(tmp10, tmp9)
tmp12 = tmp11 * tmp8
tmp13 = 1.0
tmp14 = tmp13 - tmp8
tmp16 = tmp14 * tmp15
tmp17 = tmp12 + tmp16
tl.store(in_out_ptr0 + (x3), tmp17, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf1)
del primals_4
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [gate_layer_result], Original ATen: [aten._softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__softmax_0.run(buf1, buf2, 256, grid=grid(256), stream=stream0)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [normal_layer_result, gate_layer_result, multiplyed_gate_and_normal, sub, multiplyed_gate_and_input, add], Original ATen: [aten.relu, aten._softmax, aten.mul, aten.rsub, aten.add]
triton_poi_fused__softmax_add_mul_relu_rsub_1.run(buf4, buf2, buf0, primals_3, 256, grid=grid(256), stream=stream0)
del buf2
return (buf4, primals_3, buf0, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class HighwayCNN(nn.Module):
def __init__(self, input_size, gate_bias=-1, activation_function=nn.
functional.relu, gate_activation=nn.functional.softmax):
super(HighwayCNN, self).__init__()
self.activation_function = activation_function
self.gate_activation = gate_activation
self.normal_layer = nn.Linear(input_size, input_size)
self.gate_layer = nn.Linear(input_size, input_size)
self.gate_layer.bias.data.fill_(gate_bias)
def forward(self, x):
normal_layer_result = self.activation_function(self.normal_layer(x))
gate_layer_result = self.gate_activation(self.gate_layer(x))
multiplyed_gate_and_normal = torch.mul(normal_layer_result,
gate_layer_result)
multiplyed_gate_and_input = torch.mul(1 - gate_layer_result, x)
return torch.add(multiplyed_gate_and_normal, multiplyed_gate_and_input)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_size': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x3, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_add_mul_relu_rsub_1(in_out_ptr0, in_ptr0,
in_ptr1, in_ptr2, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr1 + x3, xmask)
tmp15 = tl.load(in_ptr2 + x3, xmask)
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tmp10 = tl.full([1], 0, tl.int32)
tmp11 = triton_helpers.maximum(tmp10, tmp9)
tmp12 = tmp11 * tmp8
tmp13 = 1.0
tmp14 = tmp13 - tmp8
tmp16 = tmp14 * tmp15
tmp17 = tmp12 + tmp16
tl.store(in_out_ptr0 + x3, tmp17, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf1)
del primals_4
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__softmax_0[grid(256)](buf1, buf2, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf4 = buf3
del buf3
triton_poi_fused__softmax_add_mul_relu_rsub_1[grid(256)](buf4, buf2,
buf0, primals_3, 256, XBLOCK=256, num_warps=4, num_stages=1)
del buf2
return buf4, primals_3, buf0, buf1
class HighwayCNNNew(nn.Module):
def __init__(self, input_size, gate_bias=-1, activation_function=nn.
functional.relu, gate_activation=nn.functional.softmax):
super(HighwayCNNNew, self).__init__()
self.activation_function = activation_function
self.gate_activation = gate_activation
self.normal_layer = nn.Linear(input_size, input_size)
self.gate_layer = nn.Linear(input_size, input_size)
self.gate_layer.bias.data.fill_(gate_bias)
def forward(self, input_0):
primals_1 = self.normal_layer.weight
primals_2 = self.normal_layer.bias
primals_4 = self.gate_layer.weight
primals_5 = self.gate_layer.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
TimSYQQX/glyce
|
HighwayCNN
| false | 14,506 |
[
"Apache-2.0"
] | 396 |
1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
https://github.com/TimSYQQX/glyce/tree/1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
BatchLinear
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/bh/cbhpcgxjg3mwo4dulstw5ie26none2yzi5sysdzl34cu6pyah4fg.py
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.add, aten.view]
# Source node to ATen node mapping:
# output_1 => add, view_3
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_1, %unsqueeze), kwargs = {})
# %view_3 : [num_users=1] = call_function[target=torch.ops.aten.reshape.default](args = (%view_2, [4, 4, 4, 4]), kwargs = {})
triton_poi_fused_add_view_0 = async_compile.triton('triton_poi_fused_add_view_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_view_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_view_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x4), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x4), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf0 # reuse
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.add, aten.view]
stream0 = get_raw_stream(0)
triton_poi_fused_add_view_0.run(buf2, primals_2, 256, grid=grid(256), stream=stream0)
del primals_2
return (buf2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import re
import torch
import warnings
from torch import nn
from collections import OrderedDict
class MetaModule(nn.Module):
"""
Base class for PyTorch meta-learning modules. These modules accept an
additional argument `params` in their `forward` method.
Notes
-----
Objects inherited from `MetaModule` are fully compatible with PyTorch
modules from `torch.nn.Module`. The argument `params` is a dictionary of
tensors, with full support of the computation graph (for differentiation).
"""
def __init__(self):
super(MetaModule, self).__init__()
self._children_modules_parameters_cache = dict()
def meta_named_parameters(self, prefix='', recurse=True):
gen = self._named_members(lambda module: module._parameters.items() if
isinstance(module, MetaModule) else [], prefix=prefix, recurse=
recurse)
for elem in gen:
yield elem
def meta_parameters(self, recurse=True):
for name, param in self.meta_named_parameters(recurse=recurse):
yield param
def get_subdict(self, params, key=None):
if params is None:
return None
all_names = tuple(params.keys())
if (key, all_names) not in self._children_modules_parameters_cache:
if key is None:
self._children_modules_parameters_cache[key, all_names
] = all_names
else:
key_escape = re.escape(key)
key_re = re.compile('^{0}\\.(.+)'.format(key_escape))
self._children_modules_parameters_cache[key, all_names] = [
key_re.sub('\\1', k) for k in all_names if key_re.match
(k) is not None]
names = self._children_modules_parameters_cache[key, all_names]
if not names:
warnings.warn(
'Module `{0}` has no parameter corresponding to the submodule named `{1}` in the dictionary `params` provided as an argument to `forward()`. Using the default parameters for this submodule. The list of the parameters in `params`: [{2}].'
.format(self.__class__.__name__, key, ', '.join(all_names)),
stacklevel=2)
return None
return OrderedDict([(name, params[f'{key}.{name}']) for name in names])
class BatchLinear(nn.Linear, MetaModule):
"""A linear meta-layer that can deal with batched weight matrices and biases, as for instance output by a
hypernetwork."""
__doc__ = nn.Linear.__doc__
def forward(self, input, params=None):
if params is None:
params = OrderedDict(self.named_parameters())
bias = params.get('bias', None)
weight = params['weight']
output = input.matmul(weight.permute(*[i for i in range(len(weight.
shape) - 2)], -1, -2))
output += bias.unsqueeze(-2)
return output
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_features': 4, 'out_features': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import re
import warnings
from torch import nn
from collections import OrderedDict
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_view_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x4, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x4, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf0
buf2 = buf1
del buf1
get_raw_stream(0)
triton_poi_fused_add_view_0[grid(256)](buf2, primals_2, 256, XBLOCK
=256, num_warps=4, num_stages=1)
del primals_2
return buf2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0)
class MetaModule(nn.Module):
"""
Base class for PyTorch meta-learning modules. These modules accept an
additional argument `params` in their `forward` method.
Notes
-----
Objects inherited from `MetaModule` are fully compatible with PyTorch
modules from `torch.nn.Module`. The argument `params` is a dictionary of
tensors, with full support of the computation graph (for differentiation).
"""
def __init__(self):
super(MetaModule, self).__init__()
self._children_modules_parameters_cache = dict()
def meta_named_parameters(self, prefix='', recurse=True):
gen = self._named_members(lambda module: module._parameters.items() if
isinstance(module, MetaModule) else [], prefix=prefix, recurse=
recurse)
for elem in gen:
yield elem
def meta_parameters(self, recurse=True):
for name, param in self.meta_named_parameters(recurse=recurse):
yield param
def get_subdict(self, params, key=None):
if params is None:
return None
all_names = tuple(params.keys())
if (key, all_names) not in self._children_modules_parameters_cache:
if key is None:
self._children_modules_parameters_cache[key, all_names
] = all_names
else:
key_escape = re.escape(key)
key_re = re.compile('^{0}\\.(.+)'.format(key_escape))
self._children_modules_parameters_cache[key, all_names] = [
key_re.sub('\\1', k) for k in all_names if key_re.match
(k) is not None]
names = self._children_modules_parameters_cache[key, all_names]
if not names:
warnings.warn(
'Module `{0}` has no parameter corresponding to the submodule named `{1}` in the dictionary `params` provided as an argument to `forward()`. Using the default parameters for this submodule. The list of the parameters in `params`: [{2}].'
.format(self.__class__.__name__, key, ', '.join(all_names)),
stacklevel=2)
return None
return OrderedDict([(name, params[f'{key}.{name}']) for name in names])
class BatchLinearNew(nn.Linear, MetaModule):
"""A linear meta-layer that can deal with batched weight matrices and biases, as for instance output by a
hypernetwork."""
__doc__ = nn.Linear.__doc__
def forward(self, input_0):
primals_1 = self.weight
primals_2 = self.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Timothy102/light-field-networks
|
BatchLinear
| false | 14,507 |
[
"MIT"
] | 95 |
0d2d6099ea1df4332b173fab47e5606d579b4293
|
https://github.com/Timothy102/light-field-networks/tree/0d2d6099ea1df4332b173fab47e5606d579b4293
|
VNMaxPool
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/4b/c4bin6v3nraanry5cu4t7z3qycurd6acdwc4l5azpezbv2erqt7w.py
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# linear => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%permute,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex % 4
x3 = (xindex // 4)
y0 = yindex % 4
y1 = (yindex // 4)
x5 = xindex
y4 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x3) + (16*x2) + (64*y1)), xmask & ymask)
tl.store(out_ptr0 + (x5 + (16*y4)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ff/cff4pauyodt3bzn5rggprurywn64qeyzwyhkbgmdbs4wzkmbk24h.py
# Topologically Sorted Source Nodes: [mul, dotprod, max_1], Original ATen: [aten.mul, aten.sum, aten.max]
# Source node to ATen node mapping:
# dotprod => sum_1
# max_1 => max_1
# mul => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, %permute_2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [2], True), kwargs = {})
# %max_1 : [num_users=1] = call_function[target=torch.ops.aten.max.dim](args = (%sum_1, -1), kwargs = {})
triton_poi_fused_max_mul_sum_1 = async_compile.triton('triton_poi_fused_max_mul_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i64', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_mul_sum_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 32, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_mul_sum_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (16*x2), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x0 + (64*x1)), xmask)
tmp3 = tl.load(in_ptr0 + (4 + (16*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (4 + x0 + (64*x1)), xmask)
tmp7 = tl.load(in_ptr0 + (8 + (16*x2)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (8 + x0 + (64*x1)), xmask)
tmp11 = tl.load(in_ptr0 + (12 + (16*x2)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (12 + x0 + (64*x1)), xmask)
tmp15 = tl.load(in_ptr0 + (1 + (16*x2)), xmask, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr1 + (16 + x0 + (64*x1)), xmask)
tmp18 = tl.load(in_ptr0 + (5 + (16*x2)), xmask, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr1 + (20 + x0 + (64*x1)), xmask)
tmp22 = tl.load(in_ptr0 + (9 + (16*x2)), xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr1 + (24 + x0 + (64*x1)), xmask)
tmp26 = tl.load(in_ptr0 + (13 + (16*x2)), xmask, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr1 + (28 + x0 + (64*x1)), xmask)
tmp45 = tl.load(in_ptr0 + (2 + (16*x2)), xmask, eviction_policy='evict_last')
tmp46 = tl.load(in_ptr1 + (32 + x0 + (64*x1)), xmask)
tmp48 = tl.load(in_ptr0 + (6 + (16*x2)), xmask, eviction_policy='evict_last')
tmp49 = tl.load(in_ptr1 + (36 + x0 + (64*x1)), xmask)
tmp52 = tl.load(in_ptr0 + (10 + (16*x2)), xmask, eviction_policy='evict_last')
tmp53 = tl.load(in_ptr1 + (40 + x0 + (64*x1)), xmask)
tmp56 = tl.load(in_ptr0 + (14 + (16*x2)), xmask, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr1 + (44 + x0 + (64*x1)), xmask)
tmp74 = tl.load(in_ptr0 + (3 + (16*x2)), xmask, eviction_policy='evict_last')
tmp75 = tl.load(in_ptr1 + (48 + x0 + (64*x1)), xmask)
tmp77 = tl.load(in_ptr0 + (7 + (16*x2)), xmask, eviction_policy='evict_last')
tmp78 = tl.load(in_ptr1 + (52 + x0 + (64*x1)), xmask)
tmp81 = tl.load(in_ptr0 + (11 + (16*x2)), xmask, eviction_policy='evict_last')
tmp82 = tl.load(in_ptr1 + (56 + x0 + (64*x1)), xmask)
tmp85 = tl.load(in_ptr0 + (15 + (16*x2)), xmask, eviction_policy='evict_last')
tmp86 = tl.load(in_ptr1 + (60 + x0 + (64*x1)), xmask)
tmp2 = tmp0 * tmp1
tmp5 = tmp3 * tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 * tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 * tmp12
tmp14 = tmp10 + tmp13
tmp17 = tmp15 * tmp16
tmp20 = tmp18 * tmp19
tmp21 = tmp17 + tmp20
tmp24 = tmp22 * tmp23
tmp25 = tmp21 + tmp24
tmp28 = tmp26 * tmp27
tmp29 = tmp25 + tmp28
tmp30 = tmp14 > tmp29
tmp31 = tmp14 == tmp29
tmp32 = tmp14 != tmp14
tmp33 = tmp29 != tmp29
tmp34 = tmp32 > tmp33
tmp35 = tmp30 | tmp34
tmp36 = tmp32 & tmp33
tmp37 = tmp31 | tmp36
tmp38 = tl.full([1], 0, tl.int64)
tmp39 = tl.full([1], 1, tl.int64)
tmp40 = tmp38 < tmp39
tmp41 = tmp37 & tmp40
tmp42 = tmp35 | tmp41
tmp43 = tl.where(tmp42, tmp14, tmp29)
tmp44 = tl.where(tmp42, tmp38, tmp39)
tmp47 = tmp45 * tmp46
tmp50 = tmp48 * tmp49
tmp51 = tmp47 + tmp50
tmp54 = tmp52 * tmp53
tmp55 = tmp51 + tmp54
tmp58 = tmp56 * tmp57
tmp59 = tmp55 + tmp58
tmp60 = tmp43 > tmp59
tmp61 = tmp43 == tmp59
tmp62 = tmp43 != tmp43
tmp63 = tmp59 != tmp59
tmp64 = tmp62 > tmp63
tmp65 = tmp60 | tmp64
tmp66 = tmp62 & tmp63
tmp67 = tmp61 | tmp66
tmp68 = tl.full([1], 2, tl.int64)
tmp69 = tmp44 < tmp68
tmp70 = tmp67 & tmp69
tmp71 = tmp65 | tmp70
tmp72 = tl.where(tmp71, tmp43, tmp59)
tmp73 = tl.where(tmp71, tmp44, tmp68)
tmp76 = tmp74 * tmp75
tmp79 = tmp77 * tmp78
tmp80 = tmp76 + tmp79
tmp83 = tmp81 * tmp82
tmp84 = tmp80 + tmp83
tmp87 = tmp85 * tmp86
tmp88 = tmp84 + tmp87
tmp89 = tmp72 > tmp88
tmp90 = tmp72 == tmp88
tmp91 = tmp72 != tmp72
tmp92 = tmp88 != tmp88
tmp93 = tmp91 > tmp92
tmp94 = tmp89 | tmp93
tmp95 = tmp91 & tmp92
tmp96 = tmp90 | tmp95
tmp97 = tl.full([1], 3, tl.int64)
tmp98 = tmp73 < tmp97
tmp99 = tmp96 & tmp98
tmp100 = tmp94 | tmp99
tmp101 = tl.where(tmp100, tmp72, tmp88)
tmp102 = tl.where(tmp100, tmp73, tmp97)
tl.store(out_ptr0 + (x2), tmp102, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/hh/chhmjvtnydhywuctlrzgz3hln27ndcn643agxfpeht6gnhljobu2.py
# Topologically Sorted Source Nodes: [x_max], Original ATen: [aten.index]
# Source node to ATen node mapping:
# x_max => index
# Graph fragment:
# %index : [num_users=1] = call_function[target=torch.ops.aten.index.Tensor](args = (%arg0_1, [%expand, %expand_1, %expand_2, %getitem_1]), kwargs = {})
triton_poi_fused_index_2 = async_compile.triton('triton_poi_fused_index_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_index_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_index_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = (xindex // 4)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x3), xmask, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 4, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tl.device_assert(((0 <= tmp4) & (tmp4 < 4)) | ~(xmask), "index out of bounds: 0 <= tmp4 < 4")
tmp6 = tl.load(in_ptr1 + (tmp4 + (4*x4)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x4), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(arg0_1, buf0, 16, 16, grid=grid(16, 16), stream=stream0)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (64, 4), (4, 1), 0), reinterpret_tensor(arg1_1, (4, 4), (1, 4), 0), out=buf1)
del arg1_1
del buf0
buf2 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.int64)
# Topologically Sorted Source Nodes: [mul, dotprod, max_1], Original ATen: [aten.mul, aten.sum, aten.max]
triton_poi_fused_max_mul_sum_1.run(arg0_1, buf1, buf2, 16, grid=grid(16), stream=stream0)
del buf1
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_max], Original ATen: [aten.index]
triton_poi_fused_index_2.run(buf2, arg0_1, buf3, 64, grid=grid(64), stream=stream0)
del arg0_1
del buf2
return (buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.utils.data
import torch
import torch.nn.parallel
class VNMaxPool(nn.Module):
def __init__(self, in_channels):
super(VNMaxPool, self).__init__()
self.map_to_dir = nn.Linear(in_channels, in_channels, bias=False)
def forward(self, x):
"""
x: point features of shape [B, N_feat, 3, N_samples, ...]
"""
d = self.map_to_dir(x.transpose(1, -1)).transpose(1, -1)
dotprod = (x * d).sum(2, keepdims=True)
idx = dotprod.max(dim=-1, keepdim=False)[1]
index_tuple = torch.meshgrid([torch.arange(j) for j in x.size()[:-1]]
) + (idx,)
x_max = x[index_tuple]
return x_max
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
import torch.utils.data
import torch
import torch.nn.parallel
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex % 4
x3 = xindex // 4
y0 = yindex % 4
y1 = yindex // 4
x5 = xindex
y4 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x3 + 16 * x2 + 64 * y1), xmask & ymask)
tl.store(out_ptr0 + (x5 + 16 * y4), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_max_mul_sum_1(in_ptr0, in_ptr1, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + 16 * x2, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x0 + 64 * x1), xmask)
tmp3 = tl.load(in_ptr0 + (4 + 16 * x2), xmask, eviction_policy='evict_last'
)
tmp4 = tl.load(in_ptr1 + (4 + x0 + 64 * x1), xmask)
tmp7 = tl.load(in_ptr0 + (8 + 16 * x2), xmask, eviction_policy='evict_last'
)
tmp8 = tl.load(in_ptr1 + (8 + x0 + 64 * x1), xmask)
tmp11 = tl.load(in_ptr0 + (12 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp12 = tl.load(in_ptr1 + (12 + x0 + 64 * x1), xmask)
tmp15 = tl.load(in_ptr0 + (1 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp16 = tl.load(in_ptr1 + (16 + x0 + 64 * x1), xmask)
tmp18 = tl.load(in_ptr0 + (5 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp19 = tl.load(in_ptr1 + (20 + x0 + 64 * x1), xmask)
tmp22 = tl.load(in_ptr0 + (9 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp23 = tl.load(in_ptr1 + (24 + x0 + 64 * x1), xmask)
tmp26 = tl.load(in_ptr0 + (13 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp27 = tl.load(in_ptr1 + (28 + x0 + 64 * x1), xmask)
tmp45 = tl.load(in_ptr0 + (2 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp46 = tl.load(in_ptr1 + (32 + x0 + 64 * x1), xmask)
tmp48 = tl.load(in_ptr0 + (6 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp49 = tl.load(in_ptr1 + (36 + x0 + 64 * x1), xmask)
tmp52 = tl.load(in_ptr0 + (10 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp53 = tl.load(in_ptr1 + (40 + x0 + 64 * x1), xmask)
tmp56 = tl.load(in_ptr0 + (14 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp57 = tl.load(in_ptr1 + (44 + x0 + 64 * x1), xmask)
tmp74 = tl.load(in_ptr0 + (3 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp75 = tl.load(in_ptr1 + (48 + x0 + 64 * x1), xmask)
tmp77 = tl.load(in_ptr0 + (7 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp78 = tl.load(in_ptr1 + (52 + x0 + 64 * x1), xmask)
tmp81 = tl.load(in_ptr0 + (11 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp82 = tl.load(in_ptr1 + (56 + x0 + 64 * x1), xmask)
tmp85 = tl.load(in_ptr0 + (15 + 16 * x2), xmask, eviction_policy=
'evict_last')
tmp86 = tl.load(in_ptr1 + (60 + x0 + 64 * x1), xmask)
tmp2 = tmp0 * tmp1
tmp5 = tmp3 * tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 * tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 * tmp12
tmp14 = tmp10 + tmp13
tmp17 = tmp15 * tmp16
tmp20 = tmp18 * tmp19
tmp21 = tmp17 + tmp20
tmp24 = tmp22 * tmp23
tmp25 = tmp21 + tmp24
tmp28 = tmp26 * tmp27
tmp29 = tmp25 + tmp28
tmp30 = tmp14 > tmp29
tmp31 = tmp14 == tmp29
tmp32 = tmp14 != tmp14
tmp33 = tmp29 != tmp29
tmp34 = tmp32 > tmp33
tmp35 = tmp30 | tmp34
tmp36 = tmp32 & tmp33
tmp37 = tmp31 | tmp36
tmp38 = tl.full([1], 0, tl.int64)
tmp39 = tl.full([1], 1, tl.int64)
tmp40 = tmp38 < tmp39
tmp41 = tmp37 & tmp40
tmp42 = tmp35 | tmp41
tmp43 = tl.where(tmp42, tmp14, tmp29)
tmp44 = tl.where(tmp42, tmp38, tmp39)
tmp47 = tmp45 * tmp46
tmp50 = tmp48 * tmp49
tmp51 = tmp47 + tmp50
tmp54 = tmp52 * tmp53
tmp55 = tmp51 + tmp54
tmp58 = tmp56 * tmp57
tmp59 = tmp55 + tmp58
tmp60 = tmp43 > tmp59
tmp61 = tmp43 == tmp59
tmp62 = tmp43 != tmp43
tmp63 = tmp59 != tmp59
tmp64 = tmp62 > tmp63
tmp65 = tmp60 | tmp64
tmp66 = tmp62 & tmp63
tmp67 = tmp61 | tmp66
tmp68 = tl.full([1], 2, tl.int64)
tmp69 = tmp44 < tmp68
tmp70 = tmp67 & tmp69
tmp71 = tmp65 | tmp70
tmp72 = tl.where(tmp71, tmp43, tmp59)
tmp73 = tl.where(tmp71, tmp44, tmp68)
tmp76 = tmp74 * tmp75
tmp79 = tmp77 * tmp78
tmp80 = tmp76 + tmp79
tmp83 = tmp81 * tmp82
tmp84 = tmp80 + tmp83
tmp87 = tmp85 * tmp86
tmp88 = tmp84 + tmp87
tmp89 = tmp72 > tmp88
tmp90 = tmp72 == tmp88
tmp91 = tmp72 != tmp72
tmp92 = tmp88 != tmp88
tmp93 = tmp91 > tmp92
tmp94 = tmp89 | tmp93
tmp95 = tmp91 & tmp92
tmp96 = tmp90 | tmp95
tmp97 = tl.full([1], 3, tl.int64)
tmp98 = tmp73 < tmp97
tmp99 = tmp96 & tmp98
tmp100 = tmp94 | tmp99
tl.where(tmp100, tmp72, tmp88)
tmp102 = tl.where(tmp100, tmp73, tmp97)
tl.store(out_ptr0 + x2, tmp102, xmask)
@triton.jit
def triton_poi_fused_index_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex // 4
x4 = xindex
tmp0 = tl.load(in_ptr0 + x3, xmask, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 4, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tl.device_assert((0 <= tmp4) & (tmp4 < 4) | ~xmask,
'index out of bounds: 0 <= tmp4 < 4')
tmp6 = tl.load(in_ptr1 + (tmp4 + 4 * x4), xmask, eviction_policy=
'evict_last')
tl.store(out_ptr0 + x4, tmp6, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(16, 16)](arg0_1, buf0, 16, 16, XBLOCK
=16, YBLOCK=16, num_warps=4, num_stages=1)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (64, 4), (4, 1), 0),
reinterpret_tensor(arg1_1, (4, 4), (1, 4), 0), out=buf1)
del arg1_1
del buf0
buf2 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.int64)
triton_poi_fused_max_mul_sum_1[grid(16)](arg0_1, buf1, buf2, 16,
XBLOCK=16, num_warps=1, num_stages=1)
del buf1
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_index_2[grid(64)](buf2, arg0_1, buf3, 64, XBLOCK=
64, num_warps=1, num_stages=1)
del arg0_1
del buf2
return buf3,
class VNMaxPoolNew(nn.Module):
def __init__(self, in_channels):
super(VNMaxPoolNew, self).__init__()
self.map_to_dir = nn.Linear(in_channels, in_channels, bias=False)
def forward(self, input_0):
arg1_1 = self.map_to_dir.weight
arg0_1 = input_0
output = call([arg0_1, arg1_1])
return output[0]
|
Tiamat-Tech/vnn
|
VNMaxPool
| false | 14,508 |
[
"MIT"
] | 280 |
f3197e210022b5f0015e0da6456adf66bd0cd73e
|
https://github.com/Tiamat-Tech/vnn/tree/f3197e210022b5f0015e0da6456adf66bd0cd73e
|
SoftL1
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/3g/c3grf6coinyl3okv26iopeivr4nol5nlifsiefysroxq5sjccqcy.py
# Topologically Sorted Source Nodes: [sub, l1, ret, ret_1, mean], Original ATen: [aten.sub, aten.abs, aten.clamp, aten.mean]
# Source node to ATen node mapping:
# l1 => abs_1
# mean => mean
# ret => sub_1
# ret_1 => clamp_max, clamp_min
# sub => sub
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %arg1_1), kwargs = {})
# %abs_1 : [num_users=2] = call_function[target=torch.ops.aten.abs.default](args = (%sub,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%abs_1, 0.0), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub_1, 0.0), kwargs = {})
# %clamp_max : [num_users=1] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min, 100.0), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_1,), kwargs = {})
triton_per_fused_abs_clamp_mean_sub_0 = async_compile.triton('triton_per_fused_abs_clamp_mean_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {4: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 5), equal_to_1=(4,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_abs_clamp_mean_sub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_abs_clamp_mean_sub_0(in_out_ptr0, in_ptr0, in_ptr1, out_ptr0, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = tl.load(in_ptr1 + (r0), None)
tmp2 = tmp0 - tmp1
tmp3 = tl_math.abs(tmp2)
tmp4 = 0.0
tmp5 = tmp3 - tmp4
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = 100.0
tmp8 = triton_helpers.minimum(tmp6, tmp7)
tmp9 = tl.broadcast_to(tmp3, [RBLOCK])
tmp11 = triton_helpers.promote_to_tensor(tl.sum(tmp9, 0))
tmp12 = 256.0
tmp13 = tmp11 / tmp12
tl.store(out_ptr0 + (tl.broadcast_to(r0, [RBLOCK])), tmp8, None)
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp13, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [sub, l1, ret, ret_1, mean], Original ATen: [aten.sub, aten.abs, aten.clamp, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_abs_clamp_mean_sub_0.run(buf2, arg0_1, arg1_1, buf0, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf0, buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
class SoftL1(torch.nn.Module):
def __init__(self):
super(SoftL1, self).__init__()
def forward(self, input, target, eps=0.0):
l1 = torch.abs(input - target)
ret = l1 - eps
ret = torch.clamp(ret, min=0.0, max=100.0)
return ret, torch.mean(l1.detach())
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_abs_clamp_mean_sub_0(in_out_ptr0, in_ptr0, in_ptr1,
out_ptr0, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = tl.load(in_ptr1 + r0, None)
tmp2 = tmp0 - tmp1
tmp3 = tl_math.abs(tmp2)
tmp4 = 0.0
tmp5 = tmp3 - tmp4
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = 100.0
tmp8 = triton_helpers.minimum(tmp6, tmp7)
tmp9 = tl.broadcast_to(tmp3, [RBLOCK])
tmp11 = triton_helpers.promote_to_tensor(tl.sum(tmp9, 0))
tmp12 = 256.0
tmp13 = tmp11 / tmp12
tl.store(out_ptr0 + tl.broadcast_to(r0, [RBLOCK]), tmp8, None)
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp13, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1
del buf1
get_raw_stream(0)
triton_per_fused_abs_clamp_mean_sub_0[grid(1)](buf2, arg0_1, arg1_1,
buf0, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf0, buf2
class SoftL1New(torch.nn.Module):
def __init__(self):
super(SoftL1New, self).__init__()
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0], output[1]
|
Tiamat-Tech/npms
|
SoftL1
| false | 14,509 |
[
"MIT"
] | 96 |
2d1bce8c98b0f24aa69273975c52b2fbdb101c29
|
https://github.com/Tiamat-Tech/npms/tree/2d1bce8c98b0f24aa69273975c52b2fbdb101c29
|
Correlation
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/hl/chlb6erzxkaz5er2ywfwkvvgncd76mgltyvsn22epwb3ucdjvihq.py
# Topologically Sorted Source Nodes: [mul, sum_1, loss], Original ATen: [aten.mul, aten.sum, aten.mean]
# Source node to ATen node mapping:
# loss => mean
# mul => mul
# sum_1 => sum_1
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%slice_1, %slice_2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_1,), kwargs = {})
triton_per_fused_mean_mul_sum_0 = async_compile.triton('triton_per_fused_mean_mul_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mean_mul_sum_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 16, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mean_mul_sum_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 48
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = rindex < rnumel
r0 = rindex % 16
r1 = (rindex // 16)
r2 = rindex
tmp0 = tl.load(in_ptr0 + (r0 + (64*r1)), rmask, other=0.0)
tmp1 = tl.load(in_ptr1 + (r0 + (64*r1)), rmask, other=0.0)
tmp4 = tl.load(in_ptr0 + (64 + r0 + (64*r1)), rmask, other=0.0)
tmp5 = tl.load(in_ptr1 + (64 + r0 + (64*r1)), rmask, other=0.0)
tmp9 = tl.load(in_ptr0 + (16 + r0 + (64*r1)), rmask, other=0.0)
tmp10 = tl.load(in_ptr1 + (16 + r0 + (64*r1)), rmask, other=0.0)
tmp13 = tl.load(in_ptr0 + (80 + r0 + (64*r1)), rmask, other=0.0)
tmp14 = tl.load(in_ptr1 + (80 + r0 + (64*r1)), rmask, other=0.0)
tmp19 = tl.load(in_ptr0 + (32 + r0 + (64*r1)), rmask, other=0.0)
tmp20 = tl.load(in_ptr1 + (32 + r0 + (64*r1)), rmask, other=0.0)
tmp23 = tl.load(in_ptr0 + (96 + r0 + (64*r1)), rmask, other=0.0)
tmp24 = tl.load(in_ptr1 + (96 + r0 + (64*r1)), rmask, other=0.0)
tmp29 = tl.load(in_ptr0 + (48 + r0 + (64*r1)), rmask, other=0.0)
tmp30 = tl.load(in_ptr1 + (48 + r0 + (64*r1)), rmask, other=0.0)
tmp33 = tl.load(in_ptr0 + (112 + r0 + (64*r1)), rmask, other=0.0)
tmp34 = tl.load(in_ptr1 + (112 + r0 + (64*r1)), rmask, other=0.0)
tmp2 = tmp0 - tmp1
tmp3 = tl_math.abs(tmp2)
tmp6 = tmp4 - tmp5
tmp7 = tl_math.abs(tmp6)
tmp8 = tmp3 * tmp7
tmp11 = tmp9 - tmp10
tmp12 = tl_math.abs(tmp11)
tmp15 = tmp13 - tmp14
tmp16 = tl_math.abs(tmp15)
tmp17 = tmp12 * tmp16
tmp18 = tmp8 + tmp17
tmp21 = tmp19 - tmp20
tmp22 = tl_math.abs(tmp21)
tmp25 = tmp23 - tmp24
tmp26 = tl_math.abs(tmp25)
tmp27 = tmp22 * tmp26
tmp28 = tmp18 + tmp27
tmp31 = tmp29 - tmp30
tmp32 = tl_math.abs(tmp31)
tmp35 = tmp33 - tmp34
tmp36 = tl_math.abs(tmp35)
tmp37 = tmp32 * tmp36
tmp38 = tmp28 + tmp37
tmp39 = tl.broadcast_to(tmp38, [XBLOCK, RBLOCK])
tmp41 = tl.where(rmask, tmp39, 0)
tmp42 = tl.sum(tmp41, 1)[:, None]
tmp43 = 48.0
tmp44 = tmp42 / tmp43
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp44, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [mul, sum_1, loss], Original ATen: [aten.mul, aten.sum, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_mean_mul_sum_0.run(buf2, arg0_1, arg1_1, 1, 48, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class Correlation(nn.Module):
"""Correlation Congruence for Knowledge Distillation, ICCV 2019.
The authors nicely shared the code with me. I restructured their code to be
compatible with my running framework. Credits go to the original author"""
def __init__(self):
super(Correlation, self).__init__()
def forward(self, f_s, f_t):
delta = torch.abs(f_s - f_t)
loss = torch.mean((delta[:-1] * delta[1:]).sum(1))
return loss
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_mean_mul_sum_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel,
rnumel, XBLOCK: tl.constexpr):
rnumel = 48
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
rmask = rindex < rnumel
r0 = rindex % 16
r1 = rindex // 16
tmp0 = tl.load(in_ptr0 + (r0 + 64 * r1), rmask, other=0.0)
tmp1 = tl.load(in_ptr1 + (r0 + 64 * r1), rmask, other=0.0)
tmp4 = tl.load(in_ptr0 + (64 + r0 + 64 * r1), rmask, other=0.0)
tmp5 = tl.load(in_ptr1 + (64 + r0 + 64 * r1), rmask, other=0.0)
tmp9 = tl.load(in_ptr0 + (16 + r0 + 64 * r1), rmask, other=0.0)
tmp10 = tl.load(in_ptr1 + (16 + r0 + 64 * r1), rmask, other=0.0)
tmp13 = tl.load(in_ptr0 + (80 + r0 + 64 * r1), rmask, other=0.0)
tmp14 = tl.load(in_ptr1 + (80 + r0 + 64 * r1), rmask, other=0.0)
tmp19 = tl.load(in_ptr0 + (32 + r0 + 64 * r1), rmask, other=0.0)
tmp20 = tl.load(in_ptr1 + (32 + r0 + 64 * r1), rmask, other=0.0)
tmp23 = tl.load(in_ptr0 + (96 + r0 + 64 * r1), rmask, other=0.0)
tmp24 = tl.load(in_ptr1 + (96 + r0 + 64 * r1), rmask, other=0.0)
tmp29 = tl.load(in_ptr0 + (48 + r0 + 64 * r1), rmask, other=0.0)
tmp30 = tl.load(in_ptr1 + (48 + r0 + 64 * r1), rmask, other=0.0)
tmp33 = tl.load(in_ptr0 + (112 + r0 + 64 * r1), rmask, other=0.0)
tmp34 = tl.load(in_ptr1 + (112 + r0 + 64 * r1), rmask, other=0.0)
tmp2 = tmp0 - tmp1
tmp3 = tl_math.abs(tmp2)
tmp6 = tmp4 - tmp5
tmp7 = tl_math.abs(tmp6)
tmp8 = tmp3 * tmp7
tmp11 = tmp9 - tmp10
tmp12 = tl_math.abs(tmp11)
tmp15 = tmp13 - tmp14
tmp16 = tl_math.abs(tmp15)
tmp17 = tmp12 * tmp16
tmp18 = tmp8 + tmp17
tmp21 = tmp19 - tmp20
tmp22 = tl_math.abs(tmp21)
tmp25 = tmp23 - tmp24
tmp26 = tl_math.abs(tmp25)
tmp27 = tmp22 * tmp26
tmp28 = tmp18 + tmp27
tmp31 = tmp29 - tmp30
tmp32 = tl_math.abs(tmp31)
tmp35 = tmp33 - tmp34
tmp36 = tl_math.abs(tmp35)
tmp37 = tmp32 * tmp36
tmp38 = tmp28 + tmp37
tmp39 = tl.broadcast_to(tmp38, [XBLOCK, RBLOCK])
tmp41 = tl.where(rmask, tmp39, 0)
tmp42 = tl.sum(tmp41, 1)[:, None]
tmp43 = 48.0
tmp44 = tmp42 / tmp43
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp44, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1
del buf1
get_raw_stream(0)
triton_per_fused_mean_mul_sum_0[grid(1)](buf2, arg0_1, arg1_1, 1,
48, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf2,
class CorrelationNew(nn.Module):
"""Correlation Congruence for Knowledge Distillation, ICCV 2019.
The authors nicely shared the code with me. I restructured their code to be
compatible with my running framework. Credits go to the original author"""
def __init__(self):
super(CorrelationNew, self).__init__()
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
UBCDingXin/RepDistiller
|
Correlation
| false | 14,510 |
[
"BSD-2-Clause"
] | 1,347 |
dcc043277f2820efafd679ffb82b8e8195b7e222
|
https://github.com/UBCDingXin/RepDistiller/tree/dcc043277f2820efafd679ffb82b8e8195b7e222
|
HighwayMLP
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/3n/c3nyedrc56xoj6pmjzzgnpithkx2vti6qsjnj43ybcoj67zutjs4.py
# Topologically Sorted Source Nodes: [gate_layer_result], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# gate_layer_result => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%view_3, [0], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_3, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_0 = async_compile.triton('triton_poi_fused__softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (64 + x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (128 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (192 + x0), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/br/cbr3h3nns6lgik3wmozi3gef4zd4j625tqhix5j3w7mqvislf7ye.py
# Topologically Sorted Source Nodes: [normal_layer_result, gate_layer_result, multiplyed_gate_and_normal, sub, multiplyed_gate_and_input, add], Original ATen: [aten.relu, aten._softmax, aten.mul, aten.rsub, aten.add]
# Source node to ATen node mapping:
# add => add
# gate_layer_result => div, sum_1
# multiplyed_gate_and_input => mul_1
# multiplyed_gate_and_normal => mul
# normal_layer_result => relu
# sub => sub_1
# Graph fragment:
# %relu : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [0], True), kwargs = {})
# %div : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%relu, %div), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %div), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %primals_3), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
triton_poi_fused__softmax_add_mul_relu_rsub_1 = async_compile.triton('triton_poi_fused__softmax_add_mul_relu_rsub_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_add_mul_relu_rsub_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 7, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_add_mul_relu_rsub_1(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (64 + x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (128 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (192 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr1 + (x2), xmask)
tmp15 = tl.load(in_ptr2 + (x2), xmask)
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tmp10 = tl.full([1], 0, tl.int32)
tmp11 = triton_helpers.maximum(tmp10, tmp9)
tmp12 = tmp11 * tmp8
tmp13 = 1.0
tmp14 = tmp13 - tmp8
tmp16 = tmp14 * tmp15
tmp17 = tmp12 + tmp16
tl.store(in_out_ptr0 + (x2), tmp17, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf1)
del primals_4
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [gate_layer_result], Original ATen: [aten._softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__softmax_0.run(buf1, buf2, 256, grid=grid(256), stream=stream0)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [normal_layer_result, gate_layer_result, multiplyed_gate_and_normal, sub, multiplyed_gate_and_input, add], Original ATen: [aten.relu, aten._softmax, aten.mul, aten.rsub, aten.add]
triton_poi_fused__softmax_add_mul_relu_rsub_1.run(buf4, buf2, buf0, primals_3, 256, grid=grid(256), stream=stream0)
del buf2
return (buf4, primals_3, buf0, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class HighwayMLP(nn.Module):
def __init__(self, input_size, gate_bias=-2, activation_function=nn.
functional.relu, gate_activation=nn.functional.softmax):
super(HighwayMLP, self).__init__()
self.activation_function = activation_function
self.gate_activation = gate_activation
self.normal_layer = nn.Linear(input_size, input_size)
self.gate_layer = nn.Linear(input_size, input_size)
self.gate_layer.bias.data.fill_(gate_bias)
def forward(self, x):
normal_layer_result = self.activation_function(self.normal_layer(x))
gate_layer_result = self.gate_activation(self.gate_layer(x), dim=0)
multiplyed_gate_and_normal = torch.mul(normal_layer_result,
gate_layer_result)
multiplyed_gate_and_input = torch.mul(1 - gate_layer_result, x)
return torch.add(multiplyed_gate_and_normal, multiplyed_gate_and_input)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_size': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (64 + x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (128 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (192 + x0), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_add_mul_relu_rsub_1(in_out_ptr0, in_ptr0,
in_ptr1, in_ptr2, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (64 + x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (128 + x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (192 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr1 + x2, xmask)
tmp15 = tl.load(in_ptr2 + x2, xmask)
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tmp10 = tl.full([1], 0, tl.int32)
tmp11 = triton_helpers.maximum(tmp10, tmp9)
tmp12 = tmp11 * tmp8
tmp13 = 1.0
tmp14 = tmp13 - tmp8
tmp16 = tmp14 * tmp15
tmp17 = tmp12 + tmp16
tl.store(in_out_ptr0 + x2, tmp17, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf1)
del primals_4
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__softmax_0[grid(256)](buf1, buf2, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf4 = buf3
del buf3
triton_poi_fused__softmax_add_mul_relu_rsub_1[grid(256)](buf4, buf2,
buf0, primals_3, 256, XBLOCK=256, num_warps=4, num_stages=1)
del buf2
return buf4, primals_3, buf0, buf1
class HighwayMLPNew(nn.Module):
def __init__(self, input_size, gate_bias=-2, activation_function=nn.
functional.relu, gate_activation=nn.functional.softmax):
super(HighwayMLPNew, self).__init__()
self.activation_function = activation_function
self.gate_activation = gate_activation
self.normal_layer = nn.Linear(input_size, input_size)
self.gate_layer = nn.Linear(input_size, input_size)
self.gate_layer.bias.data.fill_(gate_bias)
def forward(self, input_0):
primals_1 = self.normal_layer.weight
primals_2 = self.normal_layer.bias
primals_4 = self.gate_layer.weight
primals_5 = self.gate_layer.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
TimSYQQX/glyce
|
HighwayMLP
| false | 14,511 |
[
"Apache-2.0"
] | 396 |
1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
https://github.com/TimSYQQX/glyce/tree/1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
DeConv
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/oj/cojl5mb3pzv5jbmfzjkbac5hekbmpvb72kof6ouyyasitrogdd6n.py
# Topologically Sorted Source Nodes: [h], Original ATen: [aten._unsafe_index]
# Source node to ATen node mapping:
# h => _unsafe_index
# Graph fragment:
# %_unsafe_index : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%primals_1, [None, None, %unsqueeze, %convert_element_type_1]), kwargs = {})
triton_poi_fused__unsafe_index_0 = async_compile.triton('triton_poi_fused__unsafe_index_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__unsafe_index_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__unsafe_index_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 8) % 8
x0 = xindex % 8
x2 = (xindex // 64)
x4 = xindex
tmp0 = x1
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tmp5 = x0
tmp6 = tmp5.to(tl.float32)
tmp7 = tmp6 * tmp2
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.load(in_ptr0 + (tmp8 + (4*tmp4) + (16*x2)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x4), tmp9, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3, 3), (36, 9, 3, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 8, 8), (256, 64, 8, 1), torch.float32)
# Topologically Sorted Source Nodes: [h], Original ATen: [aten._unsafe_index]
stream0 = get_raw_stream(0)
triton_poi_fused__unsafe_index_0.run(primals_1, buf0, 1024, grid=grid(1024), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [h_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 8, 8), (256, 64, 8, 1))
return (buf1, primals_2, buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
import torch.onnx
class DeConv(nn.Module):
def __init__(self, in_channels, out_channels, kernel_size=3,
upsampl_scale=2):
super().__init__()
self.upsampling = nn.UpsamplingNearest2d(scale_factor=upsampl_scale)
padding_size = int((kernel_size - 1) / 2)
self.conv = nn.Conv2d(in_channels=in_channels, padding=padding_size,
out_channels=out_channels, kernel_size=kernel_size, bias=False)
self.__weights_init__()
def __weights_init__(self):
nn.init.xavier_uniform_(self.conv.weight)
def forward(self, input):
h = self.upsampling(input)
h = self.conv(h)
return h
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch import nn
import torch.onnx
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused__unsafe_index_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 8 % 8
x0 = xindex % 8
x2 = xindex // 64
x4 = xindex
tmp0 = x1
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tmp5 = x0
tmp6 = tmp5.to(tl.float32)
tmp7 = tmp6 * tmp2
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.load(in_ptr0 + (tmp8 + 4 * tmp4 + 16 * x2), xmask,
eviction_policy='evict_last')
tl.store(out_ptr0 + x4, tmp9, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3, 3), (36, 9, 3, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 8, 8), (256, 64, 8, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__unsafe_index_0[grid(1024)](primals_1, buf0, 1024,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 8, 8), (256, 64, 8, 1))
return buf1, primals_2, buf0
class DeConvNew(nn.Module):
def __init__(self, in_channels, out_channels, kernel_size=3,
upsampl_scale=2):
super().__init__()
self.upsampling = nn.UpsamplingNearest2d(scale_factor=upsampl_scale)
padding_size = int((kernel_size - 1) / 2)
self.conv = nn.Conv2d(in_channels=in_channels, padding=padding_size,
out_channels=out_channels, kernel_size=kernel_size, bias=False)
self.__weights_init__()
def __weights_init__(self):
nn.init.xavier_uniform_(self.conv.weight)
def forward(self, input_0):
primals_2 = self.conv.weight
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0]
|
TriceHelix/ASMAGAN
|
DeConv
| false | 14,512 |
[
"Apache-2.0"
] | 121 |
6e2b5b587f88f641fdcc05a81cf5f0b4d6a9f3e1
|
https://github.com/TriceHelix/ASMAGAN/tree/6e2b5b587f88f641fdcc05a81cf5f0b4d6a9f3e1
|
CustomizeLayer
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ne/cnem3ewvyvy2ry3vceibv7j67edhj5jen3qlzvzibg5tpkhxkwsq.py
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, x, mul, add], Original ATen: [aten.pow, aten.sum, aten.sqrt, aten.div, aten.mul, aten.add]
# Source node to ATen node mapping:
# add => add
# mul => mul
# norm => sqrt
# pow_1 => pow_1
# sum_1 => sum_1
# x => div
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%primals_1, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [1], True), kwargs = {})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%sum_1,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%primals_1, %sqrt), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %primals_2), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %primals_3), kwargs = {})
triton_poi_fused_add_div_mul_pow_sqrt_sum_0 = async_compile.triton('triton_poi_fused_add_div_mul_pow_sqrt_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_pow_sqrt_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 7, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mul_pow_sqrt_sum_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x3 = (xindex // 64)
x5 = xindex % 16
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x4), xmask)
tmp1 = tl.load(in_ptr0 + (x5 + (64*x3)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x5 + (64*x3)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x5 + (64*x3)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x5 + (64*x3)), xmask, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr2 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = tmp0 / tmp12
tmp15 = tmp13 * tmp14
tmp17 = tmp15 + tmp16
tl.store(out_ptr0 + (x4), tmp17, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, sum_1, norm, x, mul, add], Original ATen: [aten.pow, aten.sum, aten.sqrt, aten.div, aten.mul, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mul_pow_sqrt_sum_0.run(primals_1, primals_2, primals_3, buf0, 256, grid=grid(256), stream=stream0)
del primals_2
del primals_3
return (buf0, primals_1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class CustomizeLayer(nn.Module):
def __init__(self, in_dim):
super().__init__()
self.in_dim = in_dim
self.scale = nn.Parameter(torch.Tensor(self.in_dim))
self.bias = nn.Parameter(torch.Tensor(self.in_dim))
def forward(self, x):
norm = x.pow(2).sum(dim=1, keepdim=True).sqrt()
x = torch.div(x, norm)
return x * self.scale + self.bias
def __repr__(self):
return 'CustomizedLayer(in_dim=%d)' % self.in_dim
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_dim': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_mul_pow_sqrt_sum_0(in_ptr0, in_ptr1, in_ptr2,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x4 = xindex
x3 = xindex // 64
x5 = xindex % 16
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x4, xmask)
tmp1 = tl.load(in_ptr0 + (x5 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x5 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x5 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x5 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp14 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr2 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = tmp0 / tmp12
tmp15 = tmp13 * tmp14
tmp17 = tmp15 + tmp16
tl.store(out_ptr0 + x4, tmp17, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_mul_pow_sqrt_sum_0[grid(256)](primals_1,
primals_2, primals_3, buf0, 256, XBLOCK=128, num_warps=4,
num_stages=1)
del primals_2
del primals_3
return buf0, primals_1
class CustomizeLayerNew(nn.Module):
def __init__(self, in_dim):
super().__init__()
self.in_dim = in_dim
self.scale = nn.Parameter(torch.Tensor(self.in_dim))
self.bias = nn.Parameter(torch.Tensor(self.in_dim))
def __repr__(self):
return 'CustomizedLayer(in_dim=%d)' % self.in_dim
def forward(self, input_0):
primals_2 = self.scale
primals_3 = self.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Trouble404/Torch-Pruning
|
CustomizeLayer
| false | 14,513 |
[
"MIT"
] | 468 |
80e07f66c220ac0ec52f0e19a4a71e8865d28952
|
https://github.com/Trouble404/Torch-Pruning/tree/80e07f66c220ac0ec52f0e19a4a71e8865d28952
|
MultiHeadSelfAttention
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/t4/ct4veshscig5vgvyj32uhwuca65ykr2tiu5huzxkuudzphoft4m4.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => clone_2
# Graph fragment:
# %clone_2 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%getitem_2,), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (8 + x0 + (12*x1)), xmask)
tmp1 = tl.load(in_ptr1 + (8 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ji/cjiuxsaveibxcx77pxybrxhp7ix53ahmzdmwwqw4rlhuo76kmhhj.py
# Topologically Sorted Source Nodes: [truediv], Original ATen: [aten.div]
# Source node to ATen node mapping:
# truediv => div
# Graph fragment:
# %div : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%view_5, 2.0), kwargs = {})
triton_poi_fused_div_1 = async_compile.triton('triton_poi_fused_div_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (12*x1)), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/54/c545jcidaxblcqouel6ztiwyytwzuaasohh4q2qelpa3u5sbntsf.py
# Topologically Sorted Source Nodes: [keys_1], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# keys_1 => clone_1
# Graph fragment:
# %clone_1 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%getitem_1,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_2 = async_compile.triton('triton_poi_fused_clone_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (4 + x0 + (12*x1)), xmask)
tmp1 = tl.load(in_ptr1 + (4 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/7s/c7spagnqvsgjrukyw5jujzjmswxuigeuvpyhxgdob766q2gfvgzr.py
# Topologically Sorted Source Nodes: [result], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# result => amax, exp, sub_1
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%bmm, [-1], True), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%bmm, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_1,), kwargs = {})
triton_poi_fused__softmax_3 = async_compile.triton('triton_poi_fused__softmax_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/dw/cdwqsjnh2osfmjr2utzzaqdg2vrfivzkuhareq3urgidllj2bsvr.py
# Topologically Sorted Source Nodes: [result], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# result => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_4 = async_compile.triton('triton_poi_fused__softmax_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (12, 4), (4, 1))
assert_size_stride(primals_3, (12, ), (1, ))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 12), (12, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 12), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(buf0, primals_3, buf1, 64, grid=grid(64), stream=stream0)
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [truediv], Original ATen: [aten.div]
triton_poi_fused_div_1.run(buf0, primals_3, buf2, 64, grid=grid(64), stream=stream0)
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [keys_1], Original ATen: [aten.clone]
triton_poi_fused_clone_2.run(buf0, primals_3, buf3, 64, grid=grid(64), stream=stream0)
del buf0
del primals_3
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
extern_kernels.bmm(buf2, reinterpret_tensor(buf3, (4, 4, 4), (16, 1, 4), 0), out=buf4)
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [result], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf4, buf5, 64, grid=grid(64), stream=stream0)
buf6 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [result], Original ATen: [aten._softmax]
triton_poi_fused__softmax_4.run(buf5, buf6, 64, grid=grid(64), stream=stream0)
buf7 = buf5; del buf5 # reuse
# Topologically Sorted Source Nodes: [outputs], Original ATen: [aten.bmm]
extern_kernels.bmm(buf6, buf1, out=buf7)
buf8 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [outputs_4], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(buf7, (16, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf8)
del primals_5
return (reinterpret_tensor(buf8, (4, 4, 4), (16, 4, 1), 0), reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), buf6, reinterpret_tensor(buf7, (16, 4), (4, 1), 0), primals_4, reinterpret_tensor(buf1, (4, 4, 4), (16, 1, 4), 0), reinterpret_tensor(buf2, (4, 4, 4), (16, 1, 4), 0), buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((12, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((12, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
from torch.nn import Module
import torch
from torch.nn import Dropout
from torch.nn import Linear
from torch.nn.modules import Dropout
def masked_softmax(vector: 'torch.Tensor', mask: 'torch.Tensor', dim: 'int'=-1
) ->torch.Tensor:
"""
``torch.nn.functional.softmax(vector)`` does not work if some elements of ``vector`` should be
masked. This performs a softmax on just the non-masked portions of ``vector``. Passing
``None`` in for the mask is also acceptable; you'll just get a regular softmax.
``vector`` can have an arbitrary number of dimensions; the only requirement is that ``mask`` is
broadcastable to ``vector's`` shape. If ``mask`` has fewer dimensions than ``vector``, we will
unsqueeze on dimension 1 until they match. If you need a different unsqueezing of your mask,
do it yourself before passing the mask into this function.
In the case that the input vector is completely masked, this function returns an array
of ``0.0``. This behavior may cause ``NaN`` if this is used as the last layer of a model
that uses categorical cross-entropy loss.
"""
if mask is None:
result = torch.nn.functional.softmax(vector, dim=dim)
else:
mask = mask.float()
while mask.dim() < vector.dim():
mask = mask.unsqueeze(1)
result = torch.nn.functional.softmax(vector + (1 - mask) * -
10000000000.0, dim=dim)
return result
def weighted_sum(matrix: 'torch.Tensor', attention: 'torch.Tensor'
) ->torch.Tensor:
"""
Takes a matrix of vectors and a set of weights over the rows in the matrix (which we call an
"attention" vector), and returns a weighted sum of the rows in the matrix. This is the typical
computation performed after an attention mechanism.
Note that while we call this a "matrix" of vectors and an attention "vector", we also handle
higher-order tensors. We always sum over the second-to-last dimension of the "matrix", and we
assume that all dimensions in the "matrix" prior to the last dimension are matched in the
"vector". Non-matched dimensions in the "vector" must be `directly after the batch dimension`.
For example, say I have a "matrix" with dimensions ``(batch_size, num_queries, num_words,
embedding_dim)``. The attention "vector" then must have at least those dimensions, and could
have more. Both:
- ``(batch_size, num_queries, num_words)`` (distribution over words for each query)
- ``(batch_size, num_documents, num_queries, num_words)`` (distribution over words in a
query for each document)
are valid input "vectors", producing tensors of shape:
``(batch_size, num_queries, embedding_dim)`` and
``(batch_size, num_documents, num_queries, embedding_dim)`` respectively.
"""
if attention.dim() == 2 and matrix.dim() == 3:
return attention.unsqueeze(1).bmm(matrix).squeeze(1)
if attention.dim() == 3 and matrix.dim() == 3:
return attention.bmm(matrix)
if matrix.dim() - 1 < attention.dim():
expanded_size = list(matrix.size())
for i in range(attention.dim() - matrix.dim() + 1):
matrix = matrix.unsqueeze(1)
expanded_size.insert(i + 1, attention.size(i + 1))
matrix = matrix.expand(*expanded_size)
intermediate = attention.unsqueeze(-1).expand_as(matrix) * matrix
return intermediate.sum(dim=-2)
class MultiHeadSelfAttention(Module):
"""
This class implements the key-value scaled dot product attention mechanism
detailed in the paper `Attention is all you Need
<https://www.semanticscholar.org/paper/Attention-Is-All-You-Need-Vaswani-Shazeer/0737da0767d77606169cbf4187b83e1ab62f6077>`_ .
The attention mechanism is a weighted sum of a projection V of the inputs, with respect
to the scaled, normalised dot product of Q and K, which are also both linear projections
of the input. This procedure is repeated for each attention head, using different parameters.
Parameters
----------
num_heads : ``int``, required.
The number of attention heads to use.
input_dim : ``int``, required.
The size of the last dimension of the input tensor.
attention_dim ``int``, required.
The total dimension of the query and key projections which comprise the
dot product attention function. Must be divisible by ``num_heads``.
values_dim : ``int``, required.
The total dimension which the input is projected to for representing the values,
which are combined using the attention. Must be divisible by ``num_heads``.
output_projection_dim : ``int``, optional (default = None)
The dimensionality of the final output projection. If this is not passed
explicitly, the projection has size `input_size`.
attention_dropout_prob : ``float``, optional (default = 0.1).
The dropout probability applied to the normalised attention
distributions.
"""
def __init__(self, input_dim: 'int', attention_dim: 'int', values_dim:
'int', num_heads: 'int'=1, output_projection_dim: 'int'=None,
attention_dropout_prob: 'float'=0.1) ->None:
super(MultiHeadSelfAttention, self).__init__()
self._num_heads = num_heads
self._input_dim = input_dim
self._output_dim = output_projection_dim or input_dim
self._attention_dim = attention_dim
self._values_dim = values_dim
if attention_dim % num_heads != 0:
raise ValueError(
f'Key size ({attention_dim}) must be divisible by the number of attention heads ({num_heads}).'
)
if values_dim % num_heads != 0:
raise ValueError(
f'Value size ({values_dim}) must be divisible by the number of attention heads ({num_heads}).'
)
self._combined_projection = Linear(input_dim, 2 * attention_dim +
values_dim)
self._scale = (input_dim // num_heads) ** 0.5
self._output_projection = Linear(values_dim, self._output_dim)
self._attention_dropout = Dropout(attention_dropout_prob)
def get_input_dim(self):
return self._input_dim
def get_output_dim(self):
return self._output_dim
def is_bidirectional(self):
return False
def forward(self, inputs: 'torch.Tensor', mask: 'torch.LongTensor'=None
) ->torch.FloatTensor:
"""
Parameters
----------
inputs : ``torch.FloatTensor``, required.
A tensor of shape (batch_size, timesteps, input_dim)
mask : ``torch.FloatTensor``, optional (default = None).
A tensor of shape (batch_size, timesteps).
Returns
-------
A tensor of shape (batch_size, timesteps, output_projection_dim),
where output_projection_dim = input_dim by default.
"""
num_heads = self._num_heads
batch_size, timesteps, _ = inputs.size()
if mask is None:
mask = inputs.new_ones(batch_size, timesteps)
combined_projection = self._combined_projection(inputs)
queries, keys, *values = combined_projection.split(self.
_attention_dim, -1)
queries = queries.contiguous()
keys = keys.contiguous()
values = torch.cat(values, -1).contiguous()
values_per_head = values.view(batch_size, timesteps, num_heads, int
(self._values_dim / num_heads))
values_per_head = values_per_head.transpose(1, 2).contiguous()
values_per_head = values_per_head.view(batch_size * num_heads,
timesteps, int(self._values_dim / num_heads))
queries_per_head = queries.view(batch_size, timesteps, num_heads,
int(self._attention_dim / num_heads))
queries_per_head = queries_per_head.transpose(1, 2).contiguous()
queries_per_head = queries_per_head.view(batch_size * num_heads,
timesteps, int(self._attention_dim / num_heads))
keys_per_head = keys.view(batch_size, timesteps, num_heads, int(
self._attention_dim / num_heads))
keys_per_head = keys_per_head.transpose(1, 2).contiguous()
keys_per_head = keys_per_head.view(batch_size * num_heads,
timesteps, int(self._attention_dim / num_heads))
scaled_similarities = torch.bmm(queries_per_head / self._scale,
keys_per_head.transpose(1, 2))
attention = masked_softmax(scaled_similarities, mask.repeat(1,
num_heads).view(batch_size * num_heads, timesteps))
attention = self._attention_dropout(attention)
outputs = weighted_sum(values_per_head, attention)
outputs = outputs.view(batch_size, num_heads, timesteps, int(self.
_values_dim / num_heads))
outputs = outputs.transpose(1, 2).contiguous()
outputs = outputs.view(batch_size, timesteps, self._values_dim)
outputs = self._output_projection(outputs)
return outputs
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'input_dim': 4, 'attention_dim': 4, 'values_dim': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch.nn import Module
from torch.nn import Dropout
from torch.nn import Linear
from torch.nn.modules import Dropout
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (8 + x0 + 12 * x1), xmask)
tmp1 = tl.load(in_ptr1 + (8 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_div_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 12 * x1), xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_clone_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (4 + x0 + 12 * x1), xmask)
tmp1 = tl.load(in_ptr1 + (4 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (12, 4), (4, 1))
assert_size_stride(primals_3, (12,), (1,))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 12), (12, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_2, (4, 12), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(64)](buf0, primals_3, buf1, 64, XBLOCK=
64, num_warps=1, num_stages=1)
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_div_1[grid(64)](buf0, primals_3, buf2, 64, XBLOCK=
64, num_warps=1, num_stages=1)
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_clone_2[grid(64)](buf0, primals_3, buf3, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del buf0
del primals_3
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(buf2, reinterpret_tensor(buf3, (4, 4, 4), (16, 1,
4), 0), out=buf4)
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused__softmax_3[grid(64)](buf4, buf5, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf6 = buf4
del buf4
triton_poi_fused__softmax_4[grid(64)](buf5, buf6, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf7 = buf5
del buf5
extern_kernels.bmm(buf6, buf1, out=buf7)
buf8 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(buf7, (16, 4), (
4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0),
alpha=1, beta=1, out=buf8)
del primals_5
return reinterpret_tensor(buf8, (4, 4, 4), (16, 4, 1), 0
), reinterpret_tensor(primals_1, (16, 4), (4, 1), 0
), buf6, reinterpret_tensor(buf7, (16, 4), (4, 1), 0
), primals_4, reinterpret_tensor(buf1, (4, 4, 4), (16, 1, 4), 0
), reinterpret_tensor(buf2, (4, 4, 4), (16, 1, 4), 0), buf3
def masked_softmax(vector: 'torch.Tensor', mask: 'torch.Tensor', dim: 'int'=-1
) ->torch.Tensor:
"""
``torch.nn.functional.softmax(vector)`` does not work if some elements of ``vector`` should be
masked. This performs a softmax on just the non-masked portions of ``vector``. Passing
``None`` in for the mask is also acceptable; you'll just get a regular softmax.
``vector`` can have an arbitrary number of dimensions; the only requirement is that ``mask`` is
broadcastable to ``vector's`` shape. If ``mask`` has fewer dimensions than ``vector``, we will
unsqueeze on dimension 1 until they match. If you need a different unsqueezing of your mask,
do it yourself before passing the mask into this function.
In the case that the input vector is completely masked, this function returns an array
of ``0.0``. This behavior may cause ``NaN`` if this is used as the last layer of a model
that uses categorical cross-entropy loss.
"""
if mask is None:
result = torch.nn.functional.softmax(vector, dim=dim)
else:
mask = mask.float()
while mask.dim() < vector.dim():
mask = mask.unsqueeze(1)
result = torch.nn.functional.softmax(vector + (1 - mask) * -
10000000000.0, dim=dim)
return result
def weighted_sum(matrix: 'torch.Tensor', attention: 'torch.Tensor'
) ->torch.Tensor:
"""
Takes a matrix of vectors and a set of weights over the rows in the matrix (which we call an
"attention" vector), and returns a weighted sum of the rows in the matrix. This is the typical
computation performed after an attention mechanism.
Note that while we call this a "matrix" of vectors and an attention "vector", we also handle
higher-order tensors. We always sum over the second-to-last dimension of the "matrix", and we
assume that all dimensions in the "matrix" prior to the last dimension are matched in the
"vector". Non-matched dimensions in the "vector" must be `directly after the batch dimension`.
For example, say I have a "matrix" with dimensions ``(batch_size, num_queries, num_words,
embedding_dim)``. The attention "vector" then must have at least those dimensions, and could
have more. Both:
- ``(batch_size, num_queries, num_words)`` (distribution over words for each query)
- ``(batch_size, num_documents, num_queries, num_words)`` (distribution over words in a
query for each document)
are valid input "vectors", producing tensors of shape:
``(batch_size, num_queries, embedding_dim)`` and
``(batch_size, num_documents, num_queries, embedding_dim)`` respectively.
"""
if attention.dim() == 2 and matrix.dim() == 3:
return attention.unsqueeze(1).bmm(matrix).squeeze(1)
if attention.dim() == 3 and matrix.dim() == 3:
return attention.bmm(matrix)
if matrix.dim() - 1 < attention.dim():
expanded_size = list(matrix.size())
for i in range(attention.dim() - matrix.dim() + 1):
matrix = matrix.unsqueeze(1)
expanded_size.insert(i + 1, attention.size(i + 1))
matrix = matrix.expand(*expanded_size)
intermediate = attention.unsqueeze(-1).expand_as(matrix) * matrix
return intermediate.sum(dim=-2)
class MultiHeadSelfAttentionNew(Module):
"""
This class implements the key-value scaled dot product attention mechanism
detailed in the paper `Attention is all you Need
<https://www.semanticscholar.org/paper/Attention-Is-All-You-Need-Vaswani-Shazeer/0737da0767d77606169cbf4187b83e1ab62f6077>`_ .
The attention mechanism is a weighted sum of a projection V of the inputs, with respect
to the scaled, normalised dot product of Q and K, which are also both linear projections
of the input. This procedure is repeated for each attention head, using different parameters.
Parameters
----------
num_heads : ``int``, required.
The number of attention heads to use.
input_dim : ``int``, required.
The size of the last dimension of the input tensor.
attention_dim ``int``, required.
The total dimension of the query and key projections which comprise the
dot product attention function. Must be divisible by ``num_heads``.
values_dim : ``int``, required.
The total dimension which the input is projected to for representing the values,
which are combined using the attention. Must be divisible by ``num_heads``.
output_projection_dim : ``int``, optional (default = None)
The dimensionality of the final output projection. If this is not passed
explicitly, the projection has size `input_size`.
attention_dropout_prob : ``float``, optional (default = 0.1).
The dropout probability applied to the normalised attention
distributions.
"""
def __init__(self, input_dim: 'int', attention_dim: 'int', values_dim:
'int', num_heads: 'int'=1, output_projection_dim: 'int'=None,
attention_dropout_prob: 'float'=0.1) ->None:
super(MultiHeadSelfAttentionNew, self).__init__()
self._num_heads = num_heads
self._input_dim = input_dim
self._output_dim = output_projection_dim or input_dim
self._attention_dim = attention_dim
self._values_dim = values_dim
if attention_dim % num_heads != 0:
raise ValueError(
f'Key size ({attention_dim}) must be divisible by the number of attention heads ({num_heads}).'
)
if values_dim % num_heads != 0:
raise ValueError(
f'Value size ({values_dim}) must be divisible by the number of attention heads ({num_heads}).'
)
self._combined_projection = Linear(input_dim, 2 * attention_dim +
values_dim)
self._scale = (input_dim // num_heads) ** 0.5
self._output_projection = Linear(values_dim, self._output_dim)
self._attention_dropout = Dropout(attention_dropout_prob)
def get_input_dim(self):
return self._input_dim
def get_output_dim(self):
return self._output_dim
def is_bidirectional(self):
return False
def forward(self, input_0):
primals_2 = self._combined_projection.weight
primals_3 = self._combined_projection.bias
primals_4 = self._output_projection.weight
primals_5 = self._output_projection.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
TimSYQQX/glyce
|
MultiHeadSelfAttention
| false | 14,514 |
[
"Apache-2.0"
] | 396 |
1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
https://github.com/TimSYQQX/glyce/tree/1542ed30ce104c25aa5c69ffcc9cc5ef2fcda975
|
TestNet
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/h3/ch3jaerz3yy2cwhlkdxgu5zovr4b3pud2kkovjsigzdkknwn2xvn.py
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv1d => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%view, %primals_2, %primals_3, [1], [0], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 1, 1), (1, 1, 1))
assert_size_stride(primals_3, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_1, (64, 1, 4), (4, 4, 1), 0), primals_2, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf0, (64, 1, 4), (4, 4, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(buf1, primals_3, 256, grid=grid(256), stream=stream0)
del primals_3
return (reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0), primals_2, reinterpret_tensor(primals_1, (64, 1, 4), (4, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, 1, 1), (1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class TestNet(nn.Module):
def __init__(self):
super().__init__()
self.conv = nn.Conv1d(1, 1, 1)
def forward(self, x):
x_len = x.shape[-1]
return self.conv(x.view(-1, 1, x_len)).view(x.shape)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + x0, tmp3, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 1, 1), (1, 1, 1))
assert_size_stride(primals_3, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(primals_1, (64,
1, 4), (4, 4, 1), 0), primals_2, stride=(1,), padding=(0,),
dilation=(1,), transposed=False, output_padding=(0,), groups=1,
bias=None)
assert_size_stride(buf0, (64, 1, 4), (4, 4, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(256)](buf1, primals_3, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_3
return reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0
), primals_2, reinterpret_tensor(primals_1, (64, 1, 4), (4, 4, 1), 0)
class TestNetNew(nn.Module):
def __init__(self):
super().__init__()
self.conv = nn.Conv1d(1, 1, 1)
def forward(self, input_0):
primals_2 = self.conv.weight
primals_3 = self.conv.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
TuZehai/pytorch_stoi
|
TestNet
| false | 14,515 |
[
"MIT"
] | 45 |
ae58e3ef4d608fc367e522150f48c58f122716fd
|
https://github.com/TuZehai/pytorch_stoi/tree/ae58e3ef4d608fc367e522150f48c58f122716fd
|
SimulatorReward
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/5j/c5ji4mfxenghd3ccczky5osir42aijmeisydrv7ufxv2edv4ktf6.py
# Topologically Sorted Source Nodes: [conv2d, x_1], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d => convolution
# x_1 => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%view, %primals_2, %primals_3, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_0 = async_compile.triton('triton_poi_fused_convolution_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 8
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/l5/cl567j2pbsgcv6nt7ux6tlu6b7q2zlzodpyhx5vc6sfgkf4lnrcu.py
# Topologically Sorted Source Nodes: [conv2d_1, x_2], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_1 => convolution_1
# x_2 => relu_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_1,), kwargs = {})
triton_poi_fused_convolution_relu_1 = async_compile.triton('triton_poi_fused_convolution_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 16
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/om/comxhxljgusfvchwfatyy3zlyu63ectx7cc43vhpq4slnbeelc4a.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_3 => convolution_2
# Graph fragment:
# %convolution_2 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_1, %primals_6, %primals_7, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_2 = async_compile.triton('triton_poi_fused_convolution_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 2048
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 16) % 32
tmp0 = tl.load(in_out_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, None)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/hx/chxm27gfgqrmlk4dtntfyb4eusceejl3zp6b5gugzys3ftbpi4v5.py
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# x_5 => relu_2
# Graph fragment:
# %add_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default_1, %primals_9), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor_1,), kwargs = {})
triton_poi_fused_relu_3 = async_compile.triton('triton_poi_fused_relu_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 800
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 200
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/5d/c5dvgve24hbvrdcz5w4qrfj4r34h3se3x2w5fculeyheewlhp45h.py
# Topologically Sorted Source Nodes: [x_6], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# x_6 => relu_3
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_11), kwargs = {})
# %relu_3 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor,), kwargs = {})
triton_poi_fused_relu_4 = async_compile.triton('triton_poi_fused_relu_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_4', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_4(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 400
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 100
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/3y/c3yxwuwmqvwyaal5sg5yv4hb66vvtdtpd7juzykvfeirynnm6y36.py
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# softmax => amax, div, exp, sub, sum_1
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%addmm_2, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%addmm_2, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_5 = async_compile.triton('triton_poi_fused__softmax_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_5(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 12
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 3)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (3*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (3*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (3*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp6 = tmp0 - tmp5
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp1 - tmp5
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp2 - tmp5
tmp11 = tl_math.exp(tmp10)
tmp12 = tmp9 + tmp11
tmp13 = tmp4 - tmp5
tmp14 = tl_math.exp(tmp13)
tmp15 = tmp12 + tmp14
tmp16 = tmp7 / tmp15
tl.store(out_ptr0 + (x2), tmp16, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (8, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (8, ), (1, ))
assert_size_stride(primals_4, (16, 8, 3, 3), (72, 9, 3, 1))
assert_size_stride(primals_5, (16, ), (1, ))
assert_size_stride(primals_6, (32, 16, 3, 3), (144, 9, 3, 1))
assert_size_stride(primals_7, (32, ), (1, ))
assert_size_stride(primals_8, (200, 512), (512, 1))
assert_size_stride(primals_9, (200, ), (1, ))
assert_size_stride(primals_10, (100, 200), (200, 1))
assert_size_stride(primals_11, (100, ), (1, ))
assert_size_stride(primals_12, (3, 100), (100, 1))
assert_size_stride(primals_13, (3, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4, 4), (128, 16, 4, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv2d, x_1], Original ATen: [aten.convolution, aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_0.run(buf1, primals_3, 512, grid=grid(512), stream=stream0)
del primals_3
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 16, 4, 4), (256, 16, 4, 1))
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [conv2d_1, x_2], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf3, primals_5, 1024, grid=grid(1024), stream=stream0)
del primals_5
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(buf3, primals_6, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 32, 4, 4), (512, 16, 4, 1))
buf5 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
triton_poi_fused_convolution_2.run(buf5, primals_7, 2048, grid=grid(2048), stream=stream0)
del primals_7
buf6 = empty_strided_cuda((4, 200), (200, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf5, (4, 512), (512, 1), 0), reinterpret_tensor(primals_8, (512, 200), (1, 512), 0), out=buf6)
buf7 = buf6; del buf6 # reuse
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.relu]
triton_poi_fused_relu_3.run(buf7, primals_9, 800, grid=grid(800), stream=stream0)
del primals_9
buf8 = empty_strided_cuda((4, 100), (100, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf7, reinterpret_tensor(primals_10, (200, 100), (1, 200), 0), out=buf8)
buf9 = buf8; del buf8 # reuse
# Topologically Sorted Source Nodes: [x_6], Original ATen: [aten.relu]
triton_poi_fused_relu_4.run(buf9, primals_11, 400, grid=grid(400), stream=stream0)
del primals_11
buf10 = empty_strided_cuda((4, 3), (3, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_7], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_13, buf9, reinterpret_tensor(primals_12, (100, 3), (1, 100), 0), alpha=1, beta=1, out=buf10)
del primals_13
buf11 = empty_strided_cuda((4, 3), (3, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf10, buf11, 12, grid=grid(12), stream=stream0)
del buf10
return (buf11, primals_2, primals_4, primals_6, primals_1, buf1, buf3, reinterpret_tensor(buf5, (4, 512), (512, 1), 0), buf7, buf9, buf11, primals_12, primals_10, primals_8, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((8, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((16, 8, 3, 3), (72, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((16, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((32, 16, 3, 3), (144, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((200, 512), (512, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((200, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((100, 200), (200, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((100, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((3, 100), (100, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((3, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn.functional as F
class SimulatorReward(torch.nn.Module):
def __init__(self):
super(SimulatorReward, self).__init__()
self.conv1 = torch.nn.Conv2d(4, 8, kernel_size=3, padding=1)
self.conv2 = torch.nn.Conv2d(8, 16, kernel_size=3, padding=1)
self.conv3 = torch.nn.Conv2d(16, 32, kernel_size=3, padding=1)
self.fc1 = torch.nn.Linear(512, 200)
self.fc2 = torch.nn.Linear(200, 100)
self.fc3 = torch.nn.Linear(100, 3)
def forward(self, x):
x = x.reshape(-1, 4, 4, 4)
x = F.relu(self.conv1(x))
x = F.relu(self.conv2(x))
x = self.conv3(x)
x = x.view(-1, 512)
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return F.softmax(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 8
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 16
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 16 % 32
tmp0 = tl.load(in_out_ptr0 + x3, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, None)
@triton.jit
def triton_poi_fused_relu_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 800
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 200
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_relu_4(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 400
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 100
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused__softmax_5(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 12
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 3
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 3 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 3 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 3 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp6 = tmp0 - tmp5
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp1 - tmp5
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp2 - tmp5
tmp11 = tl_math.exp(tmp10)
tmp12 = tmp9 + tmp11
tmp13 = tmp4 - tmp5
tmp14 = tl_math.exp(tmp13)
tmp15 = tmp12 + tmp14
tmp16 = tmp7 / tmp15
tl.store(out_ptr0 + x2, tmp16, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (8, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (8,), (1,))
assert_size_stride(primals_4, (16, 8, 3, 3), (72, 9, 3, 1))
assert_size_stride(primals_5, (16,), (1,))
assert_size_stride(primals_6, (32, 16, 3, 3), (144, 9, 3, 1))
assert_size_stride(primals_7, (32,), (1,))
assert_size_stride(primals_8, (200, 512), (512, 1))
assert_size_stride(primals_9, (200,), (1,))
assert_size_stride(primals_10, (100, 200), (200, 1))
assert_size_stride(primals_11, (100,), (1,))
assert_size_stride(primals_12, (3, 100), (100, 1))
assert_size_stride(primals_13, (3,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_1, primals_2, stride=(1,
1), padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4, 4), (128, 16, 4, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_relu_0[grid(512)](buf1, primals_3, 512,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_3
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 16, 4, 4), (256, 16, 4, 1))
buf3 = buf2
del buf2
triton_poi_fused_convolution_relu_1[grid(1024)](buf3, primals_5,
1024, XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
buf4 = extern_kernels.convolution(buf3, primals_6, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 32, 4, 4), (512, 16, 4, 1))
buf5 = buf4
del buf4
triton_poi_fused_convolution_2[grid(2048)](buf5, primals_7, 2048,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_7
buf6 = empty_strided_cuda((4, 200), (200, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf5, (4, 512), (512, 1), 0),
reinterpret_tensor(primals_8, (512, 200), (1, 512), 0), out=buf6)
buf7 = buf6
del buf6
triton_poi_fused_relu_3[grid(800)](buf7, primals_9, 800, XBLOCK=256,
num_warps=4, num_stages=1)
del primals_9
buf8 = empty_strided_cuda((4, 100), (100, 1), torch.float32)
extern_kernels.mm(buf7, reinterpret_tensor(primals_10, (200, 100),
(1, 200), 0), out=buf8)
buf9 = buf8
del buf8
triton_poi_fused_relu_4[grid(400)](buf9, primals_11, 400, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_11
buf10 = empty_strided_cuda((4, 3), (3, 1), torch.float32)
extern_kernels.addmm(primals_13, buf9, reinterpret_tensor(
primals_12, (100, 3), (1, 100), 0), alpha=1, beta=1, out=buf10)
del primals_13
buf11 = empty_strided_cuda((4, 3), (3, 1), torch.float32)
triton_poi_fused__softmax_5[grid(12)](buf10, buf11, 12, XBLOCK=16,
num_warps=1, num_stages=1)
del buf10
return (buf11, primals_2, primals_4, primals_6, primals_1, buf1, buf3,
reinterpret_tensor(buf5, (4, 512), (512, 1), 0), buf7, buf9, buf11,
primals_12, primals_10, primals_8)
class SimulatorRewardNew(torch.nn.Module):
def __init__(self):
super(SimulatorRewardNew, self).__init__()
self.conv1 = torch.nn.Conv2d(4, 8, kernel_size=3, padding=1)
self.conv2 = torch.nn.Conv2d(8, 16, kernel_size=3, padding=1)
self.conv3 = torch.nn.Conv2d(16, 32, kernel_size=3, padding=1)
self.fc1 = torch.nn.Linear(512, 200)
self.fc2 = torch.nn.Linear(200, 100)
self.fc3 = torch.nn.Linear(100, 3)
def forward(self, input_0):
primals_2 = self.conv1.weight
primals_3 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_6 = self.conv3.weight
primals_7 = self.conv3.bias
primals_8 = self.fc1.weight
primals_9 = self.fc1.bias
primals_10 = self.fc2.weight
primals_11 = self.fc2.bias
primals_12 = self.fc3.weight
primals_13 = self.fc3.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13])
return output[0]
|
Tuantrung/DeepReinforcementLearningInAction
|
SimulatorReward
| false | 14,516 |
[
"MIT"
] | 474 |
8afda00a8211326c540b5de5a964d62a7f29a70c
|
https://github.com/Tuantrung/DeepReinforcementLearningInAction/tree/8afda00a8211326c540b5de5a964d62a7f29a70c
|
Conv_Q
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/sn/csnsms5tdtjok5uxcwcbko2ioqfann3pwnmkfhlujgvnsujd5bud.py
# Topologically Sorted Source Nodes: [conv2d, c], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# c => relu
# conv2d => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [4, 4], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_0 = async_compile.triton('triton_poi_fused_convolution_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[262144],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 156800
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 1225) % 32
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/f4/cf4q74veoggsxdgdkl43ap6cyqfylpfk3qs7wdqoebyfzzb36dvw.py
# Topologically Sorted Source Nodes: [conv2d_1, c_1], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# c_1 => relu_1
# conv2d_1 => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [2, 2], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_1,), kwargs = {})
triton_poi_fused_convolution_relu_1 = async_compile.triton('triton_poi_fused_convolution_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[65536],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 65536
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 256) % 64
tmp0 = tl.load(in_out_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/jd/cjdph23oasfased5f2dfu7kch7qcwjhegz6fxsrsn22yzjy3qj2u.py
# Topologically Sorted Source Nodes: [conv2d_2, c_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# c_2 => relu_2
# conv2d_2 => convolution_2
# Graph fragment:
# %convolution_2 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_1, %primals_6, %primals_7, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_2,), kwargs = {})
# %le_2 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_2, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_2 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[65536],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_2(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 50176
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 196) % 64
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
tl.store(out_ptr0 + (x3), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/b5/cb5bmriikeb3z65rmk4n4vz3fvd4pzjrhfemonu665rzgwpxeamm.py
# Topologically Sorted Source Nodes: [q], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# q => relu_3
# Graph fragment:
# %add_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default_1, %primals_9), kwargs = {})
# %relu_3 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor_1,), kwargs = {})
triton_poi_fused_relu_3 = async_compile.triton('triton_poi_fused_relu_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8192],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 8192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 512
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/u2/cu22r6l6j2dwad6ehxo727pi566vx4re7y7wswbfiuiqovkorg37.py
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# log_softmax => amax, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%addmm_2, [1], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%addmm_2, %amax), kwargs = {})
triton_poi_fused__log_softmax_4 = async_compile.triton('triton_poi_fused__log_softmax_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/r4/cr4lhbz4zupusb2cwexdjmmp7mdneumkf5mp763v3ocsgjeaa5ai.py
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# log_softmax => exp, log, sub_1, sum_1
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
triton_poi_fused__log_softmax_5 = async_compile.triton('triton_poi_fused__log_softmax_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_5(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tl.store(out_ptr0 + (x2), tmp13, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15 = args
args.clear()
assert_size_stride(primals_1, (32, 4, 8, 8), (256, 64, 8, 1))
assert_size_stride(primals_2, (32, ), (1, ))
assert_size_stride(primals_3, (4, 4, 144, 144), (82944, 20736, 144, 1))
assert_size_stride(primals_4, (64, 32, 4, 4), (512, 16, 4, 1))
assert_size_stride(primals_5, (64, ), (1, ))
assert_size_stride(primals_6, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_7, (64, ), (1, ))
assert_size_stride(primals_8, (512, 3136), (3136, 1))
assert_size_stride(primals_9, (512, ), (1, ))
assert_size_stride(primals_10, (512, 3136), (3136, 1))
assert_size_stride(primals_11, (512, ), (1, ))
assert_size_stride(primals_12, (4, 512), (512, 1))
assert_size_stride(primals_13, (4, ), (1, ))
assert_size_stride(primals_14, (4, 512), (512, 1))
assert_size_stride(primals_15, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(4, 4), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 32, 35, 35), (39200, 1225, 35, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv2d, c], Original ATen: [aten.convolution, aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_0.run(buf1, primals_2, 156800, grid=grid(156800), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(2, 2), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 64, 16, 16), (16384, 256, 16, 1))
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [conv2d_1, c_1], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf3, primals_5, 65536, grid=grid(65536), stream=stream0)
del primals_5
# Topologically Sorted Source Nodes: [conv2d_2], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(buf3, primals_6, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 64, 14, 14), (12544, 196, 14, 1))
buf5 = buf4; del buf4 # reuse
buf14 = empty_strided_cuda((4, 64, 14, 14), (12544, 196, 14, 1), torch.bool)
# Topologically Sorted Source Nodes: [conv2d_2, c_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_2.run(buf5, primals_7, buf14, 50176, grid=grid(50176), stream=stream0)
del primals_7
buf6 = empty_strided_cuda((16, 512), (512, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf5, (16, 3136), (3136, 1), 0), reinterpret_tensor(primals_8, (3136, 512), (1, 3136), 0), out=buf6)
buf7 = buf6; del buf6 # reuse
# Topologically Sorted Source Nodes: [q], Original ATen: [aten.relu]
triton_poi_fused_relu_3.run(buf7, primals_9, 8192, grid=grid(8192), stream=stream0)
del primals_9
buf8 = empty_strided_cuda((16, 512), (512, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf5, (16, 3136), (3136, 1), 0), reinterpret_tensor(primals_10, (3136, 512), (1, 3136), 0), out=buf8)
buf9 = buf8; del buf8 # reuse
# Topologically Sorted Source Nodes: [i], Original ATen: [aten.relu]
triton_poi_fused_relu_3.run(buf9, primals_11, 8192, grid=grid(8192), stream=stream0)
del primals_11
buf10 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [i_1], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_13, buf9, reinterpret_tensor(primals_12, (512, 4), (1, 512), 0), alpha=1, beta=1, out=buf10)
del primals_13
buf11 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_15, buf7, reinterpret_tensor(primals_14, (512, 4), (1, 512), 0), alpha=1, beta=1, out=buf11)
del primals_15
buf12 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
triton_poi_fused__log_softmax_4.run(buf10, buf12, 64, grid=grid(64), stream=stream0)
buf13 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
triton_poi_fused__log_softmax_5.run(buf12, buf13, 64, grid=grid(64), stream=stream0)
del buf12
return (buf11, buf13, buf10, primals_1, primals_3, primals_4, primals_6, buf1, buf3, reinterpret_tensor(buf5, (16, 3136), (3136, 1), 0), buf7, buf9, buf13, primals_14, primals_12, primals_10, primals_8, buf14, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((32, 4, 8, 8), (256, 64, 8, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 144, 144), (82944, 20736, 144, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((64, 32, 4, 4), (512, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((64, 64, 3, 3), (576, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((512, 3136), (3136, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((512, 3136), (3136, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, 512), (512, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_14 = rand_strided((4, 512), (512, 1), device='cuda:0', dtype=torch.float32)
primals_15 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.data
class Conv_Q(nn.Module):
def __init__(self, frames, num_actions):
super(Conv_Q, self).__init__()
self.c1 = nn.Conv2d(frames, 32, kernel_size=8, stride=4)
self.c2 = nn.Conv2d(32, 64, kernel_size=4, stride=2)
self.c3 = nn.Conv2d(64, 64, kernel_size=3, stride=1)
self.q1 = nn.Linear(3136, 512)
self.q2 = nn.Linear(512, num_actions)
self.i1 = nn.Linear(3136, 512)
self.i2 = nn.Linear(512, num_actions)
def forward(self, state):
c = F.relu(self.c1(state))
c = F.relu(self.c2(c))
c = F.relu(self.c3(c))
q = F.relu(self.q1(c.reshape(-1, 3136)))
i = F.relu(self.i1(c.reshape(-1, 3136)))
i = self.i2(i)
return self.q2(q), F.log_softmax(i, dim=1), i
def get_inputs():
return [torch.rand([4, 4, 144, 144])]
def get_init_inputs():
return [[], {'frames': 4, 'num_actions': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 156800
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 1225 % 32
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 256 % 64
tmp0 = tl.load(in_out_ptr0 + x3, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, None)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_2(in_out_ptr0,
in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 50176
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 196 % 64
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x3, tmp4, xmask)
tl.store(out_ptr0 + x3, tmp6, xmask)
@triton.jit
def triton_poi_fused_relu_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 512
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused__log_softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused__log_softmax_5(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tl.store(out_ptr0 + x2, tmp13, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13, primals_14, primals_15) = args
args.clear()
assert_size_stride(primals_1, (32, 4, 8, 8), (256, 64, 8, 1))
assert_size_stride(primals_2, (32,), (1,))
assert_size_stride(primals_3, (4, 4, 144, 144), (82944, 20736, 144, 1))
assert_size_stride(primals_4, (64, 32, 4, 4), (512, 16, 4, 1))
assert_size_stride(primals_5, (64,), (1,))
assert_size_stride(primals_6, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_7, (64,), (1,))
assert_size_stride(primals_8, (512, 3136), (3136, 1))
assert_size_stride(primals_9, (512,), (1,))
assert_size_stride(primals_10, (512, 3136), (3136, 1))
assert_size_stride(primals_11, (512,), (1,))
assert_size_stride(primals_12, (4, 512), (512, 1))
assert_size_stride(primals_13, (4,), (1,))
assert_size_stride(primals_14, (4, 512), (512, 1))
assert_size_stride(primals_15, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(4,
4), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 32, 35, 35), (39200, 1225, 35, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_relu_0[grid(156800)](buf1, primals_2,
156800, XBLOCK=512, num_warps=8, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(2, 2),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 64, 16, 16), (16384, 256, 16, 1))
buf3 = buf2
del buf2
triton_poi_fused_convolution_relu_1[grid(65536)](buf3, primals_5,
65536, XBLOCK=256, num_warps=4, num_stages=1)
del primals_5
buf4 = extern_kernels.convolution(buf3, primals_6, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 64, 14, 14), (12544, 196, 14, 1))
buf5 = buf4
del buf4
buf14 = empty_strided_cuda((4, 64, 14, 14), (12544, 196, 14, 1),
torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_2[grid(50176)](
buf5, primals_7, buf14, 50176, XBLOCK=512, num_warps=4,
num_stages=1)
del primals_7
buf6 = empty_strided_cuda((16, 512), (512, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf5, (16, 3136), (3136, 1), 0
), reinterpret_tensor(primals_8, (3136, 512), (1, 3136), 0),
out=buf6)
buf7 = buf6
del buf6
triton_poi_fused_relu_3[grid(8192)](buf7, primals_9, 8192, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_9
buf8 = empty_strided_cuda((16, 512), (512, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf5, (16, 3136), (3136, 1), 0
), reinterpret_tensor(primals_10, (3136, 512), (1, 3136), 0),
out=buf8)
buf9 = buf8
del buf8
triton_poi_fused_relu_3[grid(8192)](buf9, primals_11, 8192, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_11
buf10 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_13, buf9, reinterpret_tensor(
primals_12, (512, 4), (1, 512), 0), alpha=1, beta=1, out=buf10)
del primals_13
buf11 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_15, buf7, reinterpret_tensor(
primals_14, (512, 4), (1, 512), 0), alpha=1, beta=1, out=buf11)
del primals_15
buf12 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
triton_poi_fused__log_softmax_4[grid(64)](buf10, buf12, 64, XBLOCK=
64, num_warps=1, num_stages=1)
buf13 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
triton_poi_fused__log_softmax_5[grid(64)](buf12, buf13, 64, XBLOCK=
64, num_warps=1, num_stages=1)
del buf12
return (buf11, buf13, buf10, primals_1, primals_3, primals_4, primals_6,
buf1, buf3, reinterpret_tensor(buf5, (16, 3136), (3136, 1), 0),
buf7, buf9, buf13, primals_14, primals_12, primals_10, primals_8, buf14
)
class Conv_QNew(nn.Module):
def __init__(self, frames, num_actions):
super(Conv_QNew, self).__init__()
self.c1 = nn.Conv2d(frames, 32, kernel_size=8, stride=4)
self.c2 = nn.Conv2d(32, 64, kernel_size=4, stride=2)
self.c3 = nn.Conv2d(64, 64, kernel_size=3, stride=1)
self.q1 = nn.Linear(3136, 512)
self.q2 = nn.Linear(512, num_actions)
self.i1 = nn.Linear(3136, 512)
self.i2 = nn.Linear(512, num_actions)
def forward(self, input_0):
primals_1 = self.c1.weight
primals_2 = self.c1.bias
primals_4 = self.c2.weight
primals_5 = self.c2.bias
primals_6 = self.c3.weight
primals_7 = self.c3.bias
primals_8 = self.q1.weight
primals_9 = self.q1.bias
primals_12 = self.q2.weight
primals_13 = self.q2.bias
primals_10 = self.i1.weight
primals_11 = self.i1.bias
primals_14 = self.i2.weight
primals_15 = self.i2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13, primals_14,
primals_15])
return output[0], output[1], output[2]
|
Thibaud-Ardoin/d4rl_evaluations
|
Conv_Q
| false | 14,517 |
[
"Apache-2.0"
] | 123 |
135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
https://github.com/Thibaud-Ardoin/d4rl_evaluations/tree/135b23d3aecc234aacaeaaa019fbc7101d9b87ec
|
LayerNorm
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/sk/csk6e2ijxwozqfg5yafv56tovxgvs23ahvpatwnswiausxio766y.py
# Topologically Sorted Source Nodes: [pow_1, variance, add, rsqrt, hidden_states, mul_1], Original ATen: [aten.pow, aten.mean, aten.add, aten.rsqrt, aten.mul]
# Source node to ATen node mapping:
# add => add
# hidden_states => mul
# mul_1 => mul_1
# pow_1 => pow_1
# rsqrt => rsqrt
# variance => mean
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%primals_1, 2), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%pow_1, [-1], True), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean, 1e-06), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_1, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_2, %mul), kwargs = {})
triton_poi_fused_add_mean_mul_pow_rsqrt_0 = async_compile.triton('triton_poi_fused_add_mean_mul_pow_rsqrt_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mean_mul_pow_rsqrt_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mean_mul_pow_rsqrt_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp2 = tl.load(in_ptr1 + (4*x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr1 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp2 * tmp2
tmp5 = tmp4 * tmp4
tmp6 = tmp3 + tmp5
tmp8 = tmp7 * tmp7
tmp9 = tmp6 + tmp8
tmp11 = tmp10 * tmp10
tmp12 = tmp9 + tmp11
tmp13 = 4.0
tmp14 = tmp12 / tmp13
tmp15 = 1e-06
tmp16 = tmp14 + tmp15
tmp17 = libdevice.rsqrt(tmp16)
tmp18 = tmp1 * tmp17
tmp19 = tmp0 * tmp18
tl.store(out_ptr0 + (x2), tmp19, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, variance, add, rsqrt, hidden_states, mul_1], Original ATen: [aten.pow, aten.mean, aten.add, aten.rsqrt, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_add_mean_mul_pow_rsqrt_0.run(primals_2, primals_1, buf0, 256, grid=grid(256), stream=stream0)
del primals_2
return (buf0, primals_1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class LayerNorm(nn.Module):
def __init__(self, hidden_size, eps=1e-06):
"""
Construct a layernorm module in the T5 style No bias and no subtraction of mean.
"""
super().__init__()
self.weight = nn.Parameter(torch.ones(hidden_size))
self.eps = eps
def forward(self, hidden_states):
variance = hidden_states.pow(2).mean(-1, keepdim=True)
hidden_states = hidden_states * torch.rsqrt(variance + self.eps)
if self.weight.dtype == torch.float16:
hidden_states = hidden_states
return self.weight * hidden_states
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'hidden_size': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_mean_mul_pow_rsqrt_0(in_ptr0, in_ptr1, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp2 = tl.load(in_ptr1 + 4 * x1, xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr1 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp3 = tmp2 * tmp2
tmp5 = tmp4 * tmp4
tmp6 = tmp3 + tmp5
tmp8 = tmp7 * tmp7
tmp9 = tmp6 + tmp8
tmp11 = tmp10 * tmp10
tmp12 = tmp9 + tmp11
tmp13 = 4.0
tmp14 = tmp12 / tmp13
tmp15 = 1e-06
tmp16 = tmp14 + tmp15
tmp17 = libdevice.rsqrt(tmp16)
tmp18 = tmp1 * tmp17
tmp19 = tmp0 * tmp18
tl.store(out_ptr0 + x2, tmp19, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_mean_mul_pow_rsqrt_0[grid(256)](primals_2,
primals_1, buf0, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
return buf0, primals_1
class LayerNormNew(nn.Module):
def __init__(self, hidden_size, eps=1e-06):
"""
Construct a layernorm module in the T5 style No bias and no subtraction of mean.
"""
super().__init__()
self.weight = nn.Parameter(torch.ones(hidden_size))
self.eps = eps
def forward(self, input_0):
primals_2 = self.weight
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0]
|
TsinghuaAI/CPM-2-Pretrain
|
LayerNorm
| false | 14,518 |
[
"MIT"
] | 54 |
33003865239e7ba13a12aabf9ec2735cef66bf3b
|
https://github.com/TsinghuaAI/CPM-2-Pretrain/tree/33003865239e7ba13a12aabf9ec2735cef66bf3b
|
PKT
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/us/cuss63los66cmuzjpei2dpgpl3bbx6cgwht7porec7hj3lclnazh.py
# Topologically Sorted Source Nodes: [pow_2, sum_2, target_net_norm, add_1, target_net, setitem_1], Original ATen: [aten.pow, aten.sum, aten.sqrt, aten.add, aten.div, aten.lift_fresh, aten.index_put]
# Source node to ATen node mapping:
# add_1 => add_1
# pow_2 => pow_2
# setitem_1 => full_default_1, index_put_1
# sum_2 => sum_2
# target_net => div_1
# target_net_norm => sqrt_1
# Graph fragment:
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%arg1_1, 2), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_2, [1], True), kwargs = {})
# %sqrt_1 : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%sum_2,), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sqrt_1, 1e-07), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg1_1, %add_1), kwargs = {})
# %full_default_1 : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cpu, pin_memory: False})
# %index_put_1 : [num_users=2] = call_function[target=torch.ops.aten.index_put_.default](args = (%div_1, [%ne_1], %full_default_1), kwargs = {})
triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0 = async_compile.triton('triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tmp16 = tmp15 != tmp15
tmp17 = 0.0
tmp18 = tl.where(tmp16, tmp17, tmp15)
tl.store(in_out_ptr0 + (x2), tmp18, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/yp/cypfp6dwhkdjjlgaluiwgyjiyan74ahdhbisu33n2ow6e3gmrfz3.py
# Topologically Sorted Source Nodes: [add_3, target_similarity_1, sum_4, target_similarity_2, add_4, add_2, model_similarity_1, sum_3, model_similarity_2, add_5, truediv_6, log, mul, loss], Original ATen: [aten.add, aten.div, aten.sum, aten.log, aten.mul, aten.mean]
# Source node to ATen node mapping:
# add_2 => add_2
# add_3 => add_3
# add_4 => add_4
# add_5 => add_5
# log => log
# loss => mean
# model_similarity_1 => div_2
# model_similarity_2 => div_4
# mul => mul
# sum_3 => sum_3
# sum_4 => sum_4
# target_similarity_1 => div_3
# target_similarity_2 => div_5
# truediv_6 => div_6
# Graph fragment:
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_1, 1.0), kwargs = {})
# %div_3 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%add_3, 2.0), kwargs = {})
# %sum_4 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%div_3, [1], True), kwargs = {})
# %div_5 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%div_3, %sum_4), kwargs = {})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%div_5, 1e-07), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm, 1.0), kwargs = {})
# %div_2 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%add_2, 2.0), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%div_2, [1], True), kwargs = {})
# %div_4 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%div_2, %sum_3), kwargs = {})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%div_4, 1e-07), kwargs = {})
# %div_6 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%add_4, %add_5), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%div_6,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div_5, %log), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%mul,), kwargs = {})
triton_per_fused_add_div_log_mean_mul_sum_1 = async_compile.triton('triton_per_fused_add_div_log_mean_mul_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 16],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_log_mean_mul_sum_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 10, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_log_mean_mul_sum_1(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r2 = rindex
r1 = (rindex // 4)
tmp0 = tl.load(in_ptr0 + (r2), None)
tmp5 = tl.load(in_ptr0 + (4*r1), None, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr0 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr1 + (r2), None)
tmp26 = tl.load(in_ptr1 + (4*r1), None, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr1 + (1 + (4*r1)), None, eviction_policy='evict_last')
tmp33 = tl.load(in_ptr1 + (2 + (4*r1)), None, eviction_policy='evict_last')
tmp37 = tl.load(in_ptr1 + (3 + (4*r1)), None, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp6 = tmp5 + tmp1
tmp7 = tmp6 * tmp3
tmp9 = tmp8 + tmp1
tmp10 = tmp9 * tmp3
tmp11 = tmp7 + tmp10
tmp13 = tmp12 + tmp1
tmp14 = tmp13 * tmp3
tmp15 = tmp11 + tmp14
tmp17 = tmp16 + tmp1
tmp18 = tmp17 * tmp3
tmp19 = tmp15 + tmp18
tmp20 = tmp4 / tmp19
tmp21 = 1e-07
tmp22 = tmp20 + tmp21
tmp24 = tmp23 + tmp1
tmp25 = tmp24 * tmp3
tmp27 = tmp26 + tmp1
tmp28 = tmp27 * tmp3
tmp30 = tmp29 + tmp1
tmp31 = tmp30 * tmp3
tmp32 = tmp28 + tmp31
tmp34 = tmp33 + tmp1
tmp35 = tmp34 * tmp3
tmp36 = tmp32 + tmp35
tmp38 = tmp37 + tmp1
tmp39 = tmp38 * tmp3
tmp40 = tmp36 + tmp39
tmp41 = tmp25 / tmp40
tmp42 = tmp41 + tmp21
tmp43 = tmp22 / tmp42
tmp44 = tl_math.log(tmp43)
tmp45 = tmp20 * tmp44
tmp46 = tl.broadcast_to(tmp45, [XBLOCK, RBLOCK])
tmp48 = tl.sum(tmp46, 1)[:, None]
tmp49 = 16.0
tmp50 = tmp48 / tmp49
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp50, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4), (4, 1))
assert_size_stride(arg1_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [pow_2, sum_2, target_net_norm, add_1, target_net, setitem_1], Original ATen: [aten.pow, aten.sum, aten.sqrt, aten.add, aten.div, aten.lift_fresh, aten.index_put]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0.run(buf1, arg1_1, 16, grid=grid(16), stream=stream0)
del arg1_1
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [target_similarity], Original ATen: [aten.mm]
extern_kernels.mm(buf1, reinterpret_tensor(buf1, (4, 4), (1, 4), 0), out=buf2)
buf4 = buf1; del buf1 # reuse
buf5 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [pow_1, sum_1, output_net_norm, add, output_net, setitem], Original ATen: [aten.pow, aten.sum, aten.sqrt, aten.add, aten.div, aten.lift_fresh, aten.index_put]
triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0.run(buf5, arg0_1, 16, grid=grid(16), stream=stream0)
del arg0_1
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [model_similarity], Original ATen: [aten.mm]
extern_kernels.mm(buf5, reinterpret_tensor(buf5, (4, 4), (1, 4), 0), out=buf6)
del buf5
buf7 = empty_strided_cuda((), (), torch.float32)
buf8 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [add_3, target_similarity_1, sum_4, target_similarity_2, add_4, add_2, model_similarity_1, sum_3, model_similarity_2, add_5, truediv_6, log, mul, loss], Original ATen: [aten.add, aten.div, aten.sum, aten.log, aten.mul, aten.mean]
triton_per_fused_add_div_log_mean_mul_sum_1.run(buf8, buf2, buf6, 1, 16, grid=grid(1), stream=stream0)
del buf2
del buf6
return (buf8, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class PKT(nn.Module):
"""Probabilistic Knowledge Transfer for deep representation learning
Code from author: https://github.com/passalis/probabilistic_kt"""
def __init__(self):
super(PKT, self).__init__()
def forward(self, f_s, f_t):
return self.cosine_similarity_loss(f_s, f_t)
@staticmethod
def cosine_similarity_loss(output_net, target_net, eps=1e-07):
output_net_norm = torch.sqrt(torch.sum(output_net ** 2, dim=1,
keepdim=True))
output_net = output_net / (output_net_norm + eps)
output_net[output_net != output_net] = 0
target_net_norm = torch.sqrt(torch.sum(target_net ** 2, dim=1,
keepdim=True))
target_net = target_net / (target_net_norm + eps)
target_net[target_net != target_net] = 0
model_similarity = torch.mm(output_net, output_net.transpose(0, 1))
target_similarity = torch.mm(target_net, target_net.transpose(0, 1))
model_similarity = (model_similarity + 1.0) / 2.0
target_similarity = (target_similarity + 1.0) / 2.0
model_similarity = model_similarity / torch.sum(model_similarity,
dim=1, keepdim=True)
target_similarity = target_similarity / torch.sum(target_similarity,
dim=1, keepdim=True)
loss = torch.mean(target_similarity * torch.log((target_similarity +
eps) / (model_similarity + eps)))
return loss
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0(in_out_ptr0,
in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-07
tmp14 = tmp12 + tmp13
tmp15 = tmp0 / tmp14
tmp16 = tmp15 != tmp15
tmp17 = 0.0
tmp18 = tl.where(tmp16, tmp17, tmp15)
tl.store(in_out_ptr0 + x2, tmp18, xmask)
@triton.jit
def triton_per_fused_add_div_log_mean_mul_sum_1(in_out_ptr0, in_ptr0,
in_ptr1, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r2 = rindex
r1 = rindex // 4
tmp0 = tl.load(in_ptr0 + r2, None)
tmp5 = tl.load(in_ptr0 + 4 * r1, None, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr0 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr1 + r2, None)
tmp26 = tl.load(in_ptr1 + 4 * r1, None, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr1 + (1 + 4 * r1), None, eviction_policy='evict_last')
tmp33 = tl.load(in_ptr1 + (2 + 4 * r1), None, eviction_policy='evict_last')
tmp37 = tl.load(in_ptr1 + (3 + 4 * r1), None, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp6 = tmp5 + tmp1
tmp7 = tmp6 * tmp3
tmp9 = tmp8 + tmp1
tmp10 = tmp9 * tmp3
tmp11 = tmp7 + tmp10
tmp13 = tmp12 + tmp1
tmp14 = tmp13 * tmp3
tmp15 = tmp11 + tmp14
tmp17 = tmp16 + tmp1
tmp18 = tmp17 * tmp3
tmp19 = tmp15 + tmp18
tmp20 = tmp4 / tmp19
tmp21 = 1e-07
tmp22 = tmp20 + tmp21
tmp24 = tmp23 + tmp1
tmp25 = tmp24 * tmp3
tmp27 = tmp26 + tmp1
tmp28 = tmp27 * tmp3
tmp30 = tmp29 + tmp1
tmp31 = tmp30 * tmp3
tmp32 = tmp28 + tmp31
tmp34 = tmp33 + tmp1
tmp35 = tmp34 * tmp3
tmp36 = tmp32 + tmp35
tmp38 = tmp37 + tmp1
tmp39 = tmp38 * tmp3
tmp40 = tmp36 + tmp39
tmp41 = tmp25 / tmp40
tmp42 = tmp41 + tmp21
tmp43 = tmp22 / tmp42
tmp44 = tl_math.log(tmp43)
tmp45 = tmp20 * tmp44
tmp46 = tl.broadcast_to(tmp45, [XBLOCK, RBLOCK])
tmp48 = tl.sum(tmp46, 1)[:, None]
tmp49 = 16.0
tmp50 = tmp48 / tmp49
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp50, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4), (4, 1))
assert_size_stride(arg1_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0[grid(16)](
buf1, arg1_1, 16, XBLOCK=16, num_warps=1, num_stages=1)
del arg1_1
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf1, reinterpret_tensor(buf1, (4, 4), (1, 4), 0),
out=buf2)
buf4 = buf1
del buf1
buf5 = buf4
del buf4
triton_poi_fused_add_div_index_put_lift_fresh_pow_sqrt_sum_0[grid(16)](
buf5, arg0_1, 16, XBLOCK=16, num_warps=1, num_stages=1)
del arg0_1
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(buf5, reinterpret_tensor(buf5, (4, 4), (1, 4), 0),
out=buf6)
del buf5
buf7 = empty_strided_cuda((), (), torch.float32)
buf8 = buf7
del buf7
triton_per_fused_add_div_log_mean_mul_sum_1[grid(1)](buf8, buf2,
buf6, 1, 16, XBLOCK=1, num_warps=2, num_stages=1)
del buf2
del buf6
return buf8,
class PKTNew(nn.Module):
"""Probabilistic Knowledge Transfer for deep representation learning
Code from author: https://github.com/passalis/probabilistic_kt"""
def __init__(self):
super(PKTNew, self).__init__()
@staticmethod
def cosine_similarity_loss(output_net, target_net, eps=1e-07):
output_net_norm = torch.sqrt(torch.sum(output_net ** 2, dim=1,
keepdim=True))
output_net = output_net / (output_net_norm + eps)
output_net[output_net != output_net] = 0
target_net_norm = torch.sqrt(torch.sum(target_net ** 2, dim=1,
keepdim=True))
target_net = target_net / (target_net_norm + eps)
target_net[target_net != target_net] = 0
model_similarity = torch.mm(output_net, output_net.transpose(0, 1))
target_similarity = torch.mm(target_net, target_net.transpose(0, 1))
model_similarity = (model_similarity + 1.0) / 2.0
target_similarity = (target_similarity + 1.0) / 2.0
model_similarity = model_similarity / torch.sum(model_similarity,
dim=1, keepdim=True)
target_similarity = target_similarity / torch.sum(target_similarity,
dim=1, keepdim=True)
loss = torch.mean(target_similarity * torch.log((target_similarity +
eps) / (model_similarity + eps)))
return loss
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
UBCDingXin/RepDistiller
|
PKT
| false | 14,519 |
[
"BSD-2-Clause"
] | 1,347 |
dcc043277f2820efafd679ffb82b8e8195b7e222
|
https://github.com/UBCDingXin/RepDistiller/tree/dcc043277f2820efafd679ffb82b8e8195b7e222
|
LogSoftmax
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/xe/cxeq77gbpevhf6jov7fs3c25pvswzi43xn2bxfthg2nvsuurswra.py
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# log_softmax => amax, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%arg0_1, [4], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %amax), kwargs = {})
triton_poi_fused__log_softmax_0 = async_compile.triton('triton_poi_fused__log_softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/vk/cvktls4q7cyrivezv7oyxxe47zkmbi43c4ac44xbpo44vj6ikleq.py
# Topologically Sorted Source Nodes: [log_softmax, nll, mul, sum_1, clamp, truediv], Original ATen: [aten._log_softmax, aten.neg, aten.mul, aten.sum, aten.clamp, aten.div]
# Source node to ATen node mapping:
# clamp => clamp_min
# log_softmax => exp, log, sub_1, sum_1
# mul => mul
# nll => neg
# sum_1 => sum_2
# truediv => div
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [4], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%sub_1,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%neg, %arg1_1), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%arg1_1, [1], True), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sum_2, 1), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul, %clamp_min), kwargs = {})
triton_poi_fused__log_softmax_clamp_div_mul_neg_sum_1 = async_compile.triton('triton_poi_fused__log_softmax_clamp_div_mul_neg_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_clamp_div_mul_neg_sum_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 10, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_clamp_div_mul_neg_sum_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x5 = xindex
x6 = (xindex // 4)
x7 = xindex % 256
x3 = (xindex // 64) % 4
x8 = xindex % 16
tmp0 = tl.load(in_ptr0 + (x5), xmask)
tmp1 = tl.load(in_ptr0 + (4*x6), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x6)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*x6)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*x6)), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr1 + (x7), xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr1 + (x8 + (64*x3)), xmask, eviction_policy='evict_last')
tmp18 = tl.load(in_ptr1 + (16 + x8 + (64*x3)), xmask, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr1 + (32 + x8 + (64*x3)), xmask, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr1 + (48 + x8 + (64*x3)), xmask, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tmp14 = -tmp13
tmp16 = tmp14 * tmp15
tmp19 = tmp17 + tmp18
tmp21 = tmp19 + tmp20
tmp23 = tmp21 + tmp22
tmp24 = 1.0
tmp25 = triton_helpers.maximum(tmp23, tmp24)
tmp26 = tmp16 / tmp25
tl.store(out_ptr0 + (x5), tmp26, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/kj/ckjk2npksmydlmbe5h7pvet6hpnnsr4gr3u47inkwbmigvketsr4.py
# Topologically Sorted Source Nodes: [sum_2, mean], Original ATen: [aten.sum, aten.mean]
# Source node to ATen node mapping:
# mean => mean
# sum_2 => sum_3
# Graph fragment:
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%div, [1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_3,), kwargs = {})
triton_per_fused_mean_sum_2 = async_compile.triton('triton_per_fused_mean_sum_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mean_sum_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 4, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mean_sum_2(in_out_ptr0, in_ptr0, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex % 64
r1 = (rindex // 64)
tmp0 = tl.load(in_ptr0 + (r0 + (256*r1)), None)
tmp1 = tl.load(in_ptr0 + (64 + r0 + (256*r1)), None)
tmp3 = tl.load(in_ptr0 + (128 + r0 + (256*r1)), None)
tmp5 = tl.load(in_ptr0 + (192 + r0 + (256*r1)), None)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = tl.broadcast_to(tmp6, [RBLOCK])
tmp9 = triton_helpers.promote_to_tensor(tl.sum(tmp7, 0))
tmp10 = 256.0
tmp11 = tmp9 / tmp10
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp11, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__log_softmax_0.run(arg0_1, buf0, 1024, grid=grid(1024), stream=stream0)
del arg0_1
buf1 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [log_softmax, nll, mul, sum_1, clamp, truediv], Original ATen: [aten._log_softmax, aten.neg, aten.mul, aten.sum, aten.clamp, aten.div]
triton_poi_fused__log_softmax_clamp_div_mul_neg_sum_1.run(buf0, arg1_1, buf1, 1024, grid=grid(1024), stream=stream0)
del arg1_1
del buf0
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [sum_2, mean], Original ATen: [aten.sum, aten.mean]
triton_per_fused_mean_sum_2.run(buf3, buf1, 1, 256, grid=grid(1), stream=stream0)
del buf1
return (buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn.functional as F
class LogSoftmax(torch.nn.Module):
def __init__(self, dim):
super(LogSoftmax, self).__init__()
self.dim = dim
def forward(self, x, a):
nll = -F.log_softmax(x, self.dim, _stacklevel=5)
return (nll * a / a.sum(1, keepdim=True).clamp(min=1)).sum(dim=1).mean(
)
def get_inputs():
return [torch.rand([4, 4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'dim': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused__log_softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused__log_softmax_clamp_div_mul_neg_sum_1(in_ptr0, in_ptr1,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x5 = xindex
x6 = xindex // 4
x7 = xindex % 256
x3 = xindex // 64 % 4
x8 = xindex % 16
tmp0 = tl.load(in_ptr0 + x5, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x6, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x6), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * x6), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * x6), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr1 + x7, xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr1 + (x8 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp18 = tl.load(in_ptr1 + (16 + x8 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp20 = tl.load(in_ptr1 + (32 + x8 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp22 = tl.load(in_ptr1 + (48 + x8 + 64 * x3), xmask, eviction_policy=
'evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 + tmp7
tmp10 = tl_math.exp(tmp9)
tmp11 = tmp8 + tmp10
tmp12 = tl_math.log(tmp11)
tmp13 = tmp0 - tmp12
tmp14 = -tmp13
tmp16 = tmp14 * tmp15
tmp19 = tmp17 + tmp18
tmp21 = tmp19 + tmp20
tmp23 = tmp21 + tmp22
tmp24 = 1.0
tmp25 = triton_helpers.maximum(tmp23, tmp24)
tmp26 = tmp16 / tmp25
tl.store(out_ptr0 + x5, tmp26, xmask)
@triton.jit
def triton_per_fused_mean_sum_2(in_out_ptr0, in_ptr0, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex % 64
r1 = rindex // 64
tmp0 = tl.load(in_ptr0 + (r0 + 256 * r1), None)
tmp1 = tl.load(in_ptr0 + (64 + r0 + 256 * r1), None)
tmp3 = tl.load(in_ptr0 + (128 + r0 + 256 * r1), None)
tmp5 = tl.load(in_ptr0 + (192 + r0 + 256 * r1), None)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = tl.broadcast_to(tmp6, [RBLOCK])
tmp9 = triton_helpers.promote_to_tensor(tl.sum(tmp7, 0))
tmp10 = 256.0
tmp11 = tmp9 / tmp10
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp11, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.float32)
get_raw_stream(0)
triton_poi_fused__log_softmax_0[grid(1024)](arg0_1, buf0, 1024,
XBLOCK=256, num_warps=4, num_stages=1)
del arg0_1
buf1 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.float32)
triton_poi_fused__log_softmax_clamp_div_mul_neg_sum_1[grid(1024)](buf0,
arg1_1, buf1, 1024, XBLOCK=128, num_warps=4, num_stages=1)
del arg1_1
del buf0
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2
del buf2
triton_per_fused_mean_sum_2[grid(1)](buf3, buf1, 1, 256, num_warps=
2, num_stages=1)
del buf1
return buf3,
class LogSoftmaxNew(torch.nn.Module):
def __init__(self, dim):
super(LogSoftmaxNew, self).__init__()
self.dim = dim
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
Tiamat-Tech/just-ask
|
LogSoftmax
| false | 14,520 |
[
"Apache-2.0"
] | 59 |
80725161e12ad0682b4c2091f61a5889a335ba21
|
https://github.com/Tiamat-Tech/just-ask/tree/80725161e12ad0682b4c2091f61a5889a335ba21
|
GlobalAvgPool2d
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/is/cispe7zbbl4nxt2jjus6h5iou2w7htohqj7z2oz6g7nqz6vbpbqr.py
# Topologically Sorted Source Nodes: [avg_pool2d], Original ATen: [aten.avg_pool2d]
# Source node to ATen node mapping:
# avg_pool2d => avg_pool2d
# Graph fragment:
# %avg_pool2d : [num_users=1] = call_function[target=torch.ops.aten.avg_pool2d.default](args = (%arg0_1, [4, 4]), kwargs = {})
triton_poi_fused_avg_pool2d_0 = async_compile.triton('triton_poi_fused_avg_pool2d_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_avg_pool2d_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 16, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_avg_pool2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (16*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (16*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (16*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (16*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (4 + (16*x0)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (5 + (16*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (6 + (16*x0)), xmask, eviction_policy='evict_last')
tmp13 = tl.load(in_ptr0 + (7 + (16*x0)), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr0 + (8 + (16*x0)), xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr0 + (9 + (16*x0)), xmask, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr0 + (10 + (16*x0)), xmask, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (11 + (16*x0)), xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr0 + (12 + (16*x0)), xmask, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr0 + (13 + (16*x0)), xmask, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr0 + (14 + (16*x0)), xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr0 + (15 + (16*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 + tmp0
tmp4 = tmp3 + tmp2
tmp6 = tmp5 + tmp4
tmp8 = tmp7 + tmp6
tmp10 = tmp9 + tmp8
tmp12 = tmp11 + tmp10
tmp14 = tmp13 + tmp12
tmp16 = tmp15 + tmp14
tmp18 = tmp17 + tmp16
tmp20 = tmp19 + tmp18
tmp22 = tmp21 + tmp20
tmp24 = tmp23 + tmp22
tmp26 = tmp25 + tmp24
tmp28 = tmp27 + tmp26
tmp30 = tmp29 + tmp28
tmp31 = 0.0625
tmp32 = tmp30 * tmp31
tl.store(out_ptr0 + (x0), tmp32, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [avg_pool2d], Original ATen: [aten.avg_pool2d]
stream0 = get_raw_stream(0)
triton_poi_fused_avg_pool2d_0.run(arg0_1, buf0, 16, grid=grid(16), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
from torch.nn import functional as F
class GlobalAvgPool2d(nn.Module):
def __init__(self):
"""
Global Average pooling module
"""
super(GlobalAvgPool2d, self).__init__()
def forward(self, x):
"""
The forward function of the GlobalAvgPool2d module
:param x: Input tensor x
:return: A tensor after average pooling
"""
return F.avg_pool2d(x, (x.shape[2], x.shape[3]))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_avg_pool2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 16 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp3 = tl.load(in_ptr0 + (2 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp5 = tl.load(in_ptr0 + (3 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp7 = tl.load(in_ptr0 + (4 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp9 = tl.load(in_ptr0 + (5 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp11 = tl.load(in_ptr0 + (6 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp13 = tl.load(in_ptr0 + (7 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp15 = tl.load(in_ptr0 + (8 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp17 = tl.load(in_ptr0 + (9 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp19 = tl.load(in_ptr0 + (10 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp21 = tl.load(in_ptr0 + (11 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp23 = tl.load(in_ptr0 + (12 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp25 = tl.load(in_ptr0 + (13 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp27 = tl.load(in_ptr0 + (14 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp29 = tl.load(in_ptr0 + (15 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp2 = tmp1 + tmp0
tmp4 = tmp3 + tmp2
tmp6 = tmp5 + tmp4
tmp8 = tmp7 + tmp6
tmp10 = tmp9 + tmp8
tmp12 = tmp11 + tmp10
tmp14 = tmp13 + tmp12
tmp16 = tmp15 + tmp14
tmp18 = tmp17 + tmp16
tmp20 = tmp19 + tmp18
tmp22 = tmp21 + tmp20
tmp24 = tmp23 + tmp22
tmp26 = tmp25 + tmp24
tmp28 = tmp27 + tmp26
tmp30 = tmp29 + tmp28
tmp31 = 0.0625
tmp32 = tmp30 * tmp31
tl.store(out_ptr0 + x0, tmp32, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_avg_pool2d_0[grid(16)](arg0_1, buf0, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del arg0_1
return buf0,
class GlobalAvgPool2dNew(nn.Module):
def __init__(self):
"""
Global Average pooling module
"""
super(GlobalAvgPool2dNew, self).__init__()
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
UniSerj/ai-research
|
GlobalAvgPool2d
| false | 14,521 |
[
"Apache-2.0"
] | 46 |
79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
https://github.com/UniSerj/ai-research/tree/79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
LogSumExpPooling1d
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/nm/cnmzyg2xopr6n4uczsyhkzybdkios4iac7a472zo4vnfd4yzmt7o.py
# Topologically Sorted Source Nodes: [exp_, x, log_], Original ATen: [aten.exp, aten.sum, aten.log]
# Source node to ATen node mapping:
# exp_ => exp
# log_ => log
# x => sum_1
# Graph fragment:
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%arg0_1,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
triton_poi_fused_exp_log_sum_0 = async_compile.triton('triton_poi_fused_exp_log_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_exp_log_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_exp_log_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp1 = tl_math.exp(tmp0)
tmp3 = tl_math.exp(tmp2)
tmp4 = tmp1 + tmp3
tmp6 = tl_math.exp(tmp5)
tmp7 = tmp4 + tmp6
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp7 + tmp9
tmp11 = tl_math.log(tmp10)
tl.store(out_ptr0 + (x0), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/eh/cehurbetgb6w4f4i7vhc3ohi4ycpl3lgwfv7ntj27dap4fq2dbwh.py
# Topologically Sorted Source Nodes: [exp_], Original ATen: [aten.exp]
# Source node to ATen node mapping:
# exp_ => exp
# Graph fragment:
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%arg0_1,), kwargs = {})
# %copy_ : [num_users=0] = call_function[target=torch.ops.aten.copy_.default](args = (%arg0_1, %exp), kwargs = {})
triton_poi_fused_exp_1 = async_compile.triton('triton_poi_fused_exp_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_exp_1', 'mutated_arg_names': ['in_ptr0', 'out_ptr1'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_exp_1(in_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl_math.exp(tmp0)
tl.store(out_ptr1 + (x0), tmp1, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [exp_, x, log_], Original ATen: [aten.exp, aten.sum, aten.log]
stream0 = get_raw_stream(0)
triton_poi_fused_exp_log_sum_0.run(arg0_1, buf0, 64, grid=grid(64), stream=stream0)
# Topologically Sorted Source Nodes: [exp_], Original ATen: [aten.exp]
triton_poi_fused_exp_1.run(arg0_1, arg0_1, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn as nn
class LogSumExpPooling1d(nn.Module):
"""Applies a 1D LogSumExp pooling over an input signal composed of several input planes.
LogSumExp is a smooth approximation of the max function.
Examples:
>>> m = LogSumExpPooling1d()
>>> input = autograd.Variable(torch.randn(4, 5, 10))
>>> m(input).squeeze()
"""
def __init__(self):
super(LogSumExpPooling1d, self).__init__()
def forward(self, x):
x.exp_()
x = x.sum(dim=-1, keepdim=True)
x.log_()
return x
def __repr__(self):
return self.__class__.__name__ + '()'
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch import nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_exp_log_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp1 = tl_math.exp(tmp0)
tmp3 = tl_math.exp(tmp2)
tmp4 = tmp1 + tmp3
tmp6 = tl_math.exp(tmp5)
tmp7 = tmp4 + tmp6
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp7 + tmp9
tmp11 = tl_math.log(tmp10)
tl.store(out_ptr0 + x0, tmp11, xmask)
@triton.jit
def triton_poi_fused_exp_1(in_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl_math.exp(tmp0)
tl.store(out_ptr1 + x0, tmp1, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_exp_log_sum_0[grid(64)](arg0_1, buf0, 64, XBLOCK=
64, num_warps=1, num_stages=1)
triton_poi_fused_exp_1[grid(256)](arg0_1, arg0_1, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del arg0_1
return buf0,
class LogSumExpPooling1dNew(nn.Module):
"""Applies a 1D LogSumExp pooling over an input signal composed of several input planes.
LogSumExp is a smooth approximation of the max function.
Examples:
>>> m = LogSumExpPooling1d()
>>> input = autograd.Variable(torch.randn(4, 5, 10))
>>> m(input).squeeze()
"""
def __init__(self):
super(LogSumExpPooling1dNew, self).__init__()
def __repr__(self):
return self.__class__.__name__ + '()'
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
UKPLab/coling2018-graph-neural-networks-question-answering
|
LogSumExpPooling1d
| false | 14,522 |
[
"Apache-2.0"
] | 164 |
389558d6570195debea570834944507de4f21d65
|
https://github.com/UKPLab/coling2018-graph-neural-networks-question-answering/tree/389558d6570195debea570834944507de4f21d65
|
CircleLoss
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/wo/cwoengrvscqt5xm6wza73dkovcud6nyo3yd56mgv46p7mrciqlzn.py
# Topologically Sorted Source Nodes: [add_2, an, sub_1, mul_2, logit_n, logsumexp, neg, add, add_1, ap, neg_1, sub, mul, logit_p, logsumexp_1, add_3, loss], Original ATen: [aten.add, aten.clamp_min, aten.sub, aten.mul, aten.logsumexp, aten.neg, aten.softplus]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# add_2 => add_2
# add_3 => add_5
# an => clamp_min_1
# ap => clamp_min
# logit_n => mul_3
# logit_p => mul_1
# logsumexp => abs_1, add_3, amax, eq, exp, full_default, log, sub_2, sum_1, where
# logsumexp_1 => abs_2, add_4, amax_1, eq_1, exp_1, full_default_1, log_1, sub_3, sum_2, where_1
# loss => div, exp_2, gt, log1p, mul_4, where_2
# mul => mul
# mul_2 => mul_2
# neg => neg
# neg_1 => neg_1
# sub => sub
# sub_1 => sub_1
# Graph fragment:
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg1_1, 4), kwargs = {})
# %clamp_min_1 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add_2, 0.0), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg1_1, 4), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%clamp_min_1, %sub_1), kwargs = {})
# %mul_3 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, 4), kwargs = {})
# %amax : [num_users=2] = call_function[target=torch.ops.aten.amax.default](args = (%mul_3, [0], True), kwargs = {})
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%amax,), kwargs = {})
# %eq : [num_users=1] = call_function[target=torch.ops.aten.eq.Scalar](args = (%abs_1, inf), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%eq, %full_default, %amax), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_3, %where), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub_2,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [0]), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%log, %squeeze), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%arg0_1,), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%neg, 1), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add, 4), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add_1, 0.0), kwargs = {})
# %neg_1 : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%clamp_min,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, -3), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%neg_1, %sub), kwargs = {})
# %mul_1 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, 4), kwargs = {})
# %amax_1 : [num_users=2] = call_function[target=torch.ops.aten.amax.default](args = (%mul_1, [0], True), kwargs = {})
# %abs_2 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%amax_1,), kwargs = {})
# %eq_1 : [num_users=1] = call_function[target=torch.ops.aten.eq.Scalar](args = (%abs_2, inf), kwargs = {})
# %full_default_1 : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where_1 : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%eq_1, %full_default_1, %amax_1), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_1, %where_1), kwargs = {})
# %exp_1 : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub_3,), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_1, [0]), kwargs = {})
# %log_1 : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_2,), kwargs = {})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%log_1, %squeeze_1), kwargs = {})
# %add_5 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_3, %add_4), kwargs = {})
# %mul_4 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_5, 1.0), kwargs = {})
# %gt : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%mul_4, 20.0), kwargs = {})
# %exp_2 : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%mul_4,), kwargs = {})
# %log1p : [num_users=1] = call_function[target=torch.ops.aten.log1p.default](args = (%exp_2,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%log1p, 1.0), kwargs = {})
# %where_2 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %add_5, %div), kwargs = {})
triton_poi_fused_add_clamp_min_logsumexp_mul_neg_softplus_sub_0 = async_compile.triton('triton_poi_fused_add_clamp_min_logsumexp_mul_neg_softplus_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_min_logsumexp_mul_neg_softplus_sub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_min_logsumexp_mul_neg_softplus_sub_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp8 = tl.load(in_ptr0 + (64 + x0), xmask)
tmp15 = tl.load(in_ptr0 + (128 + x0), xmask)
tmp22 = tl.load(in_ptr0 + (192 + x0), xmask)
tmp44 = tl.load(in_ptr1 + (x0), xmask)
tmp55 = tl.load(in_ptr1 + (64 + x0), xmask)
tmp65 = tl.load(in_ptr1 + (128 + x0), xmask)
tmp75 = tl.load(in_ptr1 + (192 + x0), xmask)
tmp1 = 4.0
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = tmp0 - tmp1
tmp6 = tmp4 * tmp5
tmp7 = tmp6 * tmp1
tmp9 = tmp8 + tmp1
tmp10 = triton_helpers.maximum(tmp9, tmp3)
tmp11 = tmp8 - tmp1
tmp12 = tmp10 * tmp11
tmp13 = tmp12 * tmp1
tmp14 = triton_helpers.maximum(tmp7, tmp13)
tmp16 = tmp15 + tmp1
tmp17 = triton_helpers.maximum(tmp16, tmp3)
tmp18 = tmp15 - tmp1
tmp19 = tmp17 * tmp18
tmp20 = tmp19 * tmp1
tmp21 = triton_helpers.maximum(tmp14, tmp20)
tmp23 = tmp22 + tmp1
tmp24 = triton_helpers.maximum(tmp23, tmp3)
tmp25 = tmp22 - tmp1
tmp26 = tmp24 * tmp25
tmp27 = tmp26 * tmp1
tmp28 = triton_helpers.maximum(tmp21, tmp27)
tmp29 = tl_math.abs(tmp28)
tmp30 = float("inf")
tmp31 = tmp29 == tmp30
tmp32 = tl.where(tmp31, tmp3, tmp28)
tmp33 = tmp7 - tmp32
tmp34 = tl_math.exp(tmp33)
tmp35 = tmp13 - tmp32
tmp36 = tl_math.exp(tmp35)
tmp37 = tmp34 + tmp36
tmp38 = tmp20 - tmp32
tmp39 = tl_math.exp(tmp38)
tmp40 = tmp37 + tmp39
tmp41 = tmp27 - tmp32
tmp42 = tl_math.exp(tmp41)
tmp43 = tmp40 + tmp42
tmp45 = -tmp44
tmp46 = 1.0
tmp47 = tmp45 + tmp46
tmp48 = tmp47 + tmp1
tmp49 = triton_helpers.maximum(tmp48, tmp3)
tmp50 = -tmp49
tmp51 = -3.0
tmp52 = tmp44 - tmp51
tmp53 = tmp50 * tmp52
tmp54 = tmp53 * tmp1
tmp56 = -tmp55
tmp57 = tmp56 + tmp46
tmp58 = tmp57 + tmp1
tmp59 = triton_helpers.maximum(tmp58, tmp3)
tmp60 = -tmp59
tmp61 = tmp55 - tmp51
tmp62 = tmp60 * tmp61
tmp63 = tmp62 * tmp1
tmp64 = triton_helpers.maximum(tmp54, tmp63)
tmp66 = -tmp65
tmp67 = tmp66 + tmp46
tmp68 = tmp67 + tmp1
tmp69 = triton_helpers.maximum(tmp68, tmp3)
tmp70 = -tmp69
tmp71 = tmp65 - tmp51
tmp72 = tmp70 * tmp71
tmp73 = tmp72 * tmp1
tmp74 = triton_helpers.maximum(tmp64, tmp73)
tmp76 = -tmp75
tmp77 = tmp76 + tmp46
tmp78 = tmp77 + tmp1
tmp79 = triton_helpers.maximum(tmp78, tmp3)
tmp80 = -tmp79
tmp81 = tmp75 - tmp51
tmp82 = tmp80 * tmp81
tmp83 = tmp82 * tmp1
tmp84 = triton_helpers.maximum(tmp74, tmp83)
tmp85 = tl_math.abs(tmp84)
tmp86 = tmp85 == tmp30
tmp87 = tl.where(tmp86, tmp3, tmp84)
tmp88 = tmp54 - tmp87
tmp89 = tl_math.exp(tmp88)
tmp90 = tmp63 - tmp87
tmp91 = tl_math.exp(tmp90)
tmp92 = tmp89 + tmp91
tmp93 = tmp73 - tmp87
tmp94 = tl_math.exp(tmp93)
tmp95 = tmp92 + tmp94
tmp96 = tmp83 - tmp87
tmp97 = tl_math.exp(tmp96)
tmp98 = tmp95 + tmp97
tmp99 = tl_math.log(tmp43)
tmp100 = tmp99 + tmp32
tmp101 = tl_math.log(tmp98)
tmp102 = tmp101 + tmp87
tmp103 = tmp100 + tmp102
tmp104 = tmp103 * tmp46
tmp105 = 20.0
tmp106 = tmp104 > tmp105
tmp107 = tl_math.exp(tmp104)
tmp108 = libdevice.log1p(tmp107)
tmp109 = tmp108 * tmp46
tmp110 = tl.where(tmp106, tmp103, tmp109)
tl.store(in_out_ptr0 + (x0), tmp110, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
buf5 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [add_2, an, sub_1, mul_2, logit_n, logsumexp, neg, add, add_1, ap, neg_1, sub, mul, logit_p, logsumexp_1, add_3, loss], Original ATen: [aten.add, aten.clamp_min, aten.sub, aten.mul, aten.logsumexp, aten.neg, aten.softplus]
stream0 = get_raw_stream(0)
triton_poi_fused_add_clamp_min_logsumexp_mul_neg_softplus_sub_0.run(buf5, arg1_1, arg0_1, 64, grid=grid(64), stream=stream0)
del arg0_1
del arg1_1
return (buf5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import Tensor
from torch import nn
from torchvision.transforms import *
class CircleLoss(nn.Module):
def __init__(self, m: 'float', gamma: 'float') ->None:
super(CircleLoss, self).__init__()
self.m = m
self.gamma = gamma
self.soft_plus = nn.Softplus()
def forward(self, sp: 'Tensor', sn: 'Tensor') ->Tensor:
ap = torch.clamp_min(-sp.detach() + 1 + self.m, min=0.0)
an = torch.clamp_min(sn.detach() + self.m, min=0.0)
delta_p = 1 - self.m
delta_n = self.m
logit_p = -ap * (sp - delta_p) * self.gamma
logit_n = an * (sn - delta_n) * self.gamma
loss = self.soft_plus(torch.logsumexp(logit_n, dim=0) + torch.
logsumexp(logit_p, dim=0))
return loss
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'m': 4, 'gamma': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch import nn
from torchvision.transforms import *
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_clamp_min_logsumexp_mul_neg_softplus_sub_0(in_out_ptr0
, in_ptr0, in_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp8 = tl.load(in_ptr0 + (64 + x0), xmask)
tmp15 = tl.load(in_ptr0 + (128 + x0), xmask)
tmp22 = tl.load(in_ptr0 + (192 + x0), xmask)
tmp44 = tl.load(in_ptr1 + x0, xmask)
tmp55 = tl.load(in_ptr1 + (64 + x0), xmask)
tmp65 = tl.load(in_ptr1 + (128 + x0), xmask)
tmp75 = tl.load(in_ptr1 + (192 + x0), xmask)
tmp1 = 4.0
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = tmp0 - tmp1
tmp6 = tmp4 * tmp5
tmp7 = tmp6 * tmp1
tmp9 = tmp8 + tmp1
tmp10 = triton_helpers.maximum(tmp9, tmp3)
tmp11 = tmp8 - tmp1
tmp12 = tmp10 * tmp11
tmp13 = tmp12 * tmp1
tmp14 = triton_helpers.maximum(tmp7, tmp13)
tmp16 = tmp15 + tmp1
tmp17 = triton_helpers.maximum(tmp16, tmp3)
tmp18 = tmp15 - tmp1
tmp19 = tmp17 * tmp18
tmp20 = tmp19 * tmp1
tmp21 = triton_helpers.maximum(tmp14, tmp20)
tmp23 = tmp22 + tmp1
tmp24 = triton_helpers.maximum(tmp23, tmp3)
tmp25 = tmp22 - tmp1
tmp26 = tmp24 * tmp25
tmp27 = tmp26 * tmp1
tmp28 = triton_helpers.maximum(tmp21, tmp27)
tmp29 = tl_math.abs(tmp28)
tmp30 = float('inf')
tmp31 = tmp29 == tmp30
tmp32 = tl.where(tmp31, tmp3, tmp28)
tmp33 = tmp7 - tmp32
tmp34 = tl_math.exp(tmp33)
tmp35 = tmp13 - tmp32
tmp36 = tl_math.exp(tmp35)
tmp37 = tmp34 + tmp36
tmp38 = tmp20 - tmp32
tmp39 = tl_math.exp(tmp38)
tmp40 = tmp37 + tmp39
tmp41 = tmp27 - tmp32
tmp42 = tl_math.exp(tmp41)
tmp43 = tmp40 + tmp42
tmp45 = -tmp44
tmp46 = 1.0
tmp47 = tmp45 + tmp46
tmp48 = tmp47 + tmp1
tmp49 = triton_helpers.maximum(tmp48, tmp3)
tmp50 = -tmp49
tmp51 = -3.0
tmp52 = tmp44 - tmp51
tmp53 = tmp50 * tmp52
tmp54 = tmp53 * tmp1
tmp56 = -tmp55
tmp57 = tmp56 + tmp46
tmp58 = tmp57 + tmp1
tmp59 = triton_helpers.maximum(tmp58, tmp3)
tmp60 = -tmp59
tmp61 = tmp55 - tmp51
tmp62 = tmp60 * tmp61
tmp63 = tmp62 * tmp1
tmp64 = triton_helpers.maximum(tmp54, tmp63)
tmp66 = -tmp65
tmp67 = tmp66 + tmp46
tmp68 = tmp67 + tmp1
tmp69 = triton_helpers.maximum(tmp68, tmp3)
tmp70 = -tmp69
tmp71 = tmp65 - tmp51
tmp72 = tmp70 * tmp71
tmp73 = tmp72 * tmp1
tmp74 = triton_helpers.maximum(tmp64, tmp73)
tmp76 = -tmp75
tmp77 = tmp76 + tmp46
tmp78 = tmp77 + tmp1
tmp79 = triton_helpers.maximum(tmp78, tmp3)
tmp80 = -tmp79
tmp81 = tmp75 - tmp51
tmp82 = tmp80 * tmp81
tmp83 = tmp82 * tmp1
tmp84 = triton_helpers.maximum(tmp74, tmp83)
tmp85 = tl_math.abs(tmp84)
tmp86 = tmp85 == tmp30
tmp87 = tl.where(tmp86, tmp3, tmp84)
tmp88 = tmp54 - tmp87
tmp89 = tl_math.exp(tmp88)
tmp90 = tmp63 - tmp87
tmp91 = tl_math.exp(tmp90)
tmp92 = tmp89 + tmp91
tmp93 = tmp73 - tmp87
tmp94 = tl_math.exp(tmp93)
tmp95 = tmp92 + tmp94
tmp96 = tmp83 - tmp87
tmp97 = tl_math.exp(tmp96)
tmp98 = tmp95 + tmp97
tmp99 = tl_math.log(tmp43)
tmp100 = tmp99 + tmp32
tmp101 = tl_math.log(tmp98)
tmp102 = tmp101 + tmp87
tmp103 = tmp100 + tmp102
tmp104 = tmp103 * tmp46
tmp105 = 20.0
tmp106 = tmp104 > tmp105
tmp107 = tl_math.exp(tmp104)
tmp108 = libdevice.log1p(tmp107)
tmp109 = tmp108 * tmp46
tmp110 = tl.where(tmp106, tmp103, tmp109)
tl.store(in_out_ptr0 + x0, tmp110, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
buf5 = buf2
del buf2
get_raw_stream(0)
triton_poi_fused_add_clamp_min_logsumexp_mul_neg_softplus_sub_0[grid
(64)](buf5, arg1_1, arg0_1, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del arg0_1
del arg1_1
return buf5,
class CircleLossNew(nn.Module):
def __init__(self, m: 'float', gamma: 'float') ->None:
super(CircleLossNew, self).__init__()
self.m = m
self.gamma = gamma
self.soft_plus = nn.Softplus()
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
TxuanYu/Person_reID_baseline_pytorch
|
CircleLoss
| false | 14,523 |
[
"MIT"
] | 3,358 |
10574b17cc8fd1fc8ade88f134679e281fdb01cc
|
https://github.com/TxuanYu/Person_reID_baseline_pytorch/tree/10574b17cc8fd1fc8ade88f134679e281fdb01cc
|
LatentDecoder
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/66/c666qvr725wwogti7syalhhjsndtv2n5sxzt6zi4wlesyjxocpic.py
# Topologically Sorted Source Nodes: [original_output_1], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# original_output_1 => add, rsqrt, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%view_3, [3]), kwargs = {correction: 0, keepdim: True})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-12), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
triton_poi_fused_native_layer_norm_0 = async_compile.triton('triton_poi_fused_native_layer_norm_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_0(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-12
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/qn/cqn57gmzzicd37ayescpre4xsvnluhj6f6dqz6ybhv2ig22gcn6n.py
# Topologically Sorted Source Nodes: [original_output_1, original_output_2], Original ATen: [aten.native_layer_norm, aten.tanh]
# Source node to ATen node mapping:
# original_output_1 => add, add_1, mul, mul_1, rsqrt, sub, var_mean
# original_output_2 => tanh
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%view_3, [3]), kwargs = {correction: 0, keepdim: True})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-12), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_3, %getitem_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_6), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_7), kwargs = {})
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%add_1,), kwargs = {})
triton_poi_fused_native_layer_norm_tanh_1 = async_compile.triton('triton_poi_fused_native_layer_norm_tanh_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_tanh_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_tanh_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tmp9 = libdevice.tanh(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [latent_output], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [original_output], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, buf0, reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf1)
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
# Topologically Sorted Source Nodes: [original_output_1], Original ATen: [aten.native_layer_norm]
stream0 = get_raw_stream(0)
triton_poi_fused_native_layer_norm_0.run(buf1, buf2, buf3, 64, grid=grid(64), stream=stream0)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [original_output_1, original_output_2], Original ATen: [aten.native_layer_norm, aten.tanh]
triton_poi_fused_native_layer_norm_tanh_1.run(buf1, buf2, buf3, primals_6, primals_7, buf4, 256, grid=grid(256), stream=stream0)
del buf2
del buf3
del primals_7
return (buf4, primals_6, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf0, buf1, buf4, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class LatentDecoder(nn.Module):
def __init__(self, hidden_size):
super(LatentDecoder, self).__init__()
self.dense = nn.Linear(hidden_size, hidden_size)
self.dense_mu = nn.Linear(hidden_size, hidden_size)
self.LayerNorm = nn.LayerNorm(hidden_size, eps=1e-12)
self.activation = nn.Tanh()
def forward(self, hidden_states):
latent_output = self.dense(hidden_states)
original_output = self.dense_mu(latent_output)
original_output = self.LayerNorm(original_output)
original_output = self.activation(original_output)
return original_output
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'hidden_size': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_native_layer_norm_0(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-12
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_tanh_1(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tmp9 = libdevice.tanh(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, buf0, reinterpret_tensor(primals_4,
(4, 4), (1, 4), 0), alpha=1, beta=1, out=buf1)
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
get_raw_stream(0)
triton_poi_fused_native_layer_norm_0[grid(64)](buf1, buf2, buf3, 64,
XBLOCK=64, num_warps=1, num_stages=1)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_tanh_1[grid(256)](buf1, buf2,
buf3, primals_6, primals_7, buf4, 256, XBLOCK=256, num_warps=4,
num_stages=1)
del buf2
del buf3
del primals_7
return buf4, primals_6, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf0, buf1, buf4, primals_4
class LatentDecoderNew(nn.Module):
def __init__(self, hidden_size):
super(LatentDecoderNew, self).__init__()
self.dense = nn.Linear(hidden_size, hidden_size)
self.dense_mu = nn.Linear(hidden_size, hidden_size)
self.LayerNorm = nn.LayerNorm(hidden_size, eps=1e-12)
self.activation = nn.Tanh()
def forward(self, input_0):
primals_1 = self.dense.weight
primals_2 = self.dense.bias
primals_4 = self.dense_mu.weight
primals_5 = self.dense_mu.bias
primals_6 = self.LayerNorm.weight
primals_7 = self.LayerNorm.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
|
UKPLab/MMT-Retrieval
|
LatentDecoder
| false | 14,524 |
[
"MIT"
] | 98 |
a31caaeb0da680131bf39dc855e38fdda949f38e
|
https://github.com/UKPLab/MMT-Retrieval/tree/a31caaeb0da680131bf39dc855e38fdda949f38e
|
Project3D
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/32/c32ppse2vdmak5is2nuwq2vbmvddtxyrdxkeqrxyec7bhptha7aa.py
# Topologically Sorted Source Nodes: [cam_points], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# cam_points => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_2,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 48
x1 = (xindex // 48)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (64*x1)), xmask)
tl.store(out_ptr0 + (x2), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/p3/cp33b3l57w5jlxxjkbyxuiiw6erdw3sxbh2kc6ydjd74u3ubmxdx.py
# Topologically Sorted Source Nodes: [sub, pix_coords_3], Original ATen: [aten.sub, aten.mul]
# Source node to ATen node mapping:
# pix_coords_3 => mul
# sub => sub
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%permute_16, 0.5), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, 2), kwargs = {})
triton_poi_fused_mul_sub_1 = async_compile.triton('triton_poi_fused_mul_sub_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_sub_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_sub_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 16) % 2
x0 = xindex % 16
x2 = (xindex // 32)
x3 = xindex % 32
x4 = xindex
tmp7 = tl.load(in_ptr0 + (x0 + (48*x2)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (32 + x0 + (48*x2)), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr0 + (16 + x0 + (48*x2)), xmask, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr0 + (x3 + (48*x2)), xmask)
tmp0 = x1
tmp1 = tl.full([1], 1, tl.int32)
tmp2 = tmp0 == tmp1
tmp3 = tmp1 == tmp1
tmp4 = tl.full([1], 0, tl.int32)
tmp5 = tmp1 == tmp4
tmp6 = tmp4 == tmp4
tmp9 = 1e-07
tmp10 = tmp8 + tmp9
tmp11 = tmp7 / tmp10
tmp12 = 0.3333333333333333
tmp13 = tmp11 * tmp12
tmp14 = tl.where(tmp6, tmp13, tmp11)
tmp16 = tmp15 / tmp10
tmp17 = tl.where(tmp5, tmp13, tmp16)
tmp18 = tl.where(tmp5, tmp14, tmp17)
tmp19 = tmp18 * tmp12
tmp20 = tl.where(tmp3, tmp19, tmp18)
tmp21 = tmp0 == tmp4
tmp23 = tmp22 / tmp10
tmp24 = tl.where(tmp21, tmp13, tmp23)
tmp25 = tl.where(tmp21, tmp14, tmp24)
tmp26 = tl.where(tmp2, tmp19, tmp25)
tmp27 = tl.where(tmp2, tmp20, tmp26)
tmp28 = 0.5
tmp29 = tmp27 - tmp28
tmp30 = 2.0
tmp31 = tmp29 * tmp30
tl.store(out_ptr0 + (x4), tmp31, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 3, 4, 4), (48, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(arg1_1, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(arg0_1, (16, 4, 4), (16, 4, 1), 0), out=buf0)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((4, 3, 4, 4), (48, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cam_points], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(buf0, buf1, 192, grid=grid(192), stream=stream0)
del buf0
buf2 = empty_strided_cuda((12, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cam_points], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf1, (12, 4, 4), (16, 4, 1), 0), reinterpret_tensor(arg2_1, (12, 4, 4), (16, 4, 1), 0), out=buf2)
del arg2_1
del buf1
buf3 = empty_strided_cuda((4, 4, 4, 2), (32, 4, 1, 16), torch.float32)
# Topologically Sorted Source Nodes: [sub, pix_coords_3], Original ATen: [aten.sub, aten.mul]
triton_poi_fused_mul_sub_1.run(buf2, buf3, 128, grid=grid(128), stream=stream0)
del buf2
return (buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 3, 4, 4), (48, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.utils.data
class Project3D(nn.Module):
"""Layer which projects 3D points into a camera with intrinsics K and at position T
"""
def __init__(self, batch_size, height, width, eps=1e-07):
super(Project3D, self).__init__()
self.batch_size = batch_size
self.height = height
self.width = width
self.eps = eps
def forward(self, points, K, T):
P = torch.matmul(K, T)[:, :3, :]
cam_points = torch.matmul(P, points)
pix_coords = cam_points[:, :2, :] / (cam_points[:, 2, :].unsqueeze(
1) + self.eps)
pix_coords = pix_coords.view(self.batch_size, 2, self.height, self.
width)
pix_coords = pix_coords.permute(0, 2, 3, 1)
pix_coords[..., 0] /= self.width - 1
pix_coords[..., 1] /= self.height - 1
pix_coords = (pix_coords - 0.5) * 2
return pix_coords
def get_inputs():
return [torch.rand([4, 3, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {'batch_size': 4, 'height': 4, 'width': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 48
x1 = xindex // 48
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 64 * x1), xmask)
tl.store(out_ptr0 + x2, tmp0, xmask)
@triton.jit
def triton_poi_fused_mul_sub_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 16 % 2
x0 = xindex % 16
x2 = xindex // 32
x3 = xindex % 32
x4 = xindex
tmp7 = tl.load(in_ptr0 + (x0 + 48 * x2), xmask, eviction_policy=
'evict_last')
tmp8 = tl.load(in_ptr0 + (32 + x0 + 48 * x2), xmask, eviction_policy=
'evict_last')
tmp15 = tl.load(in_ptr0 + (16 + x0 + 48 * x2), xmask, eviction_policy=
'evict_last')
tmp22 = tl.load(in_ptr0 + (x3 + 48 * x2), xmask)
tmp0 = x1
tmp1 = tl.full([1], 1, tl.int32)
tmp2 = tmp0 == tmp1
tmp3 = tmp1 == tmp1
tmp4 = tl.full([1], 0, tl.int32)
tmp5 = tmp1 == tmp4
tmp6 = tmp4 == tmp4
tmp9 = 1e-07
tmp10 = tmp8 + tmp9
tmp11 = tmp7 / tmp10
tmp12 = 0.3333333333333333
tmp13 = tmp11 * tmp12
tmp14 = tl.where(tmp6, tmp13, tmp11)
tmp16 = tmp15 / tmp10
tmp17 = tl.where(tmp5, tmp13, tmp16)
tmp18 = tl.where(tmp5, tmp14, tmp17)
tmp19 = tmp18 * tmp12
tmp20 = tl.where(tmp3, tmp19, tmp18)
tmp21 = tmp0 == tmp4
tmp23 = tmp22 / tmp10
tmp24 = tl.where(tmp21, tmp13, tmp23)
tmp25 = tl.where(tmp21, tmp14, tmp24)
tmp26 = tl.where(tmp2, tmp19, tmp25)
tmp27 = tl.where(tmp2, tmp20, tmp26)
tmp28 = 0.5
tmp29 = tmp27 - tmp28
tmp30 = 2.0
tmp31 = tmp29 * tmp30
tl.store(out_ptr0 + x4, tmp31, xmask)
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 3, 4, 4), (48, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(arg1_1, (16, 4, 4), (16, 4, 1
), 0), reinterpret_tensor(arg0_1, (16, 4, 4), (16, 4, 1), 0),
out=buf0)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((4, 3, 4, 4), (48, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(192)](buf0, buf1, 192, XBLOCK=256,
num_warps=4, num_stages=1)
del buf0
buf2 = empty_strided_cuda((12, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf1, (12, 4, 4), (16, 4, 1),
0), reinterpret_tensor(arg2_1, (12, 4, 4), (16, 4, 1), 0), out=buf2
)
del arg2_1
del buf1
buf3 = empty_strided_cuda((4, 4, 4, 2), (32, 4, 1, 16), torch.float32)
triton_poi_fused_mul_sub_1[grid(128)](buf2, buf3, 128, XBLOCK=128,
num_warps=4, num_stages=1)
del buf2
return buf3,
class Project3DNew(nn.Module):
"""Layer which projects 3D points into a camera with intrinsics K and at position T
"""
def __init__(self, batch_size, height, width, eps=1e-07):
super(Project3DNew, self).__init__()
self.batch_size = batch_size
self.height = height
self.width = width
self.eps = eps
def forward(self, input_0, input_1, input_2):
arg2_1 = input_0
arg0_1 = input_1
arg1_1 = input_2
output = call([arg0_1, arg1_1, arg2_1])
return output[0]
|
Uehwan/SimVODIS
|
Project3D
| false | 14,525 |
[
"MIT"
] | 117 |
288ae6f3bf37336f2c829b3a6371793990b23214
|
https://github.com/Uehwan/SimVODIS/tree/288ae6f3bf37336f2c829b3a6371793990b23214
|
KLD
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/qz/cqza6p5fjiie2hfiu5dfjqqugrnzziwuwxzlhzy2aa7khopxjbym.py
# Topologically Sorted Source Nodes: [targets], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# targets => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%arg0_1, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_0 = async_compile.triton('triton_poi_fused__softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x3), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/bg/cbg32drchyezvbfwshguvyopixmzwi2llws7xkhvpdruis76tr2t.py
# Topologically Sorted Source Nodes: [inputs], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# inputs => amax_1, sub_1
# Graph fragment:
# %amax_1 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%arg1_1, [1], True), kwargs = {})
# %sub_1 : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg1_1, %amax_1), kwargs = {})
triton_poi_fused__log_softmax_1 = async_compile.triton('triton_poi_fused__log_softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + (x3), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/pw/cpwz5yhkagu2oekvpj3gpugsclti32lhnve6jzokcqfcufoplweg.py
# Topologically Sorted Source Nodes: [targets, kl_div, inputs], Original ATen: [aten._softmax, aten.xlogy, aten._log_softmax, aten.mul, aten.sub, aten.sum, aten.div]
# Source node to ATen node mapping:
# inputs => exp_1, log, sub_2, sum_2
# kl_div => div_1, eq, full_default, full_default_1, isnan, log_1, mul, mul_1, sub_3, sum_3, where, where_1
# targets => div, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=5] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
# %isnan : [num_users=1] = call_function[target=torch.ops.aten.isnan.default](args = (%div,), kwargs = {})
# %full_default_1 : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], nan), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %eq : [num_users=1] = call_function[target=torch.ops.aten.eq.Scalar](args = (%div, 0), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %log_1 : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%div,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %log_1), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%eq, %full_default, %mul_1), kwargs = {})
# %where_1 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%isnan, %full_default_1, %where), kwargs = {})
# %exp_1 : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub_1,), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_1, [1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_2,), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub_1, %log), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %sub_2), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_1, %mul), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%sub_3,), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sum_3, 4), kwargs = {})
triton_per_fused__log_softmax__softmax_div_mul_sub_sum_xlogy_2 = async_compile.triton('triton_per_fused__log_softmax__softmax_div_mul_sub_sum_xlogy_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__log_softmax__softmax_div_mul_sub_sum_xlogy_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 10, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__log_softmax__softmax_div_mul_sub_sum_xlogy_2(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 16
r2 = (rindex // 64)
tmp0 = tl.load(in_ptr0 + (r3), None)
tmp1 = tl.load(in_ptr0 + (r0 + (64*r2)), None, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr1 + (r3), None)
tmp18 = tl.load(in_ptr1 + (r0 + (64*r2)), None, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr1 + (16 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr1 + (32 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp26 = tl.load(in_ptr1 + (48 + r0 + (64*r2)), None, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tmp9 = libdevice.isnan(tmp8).to(tl.int1)
tmp10 = 0.0
tmp11 = tmp8 == tmp10
tmp12 = tl_math.log(tmp8)
tmp13 = tmp8 * tmp12
tmp14 = tl.where(tmp11, tmp10, tmp13)
tmp15 = float("nan")
tmp16 = tl.where(tmp9, tmp15, tmp14)
tmp19 = tl_math.exp(tmp18)
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp27 = tl_math.exp(tmp26)
tmp28 = tmp25 + tmp27
tmp29 = tl_math.log(tmp28)
tmp30 = tmp17 - tmp29
tmp31 = tmp8 * tmp30
tmp32 = tmp16 - tmp31
tmp33 = tl.broadcast_to(tmp32, [RBLOCK])
tmp35 = triton_helpers.promote_to_tensor(tl.sum(tmp33, 0))
tmp36 = 0.25
tmp37 = tmp35 * tmp36
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp37, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [targets], Original ATen: [aten._softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__softmax_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [inputs], Original ATen: [aten._log_softmax]
triton_poi_fused__log_softmax_1.run(arg1_1, buf2, 256, grid=grid(256), stream=stream0)
del arg1_1
buf3 = empty_strided_cuda((), (), torch.float32)
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [targets, kl_div, inputs], Original ATen: [aten._softmax, aten.xlogy, aten._log_softmax, aten.mul, aten.sub, aten.sum, aten.div]
triton_per_fused__log_softmax__softmax_div_mul_sub_sum_xlogy_2.run(buf4, buf0, buf2, 1, 256, grid=grid(1), stream=stream0)
del buf0
del buf2
return (buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.nn.parallel
import torch.optim
import torch.utils.data
import torch.utils.data.distributed
import torch.nn.functional as F
class KLD(nn.Module):
def forward(self, targets, inputs):
targets = F.softmax(targets, dim=1)
inputs = F.log_softmax(inputs, dim=1)
return F.kl_div(inputs, targets, reduction='batchmean')
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn as nn
import torch.nn.parallel
import torch.optim
import torch.utils.data
import torch.utils.data.distributed
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x3, tmp9, xmask)
@triton.jit
def triton_poi_fused__log_softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + x3, tmp8, xmask)
@triton.jit
def triton_per_fused__log_softmax__softmax_div_mul_sub_sum_xlogy_2(in_out_ptr0,
in_ptr0, in_ptr1, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 16
r2 = rindex // 64
tmp0 = tl.load(in_ptr0 + r3, None)
tmp1 = tl.load(in_ptr0 + (r0 + 64 * r2), None, eviction_policy='evict_last'
)
tmp2 = tl.load(in_ptr0 + (16 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp17 = tl.load(in_ptr1 + r3, None)
tmp18 = tl.load(in_ptr1 + (r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp20 = tl.load(in_ptr1 + (16 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp23 = tl.load(in_ptr1 + (32 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp26 = tl.load(in_ptr1 + (48 + r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tmp9 = libdevice.isnan(tmp8).to(tl.int1)
tmp10 = 0.0
tmp11 = tmp8 == tmp10
tmp12 = tl_math.log(tmp8)
tmp13 = tmp8 * tmp12
tmp14 = tl.where(tmp11, tmp10, tmp13)
tmp15 = float('nan')
tmp16 = tl.where(tmp9, tmp15, tmp14)
tmp19 = tl_math.exp(tmp18)
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp27 = tl_math.exp(tmp26)
tmp28 = tmp25 + tmp27
tmp29 = tl_math.log(tmp28)
tmp30 = tmp17 - tmp29
tmp31 = tmp8 * tmp30
tmp32 = tmp16 - tmp31
tmp33 = tl.broadcast_to(tmp32, [RBLOCK])
tmp35 = triton_helpers.promote_to_tensor(tl.sum(tmp33, 0))
tmp36 = 0.25
tmp37 = tmp35 * tmp36
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp37, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__softmax_0[grid(256)](arg0_1, buf0, 256, XBLOCK=
256, num_warps=4, num_stages=1)
del arg0_1
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__log_softmax_1[grid(256)](arg1_1, buf2, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del arg1_1
buf3 = empty_strided_cuda((), (), torch.float32)
buf4 = buf3
del buf3
triton_per_fused__log_softmax__softmax_div_mul_sub_sum_xlogy_2[grid(1)
](buf4, buf0, buf2, 1, 256, num_warps=2, num_stages=1)
del buf0
del buf2
return buf4,
class KLDNew(nn.Module):
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
UMBCvision/CompReSS
|
KLD
| false | 14,526 |
[
"MIT"
] | 61 |
c5e57edce75da96482fd36eac484c5aca9676945
|
https://github.com/UMBCvision/CompReSS/tree/c5e57edce75da96482fd36eac484c5aca9676945
|
HSwish
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/jj/cjjcpa4jfom3kmx4ufnxtda3bmq466cpemkegyhzep2ymmlsg35l.py
# Topologically Sorted Source Nodes: [add, hardtanh, mul, truediv], Original ATen: [aten.add, aten.hardtanh, aten.mul, aten.div]
# Source node to ATen node mapping:
# add => add
# hardtanh => clamp_max, clamp_min
# mul => mul
# truediv => div
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 3.0), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add, 0.0), kwargs = {})
# %clamp_max : [num_users=1] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min, 6.0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, %clamp_max), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul, 6.0), kwargs = {})
triton_poi_fused_add_div_hardtanh_mul_0 = async_compile.triton('triton_poi_fused_add_div_hardtanh_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_hardtanh_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_hardtanh_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 3.0
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = 6.0
tmp6 = triton_helpers.minimum(tmp4, tmp5)
tmp7 = tmp0 * tmp6
tmp8 = 0.16666666666666666
tmp9 = tmp7 * tmp8
tl.store(out_ptr0 + (x0), tmp9, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add, hardtanh, mul, truediv], Original ATen: [aten.add, aten.hardtanh, aten.mul, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_hardtanh_mul_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class HSwish(nn.Module):
def __init__(self):
"""
An HSwish module
:param inplace: A boolean stating if the operation is inplace
"""
super(HSwish, self).__init__()
self.relu6 = nn.ReLU6()
def forward(self, x):
"""
The forward function of the HSwish module
:param x: Input tensor x
:return: A tensor after HSwish
"""
return x * self.relu6(x + 3.0) / 6.0
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_hardtanh_mul_0(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 3.0
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = 6.0
tmp6 = triton_helpers.minimum(tmp4, tmp5)
tmp7 = tmp0 * tmp6
tmp8 = 0.16666666666666666
tmp9 = tmp7 * tmp8
tl.store(out_ptr0 + x0, tmp9, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_hardtanh_mul_0[grid(256)](arg0_1, buf0,
256, XBLOCK=256, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class HSwishNew(nn.Module):
def __init__(self):
"""
An HSwish module
:param inplace: A boolean stating if the operation is inplace
"""
super(HSwishNew, self).__init__()
self.relu6 = nn.ReLU6()
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
UniSerj/ai-research
|
HSwish
| false | 14,527 |
[
"Apache-2.0"
] | 46 |
79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
https://github.com/UniSerj/ai-research/tree/79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
FreqEncoder
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/js/cjsapvatpnehmjlbhnn5yrkfkd7ffk6crkymmxe4kdpobztziuhd.py
# Topologically Sorted Source Nodes: [mul, sin, mul_1, cos, mul_2, sin_1, mul_3, cos_1, mul_4, sin_2, mul_5, cos_2, mul_6, sin_3, mul_7, cos_3, out], Original ATen: [aten.mul, aten.sin, aten.cos, aten.cat]
# Source node to ATen node mapping:
# cos => cos
# cos_1 => cos_1
# cos_2 => cos_2
# cos_3 => cos_3
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# mul_3 => mul_3
# mul_4 => mul_4
# mul_5 => mul_5
# mul_6 => mul_6
# mul_7 => mul_7
# out => cat
# sin => sin
# sin_1 => sin_1
# sin_2 => sin_2
# sin_3 => sin_3
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %sin : [num_users=1] = call_function[target=torch.ops.aten.sin.default](args = (%mul,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %cos : [num_users=1] = call_function[target=torch.ops.aten.cos.default](args = (%mul_1,), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 2.5198421478271484), kwargs = {})
# %sin_1 : [num_users=1] = call_function[target=torch.ops.aten.sin.default](args = (%mul_2,), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 2.5198421478271484), kwargs = {})
# %cos_1 : [num_users=1] = call_function[target=torch.ops.aten.cos.default](args = (%mul_3,), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 6.349603652954102), kwargs = {})
# %sin_2 : [num_users=1] = call_function[target=torch.ops.aten.sin.default](args = (%mul_4,), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 6.349603652954102), kwargs = {})
# %cos_2 : [num_users=1] = call_function[target=torch.ops.aten.cos.default](args = (%mul_5,), kwargs = {})
# %mul_6 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 16.0), kwargs = {})
# %sin_3 : [num_users=1] = call_function[target=torch.ops.aten.sin.default](args = (%mul_6,), kwargs = {})
# %mul_7 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 16.0), kwargs = {})
# %cos_3 : [num_users=1] = call_function[target=torch.ops.aten.cos.default](args = (%mul_7,), kwargs = {})
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%arg0_1, %sin, %cos, %sin_1, %cos_1, %sin_2, %cos_2, %sin_3, %cos_3], -1), kwargs = {})
triton_poi_fused_cat_cos_mul_sin_0 = async_compile.triton('triton_poi_fused_cat_cos_mul_sin_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: '*fp32', 8: '*fp32', 9: '*fp32', 10: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 5, 9, 10), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_cos_mul_sin_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_cos_mul_sin_0(in_ptr0, out_ptr0, out_ptr1, out_ptr2, out_ptr3, out_ptr4, out_ptr5, out_ptr6, out_ptr7, out_ptr8, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp3 = tl_math.sin(tmp2)
tmp4 = tl_math.cos(tmp2)
tmp5 = 2.5198421478271484
tmp6 = tmp0 * tmp5
tmp7 = tl_math.sin(tmp6)
tmp8 = tl_math.cos(tmp6)
tmp9 = 6.349603652954102
tmp10 = tmp0 * tmp9
tmp11 = tl_math.sin(tmp10)
tmp12 = tl_math.cos(tmp10)
tmp13 = 16.0
tmp14 = tmp0 * tmp13
tmp15 = tl_math.sin(tmp14)
tmp16 = tl_math.cos(tmp14)
tl.store(out_ptr0 + (x0 + (36*x1)), tmp0, xmask)
tl.store(out_ptr1 + (x0 + (36*x1)), tmp3, xmask)
tl.store(out_ptr2 + (x0 + (36*x1)), tmp4, xmask)
tl.store(out_ptr3 + (x0 + (36*x1)), tmp7, xmask)
tl.store(out_ptr4 + (x0 + (36*x1)), tmp8, xmask)
tl.store(out_ptr5 + (x0 + (36*x1)), tmp11, xmask)
tl.store(out_ptr6 + (x0 + (36*x1)), tmp12, xmask)
tl.store(out_ptr7 + (x0 + (36*x1)), tmp15, xmask)
tl.store(out_ptr8 + (x0 + (36*x1)), tmp16, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf9 = empty_strided_cuda((4, 4, 4, 36), (576, 144, 36, 1), torch.float32)
buf0 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 0) # alias
buf1 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 4) # alias
buf2 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 8) # alias
buf3 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 12) # alias
buf4 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 16) # alias
buf5 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 20) # alias
buf6 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 24) # alias
buf7 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 28) # alias
buf8 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 32) # alias
# Topologically Sorted Source Nodes: [mul, sin, mul_1, cos, mul_2, sin_1, mul_3, cos_1, mul_4, sin_2, mul_5, cos_2, mul_6, sin_3, mul_7, cos_3, out], Original ATen: [aten.mul, aten.sin, aten.cos, aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_cos_mul_sin_0.run(arg0_1, buf0, buf1, buf2, buf3, buf4, buf5, buf6, buf7, buf8, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf9, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
class FreqEncoder(nn.Module):
def __init__(self, input_dim, max_freq_log2, N_freqs, log_sampling=True,
include_input=True, periodic_fns=(torch.sin, torch.cos)):
super().__init__()
self.input_dim = input_dim
self.include_input = include_input
self.periodic_fns = periodic_fns
self.output_dim = 0
if self.include_input:
self.output_dim += self.input_dim
self.output_dim += self.input_dim * N_freqs * len(self.periodic_fns)
if log_sampling:
self.freq_bands = 2.0 ** torch.linspace(0.0, max_freq_log2, N_freqs
)
else:
self.freq_bands = torch.linspace(2.0 ** 0.0, 2.0 **
max_freq_log2, N_freqs)
self.freq_bands = self.freq_bands.numpy().tolist()
def forward(self, input, **kwargs):
out = []
if self.include_input:
out.append(input)
for i in range(len(self.freq_bands)):
freq = self.freq_bands[i]
for p_fn in self.periodic_fns:
out.append(p_fn(input * freq))
out = torch.cat(out, dim=-1)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_dim': 4, 'max_freq_log2': 4, 'N_freqs': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_cos_mul_sin_0(in_ptr0, out_ptr0, out_ptr1,
out_ptr2, out_ptr3, out_ptr4, out_ptr5, out_ptr6, out_ptr7, out_ptr8,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp3 = tl_math.sin(tmp2)
tmp4 = tl_math.cos(tmp2)
tmp5 = 2.5198421478271484
tmp6 = tmp0 * tmp5
tmp7 = tl_math.sin(tmp6)
tmp8 = tl_math.cos(tmp6)
tmp9 = 6.349603652954102
tmp10 = tmp0 * tmp9
tmp11 = tl_math.sin(tmp10)
tmp12 = tl_math.cos(tmp10)
tmp13 = 16.0
tmp14 = tmp0 * tmp13
tmp15 = tl_math.sin(tmp14)
tmp16 = tl_math.cos(tmp14)
tl.store(out_ptr0 + (x0 + 36 * x1), tmp0, xmask)
tl.store(out_ptr1 + (x0 + 36 * x1), tmp3, xmask)
tl.store(out_ptr2 + (x0 + 36 * x1), tmp4, xmask)
tl.store(out_ptr3 + (x0 + 36 * x1), tmp7, xmask)
tl.store(out_ptr4 + (x0 + 36 * x1), tmp8, xmask)
tl.store(out_ptr5 + (x0 + 36 * x1), tmp11, xmask)
tl.store(out_ptr6 + (x0 + 36 * x1), tmp12, xmask)
tl.store(out_ptr7 + (x0 + 36 * x1), tmp15, xmask)
tl.store(out_ptr8 + (x0 + 36 * x1), tmp16, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf9 = empty_strided_cuda((4, 4, 4, 36), (576, 144, 36, 1), torch.
float32)
buf0 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 0)
buf1 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 4)
buf2 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 8)
buf3 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 12)
buf4 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 16)
buf5 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 20)
buf6 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 24)
buf7 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 28)
buf8 = reinterpret_tensor(buf9, (4, 4, 4, 4), (576, 144, 36, 1), 32)
get_raw_stream(0)
triton_poi_fused_cat_cos_mul_sin_0[grid(256)](arg0_1, buf0, buf1,
buf2, buf3, buf4, buf5, buf6, buf7, buf8, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg0_1
return buf9,
class FreqEncoderNew(nn.Module):
def __init__(self, input_dim, max_freq_log2, N_freqs, log_sampling=True,
include_input=True, periodic_fns=(torch.sin, torch.cos)):
super().__init__()
self.input_dim = input_dim
self.include_input = include_input
self.periodic_fns = periodic_fns
self.output_dim = 0
if self.include_input:
self.output_dim += self.input_dim
self.output_dim += self.input_dim * N_freqs * len(self.periodic_fns)
if log_sampling:
self.freq_bands = 2.0 ** torch.linspace(0.0, max_freq_log2, N_freqs
)
else:
self.freq_bands = torch.linspace(2.0 ** 0.0, 2.0 **
max_freq_log2, N_freqs)
self.freq_bands = self.freq_bands.numpy().tolist()
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
VCAT19/torch-ngp
|
FreqEncoder
| false | 14,528 |
[
"MIT"
] | 262 |
dcbfe061b30808875a80f12a10a383b51b35f121
|
https://github.com/VCAT19/torch-ngp/tree/dcbfe061b30808875a80f12a10a383b51b35f121
|
RKDLoss
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/jl/cjl6gdowsefrajigrysj4wqaymcm2bsrwo2xgutm5v4zsn37efbb.py
# Topologically Sorted Source Nodes: [pow_1, e_square], Original ATen: [aten.pow, aten.sum]
# Source node to ATen node mapping:
# e_square => sum_1
# pow_1 => pow_1
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%view_1, 2), kwargs = {})
# %sum_1 : [num_users=2] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [1]), kwargs = {})
triton_per_fused_pow_sum_0 = async_compile.triton('triton_per_fused_pow_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_pow_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_pow_sum_0(in_ptr0, out_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp1 = tmp0 * tmp0
tmp2 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp4 = tl.where(xmask, tmp2, 0)
tmp5 = tl.sum(tmp4, 1)[:, None]
tl.store(out_ptr0 + (x0), tmp5, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ap/cap6lykjc7jtwgmu47dbzfsoyufnnnc6q4klmqmepnt2iwmmbz37.py
# Topologically Sorted Source Nodes: [add, mul, sub, res, res_1], Original ATen: [aten.add, aten.mul, aten.sub, aten.clamp, aten.sqrt]
# Source node to ATen node mapping:
# add => add
# mul => mul
# res => clamp_min
# res_1 => sqrt
# sub => sub
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%unsqueeze, %unsqueeze_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mm, 2), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %mul), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub, 1e-12), kwargs = {})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%clamp_min,), kwargs = {})
triton_poi_fused_add_clamp_mul_sqrt_sub_1 = async_compile.triton('triton_poi_fused_add_clamp_mul_sqrt_sub_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_mul_sqrt_sub_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_mul_sqrt_sub_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4)
x0 = xindex % 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_out_ptr0 + (x2), xmask)
tmp2 = tmp0 + tmp1
tmp4 = 2.0
tmp5 = tmp3 * tmp4
tmp6 = tmp2 - tmp5
tmp7 = 1e-12
tmp8 = triton_helpers.maximum(tmp6, tmp7)
tmp9 = libdevice.sqrt(tmp8)
tl.store(in_out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/qs/cqsmgohon47rlgpds2ngf2yb2q2itiyxn4p5dl5rzavw2q36vv5i.py
# Topologically Sorted Source Nodes: [add, mul, sub, res, res_1, setitem], Original ATen: [aten.add, aten.mul, aten.sub, aten.clamp, aten.sqrt, aten.lift_fresh, aten.index_put]
# Source node to ATen node mapping:
# add => add
# mul => mul
# res => clamp_min
# res_1 => sqrt
# setitem => full_default, index_put
# sub => sub
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%unsqueeze, %unsqueeze_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mm, 2), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %mul), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub, 1e-12), kwargs = {})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%clamp_min,), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %index_put : [num_users=2] = call_function[target=torch.ops.aten.index_put_.default](args = (%sqrt, [%lift_fresh_copy_1, %lift_fresh_copy_2], %full_default), kwargs = {})
triton_poi_fused_add_clamp_index_put_lift_fresh_mul_sqrt_sub_2 = async_compile.triton('triton_poi_fused_add_clamp_index_put_lift_fresh_mul_sqrt_sub_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0,), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_index_put_lift_fresh_mul_sqrt_sub_2', 'mutated_arg_names': ['out_ptr0'], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_index_put_lift_fresh_mul_sqrt_sub_2(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tl.full([1], 2, tl.int64)
tmp2 = tmp0 < tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.full([1], 0, tl.int64)
tmp6 = tl.where(tmp4, tmp5, tmp3)
tmp7 = tl.full([1], 3, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.where(tmp8, tmp1, tmp7)
tmp10 = tl.where(tmp2, tmp6, tmp9)
tmp11 = 0.0
tl.store(out_ptr0 + (tl.broadcast_to(5*tmp10, [XBLOCK])), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/dw/cdwmn7qwnbjqpgeeo5kmewgnitdozmuz4ynpnfasfptynvyt4kyr.py
# Topologically Sorted Source Nodes: [gt], Original ATen: [aten.gt]
# Source node to ATen node mapping:
# gt => gt
# Graph fragment:
# %gt : [num_users=1] = call_function[target=torch.ops.aten.gt.Scalar](args = (%index_put, 0), kwargs = {})
triton_poi_fused_gt_3 = async_compile.triton('triton_poi_fused_gt_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i1', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_gt_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_gt_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.0
tmp2 = tmp0 > tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, e_square], Original ATen: [aten.pow, aten.sum]
stream0 = get_raw_stream(0)
triton_per_fused_pow_sum_0.run(arg1_1, buf0, 4, 64, grid=grid(4), stream=stream0)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [prod], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 64), (64, 1), 0), reinterpret_tensor(arg1_1, (64, 4), (1, 64), 0), out=buf1)
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [add, mul, sub, res, res_1], Original ATen: [aten.add, aten.mul, aten.sub, aten.clamp, aten.sqrt]
triton_poi_fused_add_clamp_mul_sqrt_sub_1.run(buf2, buf0, 16, grid=grid(16), stream=stream0)
del buf0
# Topologically Sorted Source Nodes: [add, mul, sub, res, res_1, setitem], Original ATen: [aten.add, aten.mul, aten.sub, aten.clamp, aten.sqrt, aten.lift_fresh, aten.index_put]
triton_poi_fused_add_clamp_index_put_lift_fresh_mul_sqrt_sub_2.run(buf2, 4, grid=grid(4), stream=stream0)
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
# Topologically Sorted Source Nodes: [gt], Original ATen: [aten.gt]
triton_poi_fused_gt_3.run(buf2, buf4, 16, grid=grid(16), stream=stream0)
return (buf2, buf4, reinterpret_tensor(arg0_1, (4, 64), (64, 1), 0), reinterpret_tensor(arg1_1, (4, 64), (64, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
import torch.nn.functional as F
class RKDLoss(nn.Module):
"""Relational Knowledge Disitllation, CVPR2019"""
def __init__(self, w_d=25, w_a=50):
super(RKDLoss, self).__init__()
self.w_d = w_d
self.w_a = w_a
def forward(self, f_s, f_t):
student = f_s.view(f_s.shape[0], -1)
teacher = f_t.view(f_t.shape[0], -1)
with torch.no_grad():
t_d = self.pdist(teacher, squared=False)
mean_td = t_d[t_d > 0].mean()
t_d = t_d / mean_td
d = self.pdist(student, squared=False)
mean_d = d[d > 0].mean()
d = d / mean_d
loss_d = F.smooth_l1_loss(d, t_d)
with torch.no_grad():
td = teacher.unsqueeze(0) - teacher.unsqueeze(1)
norm_td = F.normalize(td, p=2, dim=2)
t_angle = torch.bmm(norm_td, norm_td.transpose(1, 2)).view(-1)
sd = student.unsqueeze(0) - student.unsqueeze(1)
norm_sd = F.normalize(sd, p=2, dim=2)
s_angle = torch.bmm(norm_sd, norm_sd.transpose(1, 2)).view(-1)
loss_a = F.smooth_l1_loss(s_angle, t_angle)
loss = self.w_d * loss_d + self.w_a * loss_a
return loss
@staticmethod
def pdist(e, squared=False, eps=1e-12):
e_square = e.pow(2).sum(dim=1)
prod = e @ e.t()
res = (e_square.unsqueeze(1) + e_square.unsqueeze(0) - 2 * prod).clamp(
min=eps)
if not squared:
res = res.sqrt()
res = res.clone()
res[range(len(e)), range(len(e))] = 0
return res
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_pow_sum_0(in_ptr0, out_ptr0, xnumel, rnumel, XBLOCK:
tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp1 = tmp0 * tmp0
tmp2 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp4 = tl.where(xmask, tmp2, 0)
tmp5 = tl.sum(tmp4, 1)[:, None]
tl.store(out_ptr0 + x0, tmp5, xmask)
@triton.jit
def triton_poi_fused_add_clamp_mul_sqrt_sub_1(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4
x0 = xindex % 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_out_ptr0 + x2, xmask)
tmp2 = tmp0 + tmp1
tmp4 = 2.0
tmp5 = tmp3 * tmp4
tmp6 = tmp2 - tmp5
tmp7 = 1e-12
tmp8 = triton_helpers.maximum(tmp6, tmp7)
tmp9 = libdevice.sqrt(tmp8)
tl.store(in_out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused_add_clamp_index_put_lift_fresh_mul_sqrt_sub_2(out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tl.full([1], 2, tl.int64)
tmp2 = tmp0 < tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.full([1], 0, tl.int64)
tmp6 = tl.where(tmp4, tmp5, tmp3)
tmp7 = tl.full([1], 3, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.where(tmp8, tmp1, tmp7)
tmp10 = tl.where(tmp2, tmp6, tmp9)
tmp11 = 0.0
tl.store(out_ptr0 + tl.broadcast_to(5 * tmp10, [XBLOCK]), tmp11, xmask)
@triton.jit
def triton_poi_fused_gt_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.0
tmp2 = tmp0 > tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4,), (1,), torch.float32)
get_raw_stream(0)
triton_per_fused_pow_sum_0[grid(4)](arg1_1, buf0, 4, 64, XBLOCK=1,
num_warps=2, num_stages=1)
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(arg1_1, (4, 64), (64, 1), 0),
reinterpret_tensor(arg1_1, (64, 4), (1, 64), 0), out=buf1)
buf2 = buf1
del buf1
triton_poi_fused_add_clamp_mul_sqrt_sub_1[grid(16)](buf2, buf0, 16,
XBLOCK=16, num_warps=1, num_stages=1)
del buf0
triton_poi_fused_add_clamp_index_put_lift_fresh_mul_sqrt_sub_2[grid(4)
](buf2, 4, XBLOCK=4, num_warps=1, num_stages=1)
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.bool)
triton_poi_fused_gt_3[grid(16)](buf2, buf4, 16, XBLOCK=16,
num_warps=1, num_stages=1)
return buf2, buf4, reinterpret_tensor(arg0_1, (4, 64), (64, 1), 0
), reinterpret_tensor(arg1_1, (4, 64), (64, 1), 0)
class RKDLossNew(nn.Module):
"""Relational Knowledge Disitllation, CVPR2019"""
def __init__(self, w_d=25, w_a=50):
super(RKDLossNew, self).__init__()
self.w_d = w_d
self.w_a = w_a
@staticmethod
def pdist(e, squared=False, eps=1e-12):
e_square = e.pow(2).sum(dim=1)
prod = e @ e.t()
res = (e_square.unsqueeze(1) + e_square.unsqueeze(0) - 2 * prod).clamp(
min=eps)
if not squared:
res = res.sqrt()
res = res.clone()
res[range(len(e)), range(len(e))] = 0
return res
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
UBCDingXin/RepDistiller
|
RKDLoss
| false | 14,529 |
[
"BSD-2-Clause"
] | 1,347 |
dcc043277f2820efafd679ffb82b8e8195b7e222
|
https://github.com/UBCDingXin/RepDistiller/tree/dcc043277f2820efafd679ffb82b8e8195b7e222
|
FactorTransfer
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/t5/ct56alqgoynhiombzpq5fm5wf3zh32egabawyhxhvug4d66ecouu.py
# Topologically Sorted Source Nodes: [normalize, normalize_1, sub], Original ATen: [aten.linalg_vector_norm, aten.div, aten.sub]
# Source node to ATen node mapping:
# normalize => div, pow_2, sum_1
# normalize_1 => div_1, pow_5, sum_2
# sub => sub
# Graph fragment:
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%view, 2.0), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_2, [1], True), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%view, %expand), kwargs = {})
# %pow_5 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%view_1, 2.0), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_5, [1], True), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%view_1, %expand_1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%div, %div_1), kwargs = {})
triton_per_fused_div_linalg_vector_norm_sub_0 = async_compile.triton('triton_per_fused_div_linalg_vector_norm_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 16],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_div_linalg_vector_norm_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 2, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_div_linalg_vector_norm_sub_0(in_ptr0, in_ptr1, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp2 = tl.load(in_ptr0 + (16 + r1 + (64*x0)), xmask, other=0.0)
tmp5 = tl.load(in_ptr0 + (32 + r1 + (64*x0)), xmask, other=0.0)
tmp8 = tl.load(in_ptr0 + (48 + r1 + (64*x0)), xmask, other=0.0)
tmp18 = tl.load(in_ptr1 + (r1 + (64*x0)), xmask, other=0.0)
tmp20 = tl.load(in_ptr1 + (16 + r1 + (64*x0)), xmask, other=0.0)
tmp23 = tl.load(in_ptr1 + (32 + r1 + (64*x0)), xmask, other=0.0)
tmp26 = tl.load(in_ptr1 + (48 + r1 + (64*x0)), xmask, other=0.0)
tmp1 = tmp0 * tmp0
tmp3 = tmp2 * tmp2
tmp4 = tmp1 + tmp3
tmp6 = tmp5 * tmp5
tmp7 = tmp4 + tmp6
tmp9 = tmp8 * tmp8
tmp10 = tmp7 + tmp9
tmp11 = 4.0
tmp12 = tmp10 / tmp11
tmp13 = tmp12 * tmp12
tmp14 = tl.broadcast_to(tmp13, [XBLOCK, RBLOCK])
tmp16 = tl.where(xmask, tmp14, 0)
tmp17 = tl.sum(tmp16, 1)[:, None]
tmp19 = tmp18 * tmp18
tmp21 = tmp20 * tmp20
tmp22 = tmp19 + tmp21
tmp24 = tmp23 * tmp23
tmp25 = tmp22 + tmp24
tmp27 = tmp26 * tmp26
tmp28 = tmp25 + tmp27
tmp29 = tmp28 / tmp11
tmp30 = tmp29 * tmp29
tmp31 = tl.broadcast_to(tmp30, [XBLOCK, RBLOCK])
tmp33 = tl.where(xmask, tmp31, 0)
tmp34 = tl.sum(tmp33, 1)[:, None]
tmp35 = libdevice.sqrt(tmp17)
tmp36 = 1e-12
tmp37 = triton_helpers.maximum(tmp35, tmp36)
tmp38 = tmp12 / tmp37
tmp39 = libdevice.sqrt(tmp34)
tmp40 = triton_helpers.maximum(tmp39, tmp36)
tmp41 = tmp29 / tmp40
tmp42 = tmp38 - tmp41
tl.store(out_ptr2 + (r1 + (16*x0)), tmp42, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/gl/cglmfxdo76j2ab7llzszfnvwdkitvivbxyr5x7kknis3tn2lhrjb.py
# Topologically Sorted Source Nodes: [abs_1, mean_2], Original ATen: [aten.abs, aten.mean]
# Source node to ATen node mapping:
# abs_1 => abs_1
# mean_2 => mean_2
# Graph fragment:
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub,), kwargs = {})
# %mean_2 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_1,), kwargs = {})
triton_per_fused_abs_mean_1 = async_compile.triton('triton_per_fused_abs_mean_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_abs_mean_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_abs_mean_1(in_out_ptr0, in_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = tl_math.abs(tmp0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp4 = tl.sum(tmp2, 1)[:, None]
tmp5 = 64.0
tmp6 = tmp4 / tmp5
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp6, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 16), (16, 1), torch.float32)
# Topologically Sorted Source Nodes: [normalize, normalize_1, sub], Original ATen: [aten.linalg_vector_norm, aten.div, aten.sub]
stream0 = get_raw_stream(0)
triton_per_fused_div_linalg_vector_norm_sub_0.run(arg0_1, arg1_1, buf2, 4, 16, grid=grid(4), stream=stream0)
del arg0_1
del arg1_1
buf3 = empty_strided_cuda((), (), torch.float32)
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [abs_1, mean_2], Original ATen: [aten.abs, aten.mean]
triton_per_fused_abs_mean_1.run(buf4, buf2, 1, 64, grid=grid(1), stream=stream0)
del buf2
return (buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
import torch.nn.functional as F
class FactorTransfer(nn.Module):
"""Paraphrasing Complex Network: Network Compression via Factor Transfer, NeurIPS 2018"""
def __init__(self, p1=2, p2=1):
super(FactorTransfer, self).__init__()
self.p1 = p1
self.p2 = p2
def forward(self, f_s, f_t):
return self.factor_loss(f_s, f_t)
def factor_loss(self, f_s, f_t):
s_H, t_H = f_s.shape[2], f_t.shape[2]
if s_H > t_H:
f_s = F.adaptive_avg_pool2d(f_s, (t_H, t_H))
elif s_H < t_H:
f_t = F.adaptive_avg_pool2d(f_t, (s_H, s_H))
else:
pass
if self.p2 == 1:
return (self.factor(f_s) - self.factor(f_t)).abs().mean()
else:
return (self.factor(f_s) - self.factor(f_t)).pow(self.p2).mean()
def factor(self, f):
return F.normalize(f.pow(self.p1).mean(1).view(f.size(0), -1))
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch import nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_div_linalg_vector_norm_sub_0(in_ptr0, in_ptr1,
out_ptr2, xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp2 = tl.load(in_ptr0 + (16 + r1 + 64 * x0), xmask, other=0.0)
tmp5 = tl.load(in_ptr0 + (32 + r1 + 64 * x0), xmask, other=0.0)
tmp8 = tl.load(in_ptr0 + (48 + r1 + 64 * x0), xmask, other=0.0)
tmp18 = tl.load(in_ptr1 + (r1 + 64 * x0), xmask, other=0.0)
tmp20 = tl.load(in_ptr1 + (16 + r1 + 64 * x0), xmask, other=0.0)
tmp23 = tl.load(in_ptr1 + (32 + r1 + 64 * x0), xmask, other=0.0)
tmp26 = tl.load(in_ptr1 + (48 + r1 + 64 * x0), xmask, other=0.0)
tmp1 = tmp0 * tmp0
tmp3 = tmp2 * tmp2
tmp4 = tmp1 + tmp3
tmp6 = tmp5 * tmp5
tmp7 = tmp4 + tmp6
tmp9 = tmp8 * tmp8
tmp10 = tmp7 + tmp9
tmp11 = 4.0
tmp12 = tmp10 / tmp11
tmp13 = tmp12 * tmp12
tmp14 = tl.broadcast_to(tmp13, [XBLOCK, RBLOCK])
tmp16 = tl.where(xmask, tmp14, 0)
tmp17 = tl.sum(tmp16, 1)[:, None]
tmp19 = tmp18 * tmp18
tmp21 = tmp20 * tmp20
tmp22 = tmp19 + tmp21
tmp24 = tmp23 * tmp23
tmp25 = tmp22 + tmp24
tmp27 = tmp26 * tmp26
tmp28 = tmp25 + tmp27
tmp29 = tmp28 / tmp11
tmp30 = tmp29 * tmp29
tmp31 = tl.broadcast_to(tmp30, [XBLOCK, RBLOCK])
tmp33 = tl.where(xmask, tmp31, 0)
tmp34 = tl.sum(tmp33, 1)[:, None]
tmp35 = libdevice.sqrt(tmp17)
tmp36 = 1e-12
tmp37 = triton_helpers.maximum(tmp35, tmp36)
tmp38 = tmp12 / tmp37
tmp39 = libdevice.sqrt(tmp34)
tmp40 = triton_helpers.maximum(tmp39, tmp36)
tmp41 = tmp29 / tmp40
tmp42 = tmp38 - tmp41
tl.store(out_ptr2 + (r1 + 16 * x0), tmp42, xmask)
@triton.jit
def triton_per_fused_abs_mean_1(in_out_ptr0, in_ptr0, xnumel, rnumel,
XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = tl_math.abs(tmp0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp4 = tl.sum(tmp2, 1)[:, None]
tmp5 = 64.0
tmp6 = tmp4 / tmp5
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp6, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 16), (16, 1), torch.float32)
get_raw_stream(0)
triton_per_fused_div_linalg_vector_norm_sub_0[grid(4)](arg0_1,
arg1_1, buf2, 4, 16, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
buf3 = empty_strided_cuda((), (), torch.float32)
buf4 = buf3
del buf3
triton_per_fused_abs_mean_1[grid(1)](buf4, buf2, 1, 64, XBLOCK=1,
num_warps=2, num_stages=1)
del buf2
return buf4,
class FactorTransferNew(nn.Module):
"""Paraphrasing Complex Network: Network Compression via Factor Transfer, NeurIPS 2018"""
def __init__(self, p1=2, p2=1):
super(FactorTransferNew, self).__init__()
self.p1 = p1
self.p2 = p2
def factor_loss(self, f_s, f_t):
s_H, t_H = f_s.shape[2], f_t.shape[2]
if s_H > t_H:
f_s = F.adaptive_avg_pool2d(f_s, (t_H, t_H))
elif s_H < t_H:
f_t = F.adaptive_avg_pool2d(f_t, (s_H, s_H))
else:
pass
if self.p2 == 1:
return (self.factor(f_s) - self.factor(f_t)).abs().mean()
else:
return (self.factor(f_s) - self.factor(f_t)).pow(self.p2).mean()
def factor(self, f):
return F.normalize(f.pow(self.p1).mean(1).view(f.size(0), -1))
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
UBCDingXin/RepDistiller
|
FactorTransfer
| false | 14,530 |
[
"BSD-2-Clause"
] | 1,347 |
dcc043277f2820efafd679ffb82b8e8195b7e222
|
https://github.com/UBCDingXin/RepDistiller/tree/dcc043277f2820efafd679ffb82b8e8195b7e222
|
BertAttention
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/x2/cx2hdvwyo7m5jvhhvtugzxqvmy6z4nsfhkkjhvgzbbm3cb6dsum2.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %mul_scalar : [num_users=1] = call_function[target=torch.ops.aten.mul.Scalar](args = (%permute_default, 1.0), kwargs = {})
# %clone_default : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_default,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_0 = async_compile.triton('triton_poi_fused_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_0(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (16*y1)), xmask & ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (y0), ymask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + (x2 + (4*y3)), tmp4, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/iz/ciztqj6kop3hxov46yrmzprkzfir3eljcic4mkqznz2j5cfeaudr.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %add_tensor : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_default_2, %primals_8), kwargs = {})
# %amax_default : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%add_tensor, [-1], True), kwargs = {})
# %sub_tensor : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_tensor, %amax_default), kwargs = {})
# %exp_default : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_tensor,), kwargs = {})
# %sum_dim_int_list : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_default, [-1], True), kwargs = {})
# %eq_scalar : [num_users=1] = call_function[target=torch.ops.aten.eq.Scalar](args = (%add_tensor, -inf), kwargs = {})
# %logical_not_default : [num_users=1] = call_function[target=torch.ops.aten.logical_not.default](args = (%eq_scalar,), kwargs = {})
# %any_dim : [num_users=1] = call_function[target=torch.ops.aten.any.dim](args = (%logical_not_default, -1, True), kwargs = {})
triton_poi_fused_1 = async_compile.triton('triton_poi_fused_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*i1', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
tmp0 = tl.load(in_ptr0 + (4*x2), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + (4*x2)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x2)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = triton_helpers.maximum(tmp2, tmp5)
tmp9 = tmp7 + tmp8
tmp10 = triton_helpers.maximum(tmp6, tmp9)
tmp13 = tmp11 + tmp12
tmp14 = triton_helpers.maximum(tmp10, tmp13)
tmp15 = tmp2 - tmp14
tmp16 = tl_math.exp(tmp15)
tmp17 = tmp5 - tmp14
tmp18 = tl_math.exp(tmp17)
tmp19 = tmp16 + tmp18
tmp20 = tmp9 - tmp14
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp23 = tmp13 - tmp14
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp26 = float("-inf")
tmp27 = tmp2 == tmp26
tmp28 = tmp27 == 0
tmp29 = tmp28.to(tl.int64)
tmp30 = (tmp29 != 0)
tmp31 = tmp5 == tmp26
tmp32 = tmp31 == 0
tmp33 = tmp32.to(tl.int64)
tmp34 = (tmp33 != 0)
tmp35 = tmp30 | tmp34
tmp36 = tmp9 == tmp26
tmp37 = tmp36 == 0
tmp38 = tmp37.to(tl.int64)
tmp39 = (tmp38 != 0)
tmp40 = tmp35 | tmp39
tmp41 = tmp13 == tmp26
tmp42 = tmp41 == 0
tmp43 = tmp42.to(tl.int64)
tmp44 = (tmp43 != 0)
tmp45 = tmp40 | tmp44
tl.store(out_ptr0 + (x2), tmp14, xmask)
tl.store(out_ptr1 + (x2), tmp25, xmask)
tl.store(out_ptr2 + (x2), tmp45, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/x5/cx5uvbfethxuwwkwxf3xaualzhlcwqsz4jxqpbhintggaypzjwqf.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %add_tensor : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_default_2, %primals_8), kwargs = {})
# %amax_default : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%add_tensor, [-1], True), kwargs = {})
# %sub_tensor : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_tensor, %amax_default), kwargs = {})
# %exp_default : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_tensor,), kwargs = {})
# %div_tensor : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_default, %sum_dim_int_list), kwargs = {})
# %logical_not_default_1 : [num_users=1] = call_function[target=torch.ops.aten.logical_not.default](args = (%any_dim,), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([4, 4, 4, 4], 0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where_self : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%logical_not_default_1, %full_default, %div_tensor), kwargs = {})
triton_poi_fused_2 = async_compile.triton('triton_poi_fused_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i1', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_2(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = (xindex // 4)
x4 = xindex
x5 = xindex % 64
tmp0 = tl.load(in_ptr0 + (x3), xmask, eviction_policy='evict_last').to(tl.int1)
tmp2 = tl.load(in_out_ptr0 + (x4), xmask)
tmp3 = tl.load(in_ptr1 + (x5), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr2 + (x3), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr3 + (x3), xmask, eviction_policy='evict_last')
tmp1 = tmp0 == 0
tmp4 = tmp2 + tmp3
tmp6 = tmp4 - tmp5
tmp7 = tl_math.exp(tmp6)
tmp9 = tmp7 / tmp8
tmp10 = 0.0
tmp11 = tl.where(tmp1, tmp10, tmp9)
tl.store(in_out_ptr0 + (x4), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/vv/cvvnhithjvmvhfjufxwwzclfobkrgbyyteg66hp24r675f7elw4c.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %clone_default_2 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_default_3,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_3 = async_compile.triton('triton_poi_fused_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_3(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (16*y1)), xmask & ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (y0), ymask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2 + (4*y3)), tmp2, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/6t/c6t5a5ere3lqjiu7zh3uu4oxmpdoujdaqqmeunxqapgzo4m74uav.py
# Topologically Sorted Source Nodes: [context_layer_1], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# context_layer_1 => clone_4
# Graph fragment:
# %clone_4 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%permute_7,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_4 = async_compile.triton('triton_poi_fused_clone_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (16*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/6m/c6mhj5zwirfhy5e4o45uaeov72uwfby4udubpm2fcz42iqvs2g57.py
# Topologically Sorted Source Nodes: [add_1, hidden_states_2], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# add_1 => add_1
# hidden_states_2 => var_mean
# Graph fragment:
# %add_1 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_17, %primals_3), kwargs = {})
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add_1, [2]), kwargs = {correction: 0, keepdim: True})
triton_poi_fused_add_native_layer_norm_5 = async_compile.triton('triton_poi_fused_add_native_layer_norm_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_5(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + (x0), tmp16, xmask)
tl.store(out_ptr1 + (x0), tmp28, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/i6/ci6ua4lfqzz3v6lbsh75noa7k5ird3udb6b5bjh7gxx4qxuz7gz3.py
# Topologically Sorted Source Nodes: [add_1, hidden_states_2], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# add_1 => add_1
# hidden_states_2 => add_2, add_3, mul, mul_1, rsqrt, sub_1
# Graph fragment:
# %add_1 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_17, %primals_3), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-12), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_2,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_1, %getitem_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_11), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_12), kwargs = {})
triton_poi_fused_add_native_layer_norm_6 = async_compile.triton('triton_poi_fused_add_native_layer_norm_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_6(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-12
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + (x2), tmp13, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_9, (4, 4), (4, 1))
assert_size_stride(primals_10, (4, ), (1, ))
assert_size_stride(primals_11, (4, ), (1, ))
assert_size_stride(primals_12, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (16, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (16, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (16, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf2)
del primals_6
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
stream0 = get_raw_stream(0)
triton_poi_fused_0.run(buf0, primals_2, buf3, 16, 4, grid=grid(16, 4), stream=stream0)
del primals_2
buf4 = reinterpret_tensor(buf0, (4, 4, 1, 4), (16, 4, 4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_0.run(buf1, primals_5, buf4, 16, 4, grid=grid(16, 4), stream=stream0)
del primals_5
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = reinterpret_tensor(buf1, (4, 4, 4, 1), (16, 4, 1, 64), 0); del buf1 # reuse
buf7 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf8 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.bool)
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_1.run(buf5, primals_8, buf6, buf7, buf8, 64, grid=grid(64), stream=stream0)
buf9 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf5 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_2.run(buf9, buf8, primals_8, buf6, buf7, 256, grid=grid(256), stream=stream0)
del buf8
del primals_8
buf10 = reinterpret_tensor(buf7, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf7 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_3.run(buf2, primals_7, buf10, 16, 4, grid=grid(16, 4), stream=stream0)
del primals_7
buf11 = reinterpret_tensor(buf2, (16, 4, 1), (4, 1, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.bmm(reinterpret_tensor(buf9, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf10, (16, 4, 1), (4, 1, 0), 0), out=buf11)
buf12 = reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf6 # reuse
# Topologically Sorted Source Nodes: [context_layer_1], Original ATen: [aten.clone]
triton_poi_fused_clone_4.run(buf11, buf12, 16, 4, grid=grid(16, 4), stream=stream0)
buf13 = reinterpret_tensor(buf11, (16, 4), (4, 1), 0); del buf11 # reuse
# Topologically Sorted Source Nodes: [hidden_states], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_10, reinterpret_tensor(buf12, (16, 4), (4, 1), 0), reinterpret_tensor(primals_9, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf13)
del primals_10
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf15 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
# Topologically Sorted Source Nodes: [add_1, hidden_states_2], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_5.run(buf13, primals_3, buf14, buf15, 16, grid=grid(16), stream=stream0)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add_1, hidden_states_2], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_6.run(buf13, primals_3, buf14, buf15, primals_11, primals_12, buf16, 64, grid=grid(64), stream=stream0)
del buf14
del buf15
del primals_12
return (buf16, primals_3, primals_11, buf9, reinterpret_tensor(buf10, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf3, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0), reinterpret_tensor(buf12, (16, 4), (4, 1), 0), buf13, primals_9, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
from _paritybench_helpers import _mock_config
import math
import torch
from torch import nn
from torch.nn import LayerNorm
class BertSelfAttention(nn.Module):
def __init__(self, config):
super(BertSelfAttention, self).__init__()
if config.hidden_size % config.num_attention_heads != 0:
raise ValueError(
'The hidden size (%d) is not a multiple of the number of attention heads (%d)'
% (config.hidden_size, config.num_attention_heads))
self.num_attention_heads = config.num_attention_heads
self.attention_head_size = int(config.hidden_size / config.
num_attention_heads)
self.all_head_size = (self.num_attention_heads * self.
attention_head_size)
self.query = nn.Linear(config.hidden_size, self.all_head_size)
self.key = nn.Linear(config.hidden_size, self.all_head_size)
self.value = nn.Linear(config.hidden_size, self.all_head_size)
self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
def transpose_for_scores(self, x):
new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.
attention_head_size)
x = x.view(*new_x_shape)
return x.permute(0, 2, 1, 3)
def forward(self, hidden_states, attention_mask):
mixed_query_layer = self.query(hidden_states)
mixed_key_layer = self.key(hidden_states)
mixed_value_layer = self.value(hidden_states)
query_layer = self.transpose_for_scores(mixed_query_layer)
key_layer = self.transpose_for_scores(mixed_key_layer)
value_layer = self.transpose_for_scores(mixed_value_layer)
attention_scores = torch.matmul(query_layer, key_layer.transpose(-1,
-2))
attention_scores = attention_scores / math.sqrt(self.
attention_head_size)
attention_scores = attention_scores + attention_mask
attention_probs = nn.Softmax(dim=-1)(attention_scores)
attention_probs = self.dropout(attention_probs)
context_layer = torch.matmul(attention_probs, value_layer)
context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
new_context_layer_shape = context_layer.size()[:-2] + (self.
all_head_size,)
context_layer = context_layer.view(*new_context_layer_shape)
return context_layer
class BertSelfOutput(nn.Module):
def __init__(self, config):
super(BertSelfOutput, self).__init__()
self.dense = nn.Linear(config.hidden_size, config.hidden_size)
self.LayerNorm = LayerNorm(config.hidden_size, eps=1e-12)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
def forward(self, hidden_states, input_tensor):
hidden_states = self.dense(hidden_states)
hidden_states = self.dropout(hidden_states)
hidden_states = self.LayerNorm(hidden_states + input_tensor)
return hidden_states
class BertAttention(nn.Module):
def __init__(self, config):
super(BertAttention, self).__init__()
self.self = BertSelfAttention(config)
self.output = BertSelfOutput(config)
def forward(self, input_tensor, attention_mask):
self_output = self.self(input_tensor, attention_mask)
attention_output = self.output(self_output, input_tensor)
return attention_output
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'config': _mock_config(hidden_size=4, num_attention_heads=
4, attention_probs_dropout_prob=0.5, hidden_dropout_prob=0.5)}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import math
from torch import nn
from torch.nn import LayerNorm
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_0(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel, YBLOCK:
tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 16 * y1), xmask & ymask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + y0, ymask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + (x2 + 4 * y3), tmp4, xmask & ymask)
@triton.jit
def triton_poi_fused_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, out_ptr2,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
tmp0 = tl.load(in_ptr0 + 4 * x2, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x2), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + 4 * x2), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp12 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = triton_helpers.maximum(tmp2, tmp5)
tmp9 = tmp7 + tmp8
tmp10 = triton_helpers.maximum(tmp6, tmp9)
tmp13 = tmp11 + tmp12
tmp14 = triton_helpers.maximum(tmp10, tmp13)
tmp15 = tmp2 - tmp14
tmp16 = tl_math.exp(tmp15)
tmp17 = tmp5 - tmp14
tmp18 = tl_math.exp(tmp17)
tmp19 = tmp16 + tmp18
tmp20 = tmp9 - tmp14
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp23 = tmp13 - tmp14
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp26 = float('-inf')
tmp27 = tmp2 == tmp26
tmp28 = tmp27 == 0
tmp29 = tmp28.to(tl.int64)
tmp30 = tmp29 != 0
tmp31 = tmp5 == tmp26
tmp32 = tmp31 == 0
tmp33 = tmp32.to(tl.int64)
tmp34 = tmp33 != 0
tmp35 = tmp30 | tmp34
tmp36 = tmp9 == tmp26
tmp37 = tmp36 == 0
tmp38 = tmp37.to(tl.int64)
tmp39 = tmp38 != 0
tmp40 = tmp35 | tmp39
tmp41 = tmp13 == tmp26
tmp42 = tmp41 == 0
tmp43 = tmp42.to(tl.int64)
tmp44 = tmp43 != 0
tmp45 = tmp40 | tmp44
tl.store(out_ptr0 + x2, tmp14, xmask)
tl.store(out_ptr1 + x2, tmp25, xmask)
tl.store(out_ptr2 + x2, tmp45, xmask)
@triton.jit
def triton_poi_fused_2(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex // 4
x4 = xindex
x5 = xindex % 64
tmp0 = tl.load(in_ptr0 + x3, xmask, eviction_policy='evict_last').to(tl
.int1)
tmp2 = tl.load(in_out_ptr0 + x4, xmask)
tmp3 = tl.load(in_ptr1 + x5, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr2 + x3, xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr3 + x3, xmask, eviction_policy='evict_last')
tmp1 = tmp0 == 0
tmp4 = tmp2 + tmp3
tmp6 = tmp4 - tmp5
tmp7 = tl_math.exp(tmp6)
tmp9 = tmp7 / tmp8
tmp10 = 0.0
tmp11 = tl.where(tmp1, tmp10, tmp9)
tl.store(in_out_ptr0 + x4, tmp11, xmask)
@triton.jit
def triton_poi_fused_3(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel, YBLOCK:
tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 16 * y1), xmask & ymask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + y0, ymask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2 + 4 * y3), tmp2, xmask & ymask)
@triton.jit
def triton_poi_fused_clone_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 16 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_5(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp12 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + x0, tmp16, xmask)
tl.store(out_ptr1 + x0, tmp28, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_6(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-12
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + x2, tmp13, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12
) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_9, (4, 4), (4, 1))
assert_size_stride(primals_10, (4,), (1,))
assert_size_stride(primals_11, (4,), (1,))
assert_size_stride(primals_12, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf2)
del primals_6
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_0[grid(16, 4)](buf0, primals_2, buf3, 16, 4,
XBLOCK=2, YBLOCK=16, num_warps=1, num_stages=1)
del primals_2
buf4 = reinterpret_tensor(buf0, (4, 4, 1, 4), (16, 4, 4, 1), 0)
del buf0
triton_poi_fused_0[grid(16, 4)](buf1, primals_5, buf4, 16, 4,
XBLOCK=2, YBLOCK=16, num_warps=1, num_stages=1)
del primals_5
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = reinterpret_tensor(buf1, (4, 4, 4, 1), (16, 4, 1, 64), 0)
del buf1
buf7 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf8 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.bool)
triton_poi_fused_1[grid(64)](buf5, primals_8, buf6, buf7, buf8, 64,
XBLOCK=64, num_warps=1, num_stages=1)
buf9 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf5
triton_poi_fused_2[grid(256)](buf9, buf8, primals_8, buf6, buf7,
256, XBLOCK=128, num_warps=4, num_stages=1)
del buf8
del primals_8
buf10 = reinterpret_tensor(buf7, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf7
triton_poi_fused_3[grid(16, 4)](buf2, primals_7, buf10, 16, 4,
XBLOCK=2, YBLOCK=16, num_warps=1, num_stages=1)
del primals_7
buf11 = reinterpret_tensor(buf2, (16, 4, 1), (4, 1, 1), 0)
del buf2
extern_kernels.bmm(reinterpret_tensor(buf9, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf10, (16, 4, 1), (4, 1, 0), 0), out=buf11)
buf12 = reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf6
triton_poi_fused_clone_4[grid(16, 4)](buf11, buf12, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf13 = reinterpret_tensor(buf11, (16, 4), (4, 1), 0)
del buf11
extern_kernels.addmm(primals_10, reinterpret_tensor(buf12, (16, 4),
(4, 1), 0), reinterpret_tensor(primals_9, (4, 4), (1, 4), 0),
alpha=1, beta=1, out=buf13)
del primals_10
buf14 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf15 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
triton_poi_fused_add_native_layer_norm_5[grid(16)](buf13, primals_3,
buf14, buf15, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_add_native_layer_norm_6[grid(64)](buf13, primals_3,
buf14, buf15, primals_11, primals_12, buf16, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf14
del buf15
del primals_12
return buf16, primals_3, primals_11, buf9, reinterpret_tensor(buf10, (
16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf3, (16, 1, 4), (4,
1, 1), 0), reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0
), reinterpret_tensor(buf12, (16, 4), (4, 1), 0), buf13, primals_9
class BertSelfAttention(nn.Module):
def __init__(self, config):
super(BertSelfAttention, self).__init__()
if config.hidden_size % config.num_attention_heads != 0:
raise ValueError(
'The hidden size (%d) is not a multiple of the number of attention heads (%d)'
% (config.hidden_size, config.num_attention_heads))
self.num_attention_heads = config.num_attention_heads
self.attention_head_size = int(config.hidden_size / config.
num_attention_heads)
self.all_head_size = (self.num_attention_heads * self.
attention_head_size)
self.query = nn.Linear(config.hidden_size, self.all_head_size)
self.key = nn.Linear(config.hidden_size, self.all_head_size)
self.value = nn.Linear(config.hidden_size, self.all_head_size)
self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
def transpose_for_scores(self, x):
new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.
attention_head_size)
x = x.view(*new_x_shape)
return x.permute(0, 2, 1, 3)
def forward(self, hidden_states, attention_mask):
mixed_query_layer = self.query(hidden_states)
mixed_key_layer = self.key(hidden_states)
mixed_value_layer = self.value(hidden_states)
query_layer = self.transpose_for_scores(mixed_query_layer)
key_layer = self.transpose_for_scores(mixed_key_layer)
value_layer = self.transpose_for_scores(mixed_value_layer)
attention_scores = torch.matmul(query_layer, key_layer.transpose(-1,
-2))
attention_scores = attention_scores / math.sqrt(self.
attention_head_size)
attention_scores = attention_scores + attention_mask
attention_probs = nn.Softmax(dim=-1)(attention_scores)
attention_probs = self.dropout(attention_probs)
context_layer = torch.matmul(attention_probs, value_layer)
context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
new_context_layer_shape = context_layer.size()[:-2] + (self.
all_head_size,)
context_layer = context_layer.view(*new_context_layer_shape)
return context_layer
class BertSelfOutput(nn.Module):
def __init__(self, config):
super(BertSelfOutput, self).__init__()
self.dense = nn.Linear(config.hidden_size, config.hidden_size)
self.LayerNorm = LayerNorm(config.hidden_size, eps=1e-12)
self.dropout = nn.Dropout(config.hidden_dropout_prob)
def forward(self, hidden_states, input_tensor):
hidden_states = self.dense(hidden_states)
hidden_states = self.dropout(hidden_states)
hidden_states = self.LayerNorm(hidden_states + input_tensor)
return hidden_states
class BertAttentionNew(nn.Module):
def __init__(self, config):
super(BertAttentionNew, self).__init__()
self.self = BertSelfAttention(config)
self.output = BertSelfOutput(config)
def forward(self, input_0, input_1):
primals_1 = self.self.query.weight
primals_2 = self.self.query.bias
primals_4 = self.self.key.weight
primals_5 = self.self.key.bias
primals_6 = self.self.value.weight
primals_7 = self.self.value.bias
primals_9 = self.output.dense.weight
primals_10 = self.output.dense.bias
primals_11 = self.output.LayerNorm.weight
primals_12 = self.output.LayerNorm.bias
primals_3 = input_0
primals_8 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12])
return output[0]
|
UKPLab/MMT-Retrieval
|
BertAttention
| false | 14,531 |
[
"MIT"
] | 98 |
a31caaeb0da680131bf39dc855e38fdda949f38e
|
https://github.com/UKPLab/MMT-Retrieval/tree/a31caaeb0da680131bf39dc855e38fdda949f38e
|
EfficientBaseQuantization
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/bt/cbtldol4x6qs4oziek3iavggk7xw6fhop3y4abxugr2tnfcr64kv.py
# Topologically Sorted Source Nodes: [q_int_float, q_int, x, sub, neg, mul, x_1, result], Original ATen: [aten.div, aten.round, aten.minimum, aten.sub, aten.neg, aten.mul, aten.maximum]
# Source node to ATen node mapping:
# mul => mul
# neg => neg
# q_int => round_1
# q_int_float => div
# result => mul_1
# sub => sub
# x => minimum
# x_1 => maximum
# Graph fragment:
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, %arg1_1), kwargs = {})
# %round_1 : [num_users=1] = call_function[target=torch.ops.aten.round.default](args = (%div,), kwargs = {})
# %minimum : [num_users=1] = call_function[target=torch.ops.aten.minimum.default](args = (%round_1, %arg2_1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg2_1, 1.0), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%sub,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%neg, %arg3_1), kwargs = {})
# %maximum : [num_users=1] = call_function[target=torch.ops.aten.maximum.default](args = (%minimum, %mul), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, %maximum), kwargs = {})
triton_poi_fused_div_maximum_minimum_mul_neg_round_sub_0 = async_compile.triton('triton_poi_fused_div_maximum_minimum_mul_neg_round_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_maximum_minimum_mul_neg_round_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_maximum_minimum_mul_neg_round_sub_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask)
tmp4 = tl.load(in_ptr2 + (x0), xmask)
tmp9 = tl.load(in_ptr3 + (x0), xmask)
tmp2 = tmp1 / tmp0
tmp3 = libdevice.nearbyint(tmp2)
tmp5 = triton_helpers.minimum(tmp3, tmp4)
tmp6 = 1.0
tmp7 = tmp4 - tmp6
tmp8 = -tmp7
tmp10 = tmp8 * tmp9
tmp11 = triton_helpers.maximum(tmp5, tmp10)
tmp12 = tmp0 * tmp11
tl.store(out_ptr0 + (x0), tmp12, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg3_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [q_int_float, q_int, x, sub, neg, mul, x_1, result], Original ATen: [aten.div, aten.round, aten.minimum, aten.sub, aten.neg, aten.mul, aten.maximum]
stream0 = get_raw_stream(0)
triton_poi_fused_div_maximum_minimum_mul_neg_round_sub_0.run(arg1_1, arg0_1, arg2_1, arg3_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg3_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1, arg3_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import numpy as np
from torch import nn
class _EfficientBaseQuantizationFunction(torch.autograd.Function):
@staticmethod
def clip(x, min_value, max_value):
x = torch.min(x, max_value)
x = torch.max(x, min_value)
return x
@staticmethod
def forward(ctx, x, delta, q_p, is_signed):
q_int_float = x / delta
q_int = torch.round(q_int_float)
q_int_clip = _EfficientBaseQuantizationFunction.clip(q_int, -(q_p -
1.0) * is_signed, q_p)
result = delta * q_int_clip
ctx.save_for_backward(q_int_float, delta, q_p, is_signed)
return result
@staticmethod
def backward(ctx, grad_output):
q_int_float = ctx.saved_tensors[0]
delta = ctx.saved_tensors[1]
q_p = ctx.saved_tensors[2]
is_signed = ctx.saved_tensors[3]
q_p_shape_array = np.asarray(q_p.shape)
delta_shape_array = np.asarray(delta.shape)
index2sum_q_p = list(range(len(q_int_float.shape))) if np.prod(
q_p_shape_array) == 1 else list(np.where(q_p_shape_array == 1)[0])
index2sum_delta = list(range(len(q_int_float.shape))) if np.prod(
delta_shape_array) == 1 else list(np.where(delta_shape_array ==
1)[0])
q_int = torch.round(q_int_float)
upper_bound = q_p
lower_bound = -(q_p - 1.0) * is_signed
enable_low = (q_int < upper_bound).float()
enable_high = (q_int > lower_bound).float()
not_enable_low = 1 - enable_low
not_enable_high = 1 - enable_high
grad_output_x = grad_output * enable_low * enable_high
delta_mid = grad_output_x * (q_int - q_int_float)
delta_high = grad_output * not_enable_high * lower_bound
delta_low = grad_output * not_enable_low * upper_bound
grad_output_delta = (delta_mid + delta_high + delta_low).sum(dim=
index2sum_delta).reshape(delta.shape)
grad_output_c = delta * (grad_output * (not_enable_low - is_signed *
not_enable_high)).sum(dim=index2sum_q_p).reshape(q_p.shape)
grad_output_p = -(delta * (grad_output * (q_p - 1) * not_enable_high)
).sum(dim=index2sum_q_p)
if is_signed.shape[0] != 1:
grad_output_p = grad_output_p.reshape(is_signed.shape)
return grad_output_x, grad_output_delta, grad_output_c, grad_output_p
class EfficientBaseQuantization(nn.Module):
def __init__(self):
"""
Memory Efficient implementation of the base quantization module
"""
super(EfficientBaseQuantization, self).__init__()
def forward(self, x, delta, q_p, is_signed):
"""
The forward function of the quantization module
:param x: Input tensor to be quantized
:param delta: The quantization step size
:param q_p: The number of quantization step's
:param is_signed: is quantization signed
:return: A quantized tensor
"""
return _EfficientBaseQuantizationFunction.apply(x, delta, q_p,
is_signed)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
import numpy as np
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_div_maximum_minimum_mul_neg_round_sub_0(in_ptr0,
in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask)
tmp4 = tl.load(in_ptr2 + x0, xmask)
tmp9 = tl.load(in_ptr3 + x0, xmask)
tmp2 = tmp1 / tmp0
tmp3 = libdevice.nearbyint(tmp2)
tmp5 = triton_helpers.minimum(tmp3, tmp4)
tmp6 = 1.0
tmp7 = tmp4 - tmp6
tmp8 = -tmp7
tmp10 = tmp8 * tmp9
tmp11 = triton_helpers.maximum(tmp5, tmp10)
tmp12 = tmp0 * tmp11
tl.store(out_ptr0 + x0, tmp12, xmask)
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg3_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_div_maximum_minimum_mul_neg_round_sub_0[grid(256)](
arg1_1, arg0_1, arg2_1, arg3_1, buf0, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
return buf0,
class _EfficientBaseQuantizationFunction(torch.autograd.Function):
@staticmethod
def clip(x, min_value, max_value):
x = torch.min(x, max_value)
x = torch.max(x, min_value)
return x
@staticmethod
def forward(ctx, x, delta, q_p, is_signed):
q_int_float = x / delta
q_int = torch.round(q_int_float)
q_int_clip = _EfficientBaseQuantizationFunction.clip(q_int, -(q_p -
1.0) * is_signed, q_p)
result = delta * q_int_clip
ctx.save_for_backward(q_int_float, delta, q_p, is_signed)
return result
@staticmethod
def backward(ctx, grad_output):
q_int_float = ctx.saved_tensors[0]
delta = ctx.saved_tensors[1]
q_p = ctx.saved_tensors[2]
is_signed = ctx.saved_tensors[3]
q_p_shape_array = np.asarray(q_p.shape)
delta_shape_array = np.asarray(delta.shape)
index2sum_q_p = list(range(len(q_int_float.shape))) if np.prod(
q_p_shape_array) == 1 else list(np.where(q_p_shape_array == 1)[0])
index2sum_delta = list(range(len(q_int_float.shape))) if np.prod(
delta_shape_array) == 1 else list(np.where(delta_shape_array ==
1)[0])
q_int = torch.round(q_int_float)
upper_bound = q_p
lower_bound = -(q_p - 1.0) * is_signed
enable_low = (q_int < upper_bound).float()
enable_high = (q_int > lower_bound).float()
not_enable_low = 1 - enable_low
not_enable_high = 1 - enable_high
grad_output_x = grad_output * enable_low * enable_high
delta_mid = grad_output_x * (q_int - q_int_float)
delta_high = grad_output * not_enable_high * lower_bound
delta_low = grad_output * not_enable_low * upper_bound
grad_output_delta = (delta_mid + delta_high + delta_low).sum(dim=
index2sum_delta).reshape(delta.shape)
grad_output_c = delta * (grad_output * (not_enable_low - is_signed *
not_enable_high)).sum(dim=index2sum_q_p).reshape(q_p.shape)
grad_output_p = -(delta * (grad_output * (q_p - 1) * not_enable_high)
).sum(dim=index2sum_q_p)
if is_signed.shape[0] != 1:
grad_output_p = grad_output_p.reshape(is_signed.shape)
return grad_output_x, grad_output_delta, grad_output_c, grad_output_p
class EfficientBaseQuantizationNew(nn.Module):
def __init__(self):
"""
Memory Efficient implementation of the base quantization module
"""
super(EfficientBaseQuantizationNew, self).__init__()
def forward(self, input_0, input_1, input_2, input_3):
arg0_1 = input_0
arg1_1 = input_1
arg2_1 = input_2
arg3_1 = input_3
output = call([arg0_1, arg1_1, arg2_1, arg3_1])
return output[0]
|
UniSerj/ai-research
|
EfficientBaseQuantization
| false | 14,532 |
[
"Apache-2.0"
] | 46 |
79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
https://github.com/UniSerj/ai-research/tree/79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
ScaledLeakyReLUSin
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/gy/cgyxz5pkmk3r4rp7o7dmnya5fjqpalaiuzo5khwo5j6ahsfthfga.py
# Topologically Sorted Source Nodes: [out, mul], Original ATen: [aten.cat, aten.mul]
# Source node to ATen node mapping:
# mul => mul_1
# out => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%where, %sin], 1), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%cat, 1.4142135623730951), kwargs = {})
triton_poi_fused_cat_mul_0 = async_compile.triton('triton_poi_fused_cat_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 16) % 4
x0 = xindex % 16
x2 = (xindex // 64)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 2, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (32*x1) + (64*x2)), tmp4 & xmask, other=0.0)
tmp6 = 0.0
tmp7 = tmp5 > tmp6
tmp8 = 0.2
tmp9 = tmp5 * tmp8
tmp10 = tl.where(tmp7, tmp5, tmp9)
tmp11 = tl.full(tmp10.shape, 0.0, tmp10.dtype)
tmp12 = tl.where(tmp4, tmp10, tmp11)
tmp13 = tmp0 >= tmp3
tmp14 = tl.full([1], 4, tl.int64)
tmp15 = tmp0 < tmp14
tmp16 = tl.load(in_ptr0 + (16 + x0 + (32*((-2) + x1)) + (64*x2)), tmp13 & xmask, other=0.0)
tmp17 = tl_math.sin(tmp16)
tmp18 = tl.full(tmp17.shape, 0.0, tmp17.dtype)
tmp19 = tl.where(tmp13, tmp17, tmp18)
tmp20 = tl.where(tmp4, tmp12, tmp19)
tmp21 = 1.4142135623730951
tmp22 = tmp20 * tmp21
tl.store(out_ptr0 + (x3), tmp22, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [out, mul], Original ATen: [aten.cat, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_mul_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import math
import torch
from torch import nn
import torch.nn.functional as F
class ScaledLeakyReLUSin(nn.Module):
def __init__(self, negative_slope=0.2):
super().__init__()
self.negative_slope = negative_slope
def forward(self, input):
out_lr = F.leaky_relu(input[:, ::2], negative_slope=self.negative_slope
)
out_sin = torch.sin(input[:, 1::2])
out = torch.cat([out_lr, out_sin], 1)
return out * math.sqrt(2)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_cat_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 16 % 4
x0 = xindex % 16
x2 = xindex // 64
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 2, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 32 * x1 + 64 * x2), tmp4 & xmask, other=0.0)
tmp6 = 0.0
tmp7 = tmp5 > tmp6
tmp8 = 0.2
tmp9 = tmp5 * tmp8
tmp10 = tl.where(tmp7, tmp5, tmp9)
tmp11 = tl.full(tmp10.shape, 0.0, tmp10.dtype)
tmp12 = tl.where(tmp4, tmp10, tmp11)
tmp13 = tmp0 >= tmp3
tl.full([1], 4, tl.int64)
tmp16 = tl.load(in_ptr0 + (16 + x0 + 32 * (-2 + x1) + 64 * x2), tmp13 &
xmask, other=0.0)
tmp17 = tl_math.sin(tmp16)
tmp18 = tl.full(tmp17.shape, 0.0, tmp17.dtype)
tmp19 = tl.where(tmp13, tmp17, tmp18)
tmp20 = tl.where(tmp4, tmp12, tmp19)
tmp21 = 1.4142135623730951
tmp22 = tmp20 * tmp21
tl.store(out_ptr0 + x3, tmp22, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_mul_0[grid(256)](arg0_1, buf0, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg0_1
return buf0,
class ScaledLeakyReLUSinNew(nn.Module):
def __init__(self, negative_slope=0.2):
super().__init__()
self.negative_slope = negative_slope
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
|
Ugness/CIPS_SR
|
ScaledLeakyReLUSin
| false | 14,533 |
[
"MIT"
] | 172 |
abce872f5bc1b84afb9634a7dd1991e8c74d7616
|
https://github.com/Ugness/CIPS_SR/tree/abce872f5bc1b84afb9634a7dd1991e8c74d7616
|
SReLU
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/ve/cvekibnibp2jzomh3kuv73u2liqkd3ap4mdsxspqzvopkso2ubyb.py
# Topologically Sorted Source Nodes: [ge, float_1, add, mul, add_1, mul_1, lt, float_2, gt, float_3, mul_2, mul_3, add_2, le, float_4, add_3, mul_4, add_4, mul_5, add_5], Original ATen: [aten.ge, aten._to_copy, aten.add, aten.mul, aten.lt, aten.gt, aten.le]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# add_2 => add_2
# add_3 => add_3
# add_4 => add_4
# add_5 => add_5
# float_1 => convert_element_type
# float_2 => convert_element_type_1
# float_3 => convert_element_type_2
# float_4 => convert_element_type_3
# ge => ge
# gt => gt
# le => le
# lt => lt
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# mul_3 => mul_3
# mul_4 => mul_4
# mul_5 => mul_5
# Graph fragment:
# %ge : [num_users=2] = call_function[target=torch.ops.aten.ge.Tensor](args = (%primals_2, %primals_1), kwargs = {})
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%ge, torch.float32), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %primals_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_3, %add), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_1, %mul), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%convert_element_type, %add_1), kwargs = {})
# %lt : [num_users=1] = call_function[target=torch.ops.aten.lt.Tensor](args = (%primals_2, %primals_1), kwargs = {})
# %convert_element_type_1 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%lt, torch.float32), kwargs = {})
# %gt : [num_users=1] = call_function[target=torch.ops.aten.gt.Tensor](args = (%primals_2, %primals_4), kwargs = {})
# %convert_element_type_2 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%gt, torch.float32), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%convert_element_type_1, %convert_element_type_2), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, %primals_2), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %mul_3), kwargs = {})
# %le : [num_users=2] = call_function[target=torch.ops.aten.le.Tensor](args = (%primals_2, %primals_4), kwargs = {})
# %convert_element_type_3 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%le, torch.float32), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %primals_4), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_5, %add_3), kwargs = {})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_4, %mul_4), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%convert_element_type_3, %add_4), kwargs = {})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_2, %mul_5), kwargs = {})
triton_poi_fused__to_copy_add_ge_gt_le_lt_mul_0 = async_compile.triton('triton_poi_fused__to_copy_add_ge_gt_le_lt_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*i1', 6: '*i1', 7: '*fp32', 8: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_add_ge_gt_le_lt_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_add_ge_gt_le_lt_mul_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 >= tmp1
tmp4 = tmp0 <= tmp3
tmp5 = tmp2.to(tl.float32)
tmp7 = tmp0 + tmp1
tmp8 = tmp6 * tmp7
tmp9 = tmp1 + tmp8
tmp10 = tmp5 * tmp9
tmp11 = tmp0 < tmp1
tmp12 = tmp11.to(tl.float32)
tmp13 = tmp0 > tmp3
tmp14 = tmp13.to(tl.float32)
tmp15 = tmp12 * tmp14
tmp16 = tmp15 * tmp0
tmp17 = tmp10 + tmp16
tmp18 = tmp4.to(tl.float32)
tmp20 = tmp0 + tmp3
tmp21 = tmp19 * tmp20
tmp22 = tmp3 + tmp21
tmp23 = tmp18 * tmp22
tmp24 = tmp17 + tmp23
tl.store(out_ptr0 + (x2), tmp2, xmask)
tl.store(out_ptr1 + (x2), tmp4, xmask)
tl.store(out_ptr2 + (x2), tmp24, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, ), (1, ))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, ), (1, ))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [ge, float_1, add, mul, add_1, mul_1, lt, float_2, gt, float_3, mul_2, mul_3, add_2, le, float_4, add_3, mul_4, add_4, mul_5, add_5], Original ATen: [aten.ge, aten._to_copy, aten.add, aten.mul, aten.lt, aten.gt, aten.le]
stream0 = get_raw_stream(0)
triton_poi_fused__to_copy_add_ge_gt_le_lt_mul_0.run(primals_2, primals_1, primals_4, primals_3, primals_5, buf0, buf1, buf2, 256, grid=grid(256), stream=stream0)
return (buf2, primals_1, primals_2, primals_3, primals_4, primals_5, buf0, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
from torch.nn.parameter import Parameter
class SReLU(nn.Module):
"""
SReLU (S-shaped Rectified Linear Activation Unit): a combination of three linear functions, which perform mapping R → R with the following formulation:
.. math::
h(x_i) = \\left\\{\\begin{matrix} t_i^r + a_i^r(x_i - t_i^r), x_i \\geq t_i^r \\\\ x_i, t_i^r > x_i > t_i^l\\\\ t_i^l + a_i^l(x_i - t_i^l), x_i \\leq t_i^l \\\\ \\end{matrix}\\right.
with 4 trainable parameters.
Shape:
- Input: (N, *) where * means, any number of additional
dimensions
- Output: (N, *), same shape as the input
Parameters:
.. math:: \\{t_i^r, a_i^r, t_i^l, a_i^l\\}
4 trainable parameters, which model an individual SReLU activation unit. The subscript i indicates that we allow SReLU to vary in different channels. Parameters can be initialized manually or randomly.
References:
- See SReLU paper:
https://arxiv.org/pdf/1512.07030.pdf
Examples:
>>> srelu_activation = srelu((2,2))
>>> t = torch.randn((2,2), dtype=torch.float, requires_grad = True)
>>> output = srelu_activation(t)
"""
def __init__(self, in_features, parameters=None):
"""
Initialization.
INPUT:
- in_features: shape of the input
- parameters: (tr, tl, ar, al) parameters for manual initialization, default value is None. If None is passed, parameters are initialized randomly.
"""
super(SReLU, self).__init__()
self.in_features = in_features
if parameters is None:
self.tr = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
self.tl = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
self.ar = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
self.al = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
else:
self.tr, self.tl, self.ar, self.al = parameters
def forward(self, x):
"""
Forward pass of the function
"""
return (x >= self.tr).float() * (self.tr + self.ar * (x + self.tr)) + (
x < self.tr).float() * (x > self.tl).float() * x + (x <= self.tl
).float() * (self.tl + self.al * (x + self.tl))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_features': 4}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused__to_copy_add_ge_gt_le_lt_mul_0(in_ptr0, in_ptr1,
in_ptr2, in_ptr3, in_ptr4, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x0, xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 >= tmp1
tmp4 = tmp0 <= tmp3
tmp5 = tmp2.to(tl.float32)
tmp7 = tmp0 + tmp1
tmp8 = tmp6 * tmp7
tmp9 = tmp1 + tmp8
tmp10 = tmp5 * tmp9
tmp11 = tmp0 < tmp1
tmp12 = tmp11.to(tl.float32)
tmp13 = tmp0 > tmp3
tmp14 = tmp13.to(tl.float32)
tmp15 = tmp12 * tmp14
tmp16 = tmp15 * tmp0
tmp17 = tmp10 + tmp16
tmp18 = tmp4.to(tl.float32)
tmp20 = tmp0 + tmp3
tmp21 = tmp19 * tmp20
tmp22 = tmp3 + tmp21
tmp23 = tmp18 * tmp22
tmp24 = tmp17 + tmp23
tl.store(out_ptr0 + x2, tmp2, xmask)
tl.store(out_ptr1 + x2, tmp4, xmask)
tl.store(out_ptr2 + x2, tmp24, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4,), (1,))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4,), (1,))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__to_copy_add_ge_gt_le_lt_mul_0[grid(256)](primals_2,
primals_1, primals_4, primals_3, primals_5, buf0, buf1, buf2,
256, XBLOCK=128, num_warps=4, num_stages=1)
return (buf2, primals_1, primals_2, primals_3, primals_4, primals_5,
buf0, buf1)
class SReLUNew(nn.Module):
"""
SReLU (S-shaped Rectified Linear Activation Unit): a combination of three linear functions, which perform mapping R → R with the following formulation:
.. math::
h(x_i) = \\left\\{\\begin{matrix} t_i^r + a_i^r(x_i - t_i^r), x_i \\geq t_i^r \\\\ x_i, t_i^r > x_i > t_i^l\\\\ t_i^l + a_i^l(x_i - t_i^l), x_i \\leq t_i^l \\\\ \\end{matrix}\\right.
with 4 trainable parameters.
Shape:
- Input: (N, *) where * means, any number of additional
dimensions
- Output: (N, *), same shape as the input
Parameters:
.. math:: \\{t_i^r, a_i^r, t_i^l, a_i^l\\}
4 trainable parameters, which model an individual SReLU activation unit. The subscript i indicates that we allow SReLU to vary in different channels. Parameters can be initialized manually or randomly.
References:
- See SReLU paper:
https://arxiv.org/pdf/1512.07030.pdf
Examples:
>>> srelu_activation = srelu((2,2))
>>> t = torch.randn((2,2), dtype=torch.float, requires_grad = True)
>>> output = srelu_activation(t)
"""
def __init__(self, in_features, parameters=None):
"""
Initialization.
INPUT:
- in_features: shape of the input
- parameters: (tr, tl, ar, al) parameters for manual initialization, default value is None. If None is passed, parameters are initialized randomly.
"""
super(SReLUNew, self).__init__()
self.in_features = in_features
if parameters is None:
self.tr = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
self.tl = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
self.ar = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
self.al = Parameter(torch.randn(in_features, dtype=torch.float,
requires_grad=True))
else:
self.tr, self.tl, self.ar, self.al = parameters
def forward(self, input_0):
primals_1 = self.tr
primals_3 = self.tl
primals_4 = self.ar
primals_5 = self.al
primals_2 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
|
VITA-Group/SViTE
|
SReLU
| false | 14,534 |
[
"MIT"
] | 50 |
b0c62fd153c8b0b99917ab935ee76925c9de1149
|
https://github.com/VITA-Group/SViTE/tree/b0c62fd153c8b0b99917ab935ee76925c9de1149
|
MultiHeadAttention
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/kq/ckqzjwsdojunslaxexvivgmdizmhm6czh2xpzemjckiggsrasgr3.py
# Topologically Sorted Source Nodes: [matmul_qk], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul_qk => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (64*y1)), xmask & ymask)
tl.store(out_ptr0 + (x2 + (16*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ss/csssqqlybvm2dlovzxhxorvbhyaj46oqebdryly2s5obgs7rshwq.py
# Topologically Sorted Source Nodes: [mul, scaled_attention_logits_1, attention_weights], Original ATen: [aten.mul, aten.add, aten._softmax]
# Source node to ATen node mapping:
# attention_weights => amax, div_1, exp, sub, sum_1
# mul => mul
# scaled_attention_logits_1 => add
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_4, -1000000000.0), kwargs = {})
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_5, %mul), kwargs = {})
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%add, [-1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_per_fused__softmax_add_mul_1 = async_compile.triton('triton_per_fused__softmax_add_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[256, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__softmax_add_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 2, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__softmax_add_mul_1(in_ptr0, in_ptr1, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 256
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (16*x0)), xmask, other=0.0)
tmp1 = tl.load(in_ptr1 + (r1 + (16*x0)), xmask, other=0.0)
tmp2 = -1000000000.0
tmp3 = tmp1 * tmp2
tmp4 = tmp0 + tmp3
tmp5 = tl.broadcast_to(tmp4, [XBLOCK, RBLOCK])
tmp7 = tl.where(xmask, tmp5, float("-inf"))
tmp8 = triton_helpers.max2(tmp7, 1)[:, None]
tmp9 = tmp4 - tmp8
tmp10 = tl_math.exp(tmp9)
tmp11 = tl.broadcast_to(tmp10, [XBLOCK, RBLOCK])
tmp13 = tl.where(xmask, tmp11, 0)
tmp14 = tl.sum(tmp13, 1)[:, None]
tmp15 = tmp10 / tmp14
tl.store(out_ptr2 + (r1 + (16*x0)), tmp15, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/mz/cmzlu2lip25blpsdqeby7ek5757op6xw3pdkxbdediou5szw32tx.py
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# output_1 => clone_3
# Graph fragment:
# %clone_3 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%view_9,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_2 = async_compile.triton('triton_poi_fused_clone_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 16
y1 = (yindex // 16)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (16*x2) + (64*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ok/cokamvfj3z4xuz3jmalftfns3huimimr3c4gzm52vaybmdliglu4.py
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# output_1 => add_1
# Graph fragment:
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_11, %primals_6), kwargs = {})
triton_poi_fused_add_3 = async_compile.triton('triton_poi_fused_add_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 16, 16), (1024, 256, 16, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 16, 1), (64, 16, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_qk], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(primals_1, buf0, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 1, 16), (64, 16, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_qk], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(primals_2, buf1, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((16, 16, 16), (256, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_qk], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (16, 16, 1), (16, 1, 0), 0), reinterpret_tensor(buf1, (16, 1, 16), (16, 0, 1), 0), out=buf2)
buf5 = empty_strided_cuda((4, 4, 16, 16), (1024, 256, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, scaled_attention_logits_1, attention_weights], Original ATen: [aten.mul, aten.add, aten._softmax]
triton_per_fused__softmax_add_mul_1.run(buf2, primals_4, buf5, 256, 16, grid=grid(256), stream=stream0)
del buf2
del primals_4
buf6 = reinterpret_tensor(buf1, (4, 4, 16, 1), (64, 16, 1, 1), 0); del buf1 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(primals_3, buf6, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_3
buf7 = reinterpret_tensor(buf0, (16, 16, 1), (16, 1, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf5, (16, 16, 16), (256, 16, 1), 0), reinterpret_tensor(buf6, (16, 16, 1), (16, 1, 0), 0), out=buf7)
buf8 = reinterpret_tensor(buf6, (4, 16, 4), (64, 4, 1), 0); del buf6 # reuse
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.clone]
triton_poi_fused_clone_2.run(buf7, buf8, 64, 4, grid=grid(64, 4), stream=stream0)
buf9 = reinterpret_tensor(buf7, (64, 4), (4, 1), 0); del buf7 # reuse
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf8, (64, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf9)
del primals_5
buf10 = reinterpret_tensor(buf9, (4, 16, 4), (64, 4, 1), 0); del buf9 # reuse
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.add]
triton_poi_fused_add_3.run(buf10, primals_6, 256, grid=grid(256), stream=stream0)
del primals_6
return (buf10, buf5, reinterpret_tensor(buf8, (64, 4), (4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 16, 16), (1024, 256, 16, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
def scaled_dot_product_attention(q, k, v, mask=None):
"""Calculate the attention weights.
q, k, v must have matching leading dimensions.
k, v must have matching penultimate dimension, i.e.: seq_len_k = seq_len_v.
The mask has different shapes depending on its type(padding or look ahead)
but it must be broadcastable for addition.
Args:
q: query shape == (..., seq_len_q, depth)
k: key shape == (..., seq_len_k, depth)
v: value shape == (..., seq_len_v, depth_v)
mask: Float tensor with shape broadcastable
to (..., seq_len_q, seq_len_k). Defaults to None.
Returns:
output, attention_weights
"""
matmul_qk = torch.matmul(q, k.permute(0, 1, 3, 2))
None
dk = torch.tensor(k.size()[-1], dtype=torch.float32)
scaled_attention_logits = matmul_qk / torch.sqrt(dk)
if mask is not None:
scaled_attention_logits += mask * -1000000000.0
softmax = nn.Softmax(dim=-1)
attention_weights = softmax(scaled_attention_logits)
output = torch.matmul(attention_weights, v)
return output, attention_weights
class MultiHeadAttention(nn.Module):
def __init__(self, input_feature_size, d_model, num_heads):
super(MultiHeadAttention, self).__init__()
self.num_heads = num_heads
self.d_model = d_model
assert d_model % self.num_heads == 0
self.depth = d_model // self.num_heads
self.dense = nn.Linear(input_feature_size, d_model)
def split_heads(self, x, batch_size):
"""Split the last dimension into (num_heads, depth).
Transpose the result such that the shape is (batch_size, num_heads, seq_len, depth)
"""
x = torch.reshape(x, (batch_size, -1, self.num_heads, self.depth))
return x.permute(0, 2, 1, 3)
def forward(self, v, k, q, mask, training=True):
batch_size = q.size()[0]
if q.size()[-2] > 10:
pass
else:
pass
q = q
k = k
v = v
q = self.split_heads(q, batch_size)
k = self.split_heads(k, batch_size)
v = self.split_heads(v, batch_size)
scaled_attention, attention_weights = scaled_dot_product_attention(q,
k, v, mask)
scaled_attention = scaled_attention.permute(0, 2, 1, 3)
concat_attention = torch.reshape(scaled_attention, (batch_size, -1,
self.d_model))
output = self.dense(concat_attention)
return output, attention_weights
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4]), torch.rand([4, 4, 16, 16])]
def get_init_inputs():
return [[], {'input_feature_size': 4, 'd_model': 4, 'num_heads': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 64 * y1), xmask & ymask)
tl.store(out_ptr0 + (x2 + 16 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_per_fused__softmax_add_mul_1(in_ptr0, in_ptr1, out_ptr2, xnumel,
rnumel, XBLOCK: tl.constexpr):
xnumel = 256
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 16 * x0), xmask, other=0.0)
tmp1 = tl.load(in_ptr1 + (r1 + 16 * x0), xmask, other=0.0)
tmp2 = -1000000000.0
tmp3 = tmp1 * tmp2
tmp4 = tmp0 + tmp3
tmp5 = tl.broadcast_to(tmp4, [XBLOCK, RBLOCK])
tmp7 = tl.where(xmask, tmp5, float('-inf'))
tmp8 = triton_helpers.max2(tmp7, 1)[:, None]
tmp9 = tmp4 - tmp8
tmp10 = tl_math.exp(tmp9)
tmp11 = tl.broadcast_to(tmp10, [XBLOCK, RBLOCK])
tmp13 = tl.where(xmask, tmp11, 0)
tmp14 = tl.sum(tmp13, 1)[:, None]
tmp15 = tmp10 / tmp14
tl.store(out_ptr2 + (r1 + 16 * x0), tmp15, xmask)
@triton.jit
def triton_poi_fused_clone_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 16
y1 = yindex // 16
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 16 * x2 + 64 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4, 16, 16), (1024, 256, 16, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 16, 1), (64, 16, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(16, 16)](primals_1, buf0, 16, 16,
XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
del primals_1
buf1 = empty_strided_cuda((4, 4, 1, 16), (64, 16, 16, 1), torch.float32
)
triton_poi_fused_clone_0[grid(16, 16)](primals_2, buf1, 16, 16,
XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((16, 16, 16), (256, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (16, 16, 1), (16, 1, 0),
0), reinterpret_tensor(buf1, (16, 1, 16), (16, 0, 1), 0), out=buf2)
buf5 = empty_strided_cuda((4, 4, 16, 16), (1024, 256, 16, 1), torch
.float32)
triton_per_fused__softmax_add_mul_1[grid(256)](buf2, primals_4,
buf5, 256, 16, XBLOCK=128, num_warps=8, num_stages=1)
del buf2
del primals_4
buf6 = reinterpret_tensor(buf1, (4, 4, 16, 1), (64, 16, 1, 1), 0)
del buf1
triton_poi_fused_clone_0[grid(16, 16)](primals_3, buf6, 16, 16,
XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
del primals_3
buf7 = reinterpret_tensor(buf0, (16, 16, 1), (16, 1, 1), 0)
del buf0
extern_kernels.bmm(reinterpret_tensor(buf5, (16, 16, 16), (256, 16,
1), 0), reinterpret_tensor(buf6, (16, 16, 1), (16, 1, 0), 0),
out=buf7)
buf8 = reinterpret_tensor(buf6, (4, 16, 4), (64, 4, 1), 0)
del buf6
triton_poi_fused_clone_2[grid(64, 4)](buf7, buf8, 64, 4, XBLOCK=4,
YBLOCK=32, num_warps=4, num_stages=1)
buf9 = reinterpret_tensor(buf7, (64, 4), (4, 1), 0)
del buf7
extern_kernels.mm(reinterpret_tensor(buf8, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf9)
del primals_5
buf10 = reinterpret_tensor(buf9, (4, 16, 4), (64, 4, 1), 0)
del buf9
triton_poi_fused_add_3[grid(256)](buf10, primals_6, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del primals_6
return buf10, buf5, reinterpret_tensor(buf8, (64, 4), (4, 1), 0)
def scaled_dot_product_attention(q, k, v, mask=None):
"""Calculate the attention weights.
q, k, v must have matching leading dimensions.
k, v must have matching penultimate dimension, i.e.: seq_len_k = seq_len_v.
The mask has different shapes depending on its type(padding or look ahead)
but it must be broadcastable for addition.
Args:
q: query shape == (..., seq_len_q, depth)
k: key shape == (..., seq_len_k, depth)
v: value shape == (..., seq_len_v, depth_v)
mask: Float tensor with shape broadcastable
to (..., seq_len_q, seq_len_k). Defaults to None.
Returns:
output, attention_weights
"""
matmul_qk = torch.matmul(q, k.permute(0, 1, 3, 2))
None
dk = torch.tensor(k.size()[-1], dtype=torch.float32)
scaled_attention_logits = matmul_qk / torch.sqrt(dk)
if mask is not None:
scaled_attention_logits += mask * -1000000000.0
softmax = nn.Softmax(dim=-1)
attention_weights = softmax(scaled_attention_logits)
output = torch.matmul(attention_weights, v)
return output, attention_weights
class MultiHeadAttentionNew(nn.Module):
def __init__(self, input_feature_size, d_model, num_heads):
super(MultiHeadAttentionNew, self).__init__()
self.num_heads = num_heads
self.d_model = d_model
assert d_model % self.num_heads == 0
self.depth = d_model // self.num_heads
self.dense = nn.Linear(input_feature_size, d_model)
def split_heads(self, x, batch_size):
"""Split the last dimension into (num_heads, depth).
Transpose the result such that the shape is (batch_size, num_heads, seq_len, depth)
"""
x = torch.reshape(x, (batch_size, -1, self.num_heads, self.depth))
return x.permute(0, 2, 1, 3)
def forward(self, input_0, input_1, input_2, input_3):
primals_5 = self.dense.weight
primals_6 = self.dense.bias
primals_1 = input_0
primals_2 = input_1
primals_3 = input_2
primals_4 = input_3
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6])
return output[0], output[1]
|
ULTR-Community/ULTRA_Pytorch
|
MultiHeadAttention
| false | 14,535 |
[
"Apache-2.0"
] | 46 |
ec4fe329e4239b588a940cb4bcdd6a321aade679
|
https://github.com/ULTR-Community/ULTRA_Pytorch/tree/ec4fe329e4239b588a940cb4bcdd6a321aade679
|
BaseQuantization
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/oy/coygcgfvjfikzpokuld4c67gmbuy2xthv6c4bfkuetfawc3yk4bp.py
# Topologically Sorted Source Nodes: [truediv, round_1, x_error, add, x, sub, symmetirc_q_p, neg, x_1, mul_1], Original ATen: [aten.div, aten.round, aten.sub, aten.add, aten.minimum, aten.mul, aten.neg, aten.maximum]
# Source node to ATen node mapping:
# add => add
# mul_1 => mul_1
# neg => neg
# round_1 => round_1
# sub => sub
# symmetirc_q_p => mul
# truediv => div
# x => minimum
# x_1 => maximum
# x_error => sub_1
# Graph fragment:
# %div : [num_users=3] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg2_1, %arg3_1), kwargs = {})
# %round_1 : [num_users=1] = call_function[target=torch.ops.aten.round.default](args = (%div,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%round_1, %div), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%div, %sub_1), kwargs = {})
# %minimum : [num_users=1] = call_function[target=torch.ops.aten.minimum.default](args = (%add, %arg0_1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, 1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %arg1_1), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%mul,), kwargs = {})
# %maximum : [num_users=1] = call_function[target=torch.ops.aten.maximum.default](args = (%minimum, %neg), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg3_1, %maximum), kwargs = {})
triton_poi_fused_add_div_maximum_minimum_mul_neg_round_sub_0 = async_compile.triton('triton_poi_fused_add_div_maximum_minimum_mul_neg_round_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_maximum_minimum_mul_neg_round_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_maximum_minimum_mul_neg_round_sub_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask)
tmp6 = tl.load(in_ptr2 + (x0), xmask)
tmp10 = tl.load(in_ptr3 + (x0), xmask)
tmp2 = tmp1 / tmp0
tmp3 = libdevice.nearbyint(tmp2)
tmp4 = tmp3 - tmp2
tmp5 = tmp2 + tmp4
tmp7 = triton_helpers.minimum(tmp5, tmp6)
tmp8 = 1.0
tmp9 = tmp6 - tmp8
tmp11 = tmp9 * tmp10
tmp12 = -tmp11
tmp13 = triton_helpers.maximum(tmp7, tmp12)
tmp14 = tmp0 * tmp13
tl.store(out_ptr0 + (x0), tmp14, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg3_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [truediv, round_1, x_error, add, x, sub, symmetirc_q_p, neg, x_1, mul_1], Original ATen: [aten.div, aten.round, aten.sub, aten.add, aten.minimum, aten.mul, aten.neg, aten.maximum]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_maximum_minimum_mul_neg_round_sub_0.run(arg3_1, arg2_1, arg0_1, arg1_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg3_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1, arg3_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class Clipping(nn.Module):
def __init__(self):
"""
This module perform element-wise clipping.
"""
super(Clipping, self).__init__()
def forward(self, x, max_value, min_value):
"""
The forward function of the clipping module
:param x: Input tensor to be clipped
:param max_value: The maximal value of the tensor after clipping
:param min_value: The minimal value of the tensor after clipping
:return: A clipped tensor
"""
x = torch.min(x, max_value)
x = torch.max(x, min_value)
return x
class RoundSTE(nn.Module):
def __init__(self):
"""
This module perform element-wise rounding with straight through estimator (STE).
"""
super(RoundSTE, self).__init__()
def forward(self, x):
"""
The forward function of the rounding module
:param x: Input tensor to be rounded
:return: A rounded tensor
"""
x_error = torch.round(x) - x
return x + x_error.detach()
class BaseQuantization(nn.Module):
def __init__(self):
"""
This module perform element-wise quantization.
"""
super(BaseQuantization, self).__init__()
self.round = RoundSTE()
self.clip = Clipping()
self.symmetric = False
def forward(self, x, delta, q_p, is_signed):
"""
The forward function of the quantization module
:param x: Input tensor to be quantized
:param delta: The quantization step size
:param q_p: The number of quantization step's
:param is_signed: is quantization signed
:return: A quantized tensor
"""
symmetirc_q_p = (q_p - int(not self.symmetric)) * is_signed
return delta * self.clip(self.round(x / delta), q_p, -symmetirc_q_p)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_maximum_minimum_mul_neg_round_sub_0(in_ptr0,
in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask)
tmp6 = tl.load(in_ptr2 + x0, xmask)
tmp10 = tl.load(in_ptr3 + x0, xmask)
tmp2 = tmp1 / tmp0
tmp3 = libdevice.nearbyint(tmp2)
tmp4 = tmp3 - tmp2
tmp5 = tmp2 + tmp4
tmp7 = triton_helpers.minimum(tmp5, tmp6)
tmp8 = 1.0
tmp9 = tmp6 - tmp8
tmp11 = tmp9 * tmp10
tmp12 = -tmp11
tmp13 = triton_helpers.maximum(tmp7, tmp12)
tmp14 = tmp0 * tmp13
tl.store(out_ptr0 + x0, tmp14, xmask)
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg3_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_maximum_minimum_mul_neg_round_sub_0[grid(256)
](arg3_1, arg2_1, arg0_1, arg1_1, buf0, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
return buf0,
class Clipping(nn.Module):
def __init__(self):
"""
This module perform element-wise clipping.
"""
super(Clipping, self).__init__()
def forward(self, x, max_value, min_value):
"""
The forward function of the clipping module
:param x: Input tensor to be clipped
:param max_value: The maximal value of the tensor after clipping
:param min_value: The minimal value of the tensor after clipping
:return: A clipped tensor
"""
x = torch.min(x, max_value)
x = torch.max(x, min_value)
return x
class RoundSTE(nn.Module):
def __init__(self):
"""
This module perform element-wise rounding with straight through estimator (STE).
"""
super(RoundSTE, self).__init__()
def forward(self, x):
"""
The forward function of the rounding module
:param x: Input tensor to be rounded
:return: A rounded tensor
"""
x_error = torch.round(x) - x
return x + x_error.detach()
class BaseQuantizationNew(nn.Module):
def __init__(self):
"""
This module perform element-wise quantization.
"""
super(BaseQuantizationNew, self).__init__()
self.round = RoundSTE()
self.clip = Clipping()
self.symmetric = False
def forward(self, input_0, input_1, input_2, input_3):
arg0_1 = input_0
arg1_1 = input_1
arg2_1 = input_2
arg3_1 = input_3
output = call([arg0_1, arg1_1, arg2_1, arg3_1])
return output[0]
|
UniSerj/ai-research
|
BaseQuantization
| false | 14,536 |
[
"Apache-2.0"
] | 46 |
79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
https://github.com/UniSerj/ai-research/tree/79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
AdaIN
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/sk/csk3ycujt4oqsdqux3kwturlhz5jptl4gi3ixgnxxpjkkmqpjchy.py
# Topologically Sorted Source Nodes: [mean_1, sub, var_1, normalized_feat, var, mul, mean, add_2], Original ATen: [aten.mean, aten.sub, aten.var, aten.div, aten.mul, aten.add]
# Source node to ATen node mapping:
# add_2 => add_2
# mean => mean
# mean_1 => mean_1
# mul => mul
# normalized_feat => div
# sub => sub
# var => var
# var_1 => var_1
# Graph fragment:
# %mean_1 : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%view_6, [2]), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %expand), kwargs = {})
# %var_1 : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%view_4, [2]), kwargs = {correction: 1})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %expand_1), kwargs = {})
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%view, [2]), kwargs = {correction: 1})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %expand_2), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%view_2, [2]), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %expand_3), kwargs = {})
triton_per_fused_add_div_mean_mul_sub_var_0 = async_compile.triton('triton_per_fused_add_div_mean_mul_sub_var_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[16, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_mean_mul_sub_var_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 8, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_mean_mul_sub_var_0(in_ptr0, in_ptr1, out_ptr4, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 16
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (16*x0)), xmask, other=0.0)
tmp19 = tl.load(in_ptr1 + (r1 + (16*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp6 = tl.where(xmask, tmp4, 0)
tmp7 = tl.sum(tmp6, 1)[:, None]
tmp8 = tl.full([XBLOCK, 1], 16, tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 / tmp9
tmp11 = tmp1 - tmp10
tmp12 = tmp11 * tmp11
tmp13 = tl.broadcast_to(tmp12, [XBLOCK, RBLOCK])
tmp15 = tl.where(xmask, tmp13, 0)
tmp16 = tl.sum(tmp15, 1)[:, None]
tmp18 = tl.sum(tmp3, 1)[:, None]
tmp20 = tl.broadcast_to(tmp19, [XBLOCK, RBLOCK])
tmp22 = tl.where(xmask, tmp20, 0)
tmp23 = tl.sum(tmp22, 1)[:, None]
tmp25 = tl.broadcast_to(tmp20, [XBLOCK, RBLOCK])
tmp27 = tl.where(xmask, tmp25, 0)
tmp28 = tl.sum(tmp27, 1)[:, None]
tmp29 = tmp28 / tmp9
tmp30 = tmp20 - tmp29
tmp31 = tmp30 * tmp30
tmp32 = tl.broadcast_to(tmp31, [XBLOCK, RBLOCK])
tmp34 = tl.where(xmask, tmp32, 0)
tmp35 = tl.sum(tmp34, 1)[:, None]
tmp36 = 16.0
tmp37 = tmp23 / tmp36
tmp38 = tmp19 - tmp37
tmp39 = 15.0
tmp40 = tmp35 / tmp39
tmp41 = 1e-05
tmp42 = tmp40 + tmp41
tmp43 = libdevice.sqrt(tmp42)
tmp44 = tmp38 / tmp43
tmp45 = tmp16 / tmp39
tmp46 = tmp45 + tmp41
tmp47 = libdevice.sqrt(tmp46)
tmp48 = tmp44 * tmp47
tmp49 = tmp18 / tmp36
tmp50 = tmp48 + tmp49
tl.store(out_ptr4 + (r1 + (16*x0)), tmp50, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mean_1, sub, var_1, normalized_feat, var, mul, mean, add_2], Original ATen: [aten.mean, aten.sub, aten.var, aten.div, aten.mul, aten.add]
stream0 = get_raw_stream(0)
triton_per_fused_add_div_mean_mul_sub_var_0.run(arg1_1, arg0_1, buf8, 16, 16, grid=grid(16), stream=stream0)
del arg0_1
del arg1_1
return (buf8, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn as nn
import torch.optim
import torch.utils.data
def calc_mean_std(feat, eps=1e-05):
size = feat.size()
assert len(size) == 4
N, C = size[:2]
feat_var = feat.view(N, C, -1).var(dim=2) + eps
feat_std = feat_var.sqrt().view(N, C, 1, 1)
feat_mean = feat.view(N, C, -1).mean(dim=2).view(N, C, 1, 1)
return feat_mean, feat_std
class AdaIN(nn.Module):
"""
Adaptive-Instance-Normalization
"""
def __init__(self):
super().__init__()
def forward(self, content_feat, style_feat):
assert content_feat.size()[:2] == style_feat.size()[:2]
size = content_feat.size()
style_mean, style_std = calc_mean_std(style_feat)
content_mean, content_std = calc_mean_std(content_feat)
normalized_feat = (content_feat - content_mean.expand(size)
) / content_std.expand(size)
return normalized_feat * style_std.expand(size) + style_mean.expand(
size)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
import torch.optim
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_add_div_mean_mul_sub_var_0(in_ptr0, in_ptr1, out_ptr4,
xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 16 * x0), xmask, other=0.0)
tmp19 = tl.load(in_ptr1 + (r1 + 16 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp6 = tl.where(xmask, tmp4, 0)
tmp7 = tl.sum(tmp6, 1)[:, None]
tmp8 = tl.full([XBLOCK, 1], 16, tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 / tmp9
tmp11 = tmp1 - tmp10
tmp12 = tmp11 * tmp11
tmp13 = tl.broadcast_to(tmp12, [XBLOCK, RBLOCK])
tmp15 = tl.where(xmask, tmp13, 0)
tmp16 = tl.sum(tmp15, 1)[:, None]
tmp18 = tl.sum(tmp3, 1)[:, None]
tmp20 = tl.broadcast_to(tmp19, [XBLOCK, RBLOCK])
tmp22 = tl.where(xmask, tmp20, 0)
tmp23 = tl.sum(tmp22, 1)[:, None]
tmp25 = tl.broadcast_to(tmp20, [XBLOCK, RBLOCK])
tmp27 = tl.where(xmask, tmp25, 0)
tmp28 = tl.sum(tmp27, 1)[:, None]
tmp29 = tmp28 / tmp9
tmp30 = tmp20 - tmp29
tmp31 = tmp30 * tmp30
tmp32 = tl.broadcast_to(tmp31, [XBLOCK, RBLOCK])
tmp34 = tl.where(xmask, tmp32, 0)
tmp35 = tl.sum(tmp34, 1)[:, None]
tmp36 = 16.0
tmp37 = tmp23 / tmp36
tmp38 = tmp19 - tmp37
tmp39 = 15.0
tmp40 = tmp35 / tmp39
tmp41 = 1e-05
tmp42 = tmp40 + tmp41
tmp43 = libdevice.sqrt(tmp42)
tmp44 = tmp38 / tmp43
tmp45 = tmp16 / tmp39
tmp46 = tmp45 + tmp41
tmp47 = libdevice.sqrt(tmp46)
tmp48 = tmp44 * tmp47
tmp49 = tmp18 / tmp36
tmp50 = tmp48 + tmp49
tl.store(out_ptr4 + (r1 + 16 * x0), tmp50, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_per_fused_add_div_mean_mul_sub_var_0[grid(16)](arg1_1,
arg0_1, buf8, 16, 16, XBLOCK=8, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf8,
def calc_mean_std(feat, eps=1e-05):
size = feat.size()
assert len(size) == 4
N, C = size[:2]
feat_var = feat.view(N, C, -1).var(dim=2) + eps
feat_std = feat_var.sqrt().view(N, C, 1, 1)
feat_mean = feat.view(N, C, -1).mean(dim=2).view(N, C, 1, 1)
return feat_mean, feat_std
class AdaINNew(nn.Module):
"""
Adaptive-Instance-Normalization
"""
def __init__(self):
super().__init__()
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
VITA-Group/Sandwich-Batch-Normalization
|
AdaIN
| false | 14,537 |
[
"MIT"
] | 46 |
25e7df6e64a67cebd7e70b911f874cfc1bd19df0
|
https://github.com/VITA-Group/Sandwich-Batch-Normalization/tree/25e7df6e64a67cebd7e70b911f874cfc1bd19df0
|
SCRM
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/fh/cfhnguw4v6uy4ysjg54ojclakwi3bj2lte6oqizl4rpf4lcxpiyp.py
# Topologically Sorted Source Nodes: [f_s_1], Original ATen: [aten.div]
# Source node to ATen node mapping:
# f_s_1 => div
# Graph fragment:
# %div : [num_users=8] = call_function[target=torch.ops.aten.div.Tensor](args = (%select, %expand), kwargs = {})
triton_poi_fused_div_0 = async_compile.triton('triton_poi_fused_div_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x3), tmp15, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/vb/cvbckpx36pmi43pm4nuhmb6mk2i2izcpcwp6w2mt4dtyxosapprc.py
# Topologically Sorted Source Nodes: [f_s_3], Original ATen: [aten.div]
# Source node to ATen node mapping:
# f_s_3 => div_6
# Graph fragment:
# %div_6 : [num_users=8] = call_function[target=torch.ops.aten.div.Tensor](args = (%select_1, %expand_4), kwargs = {})
triton_poi_fused_div_1 = async_compile.triton('triton_poi_fused_div_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (256 + x3), xmask)
tmp1 = tl.load(in_ptr0 + (256 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (272 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (288 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (304 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x3), tmp15, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/gx/cgxjobnzpzl5xuw5yvex6rq6zcbzjr3ccnjs2perkcswcbvp6qmi.py
# Topologically Sorted Source Nodes: [f_s_5], Original ATen: [aten.div]
# Source node to ATen node mapping:
# f_s_5 => div_12
# Graph fragment:
# %div_12 : [num_users=8] = call_function[target=torch.ops.aten.div.Tensor](args = (%select_2, %expand_8), kwargs = {})
triton_poi_fused_div_2 = async_compile.triton('triton_poi_fused_div_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (512 + x3), xmask)
tmp1 = tl.load(in_ptr0 + (512 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (528 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (544 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (560 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x3), tmp15, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/n2/cn2spcpuvudmkpvrlw2achoyo7v7k2vktctfincso3dyfom5oz4x.py
# Topologically Sorted Source Nodes: [f_s_7], Original ATen: [aten.div]
# Source node to ATen node mapping:
# f_s_7 => div_18
# Graph fragment:
# %div_18 : [num_users=8] = call_function[target=torch.ops.aten.div.Tensor](args = (%select_3, %expand_12), kwargs = {})
triton_poi_fused_div_3 = async_compile.triton('triton_poi_fused_div_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (768 + x3), xmask)
tmp1 = tl.load(in_ptr0 + (768 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (784 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (800 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (816 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x3), tmp15, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/4m/c4m2wk5rzhradyfdzzmfqu2fhq5ceu6npmbjot6ovfl5pvukl5cw.py
# Topologically Sorted Source Nodes: [energy_new], Original ATen: [aten.sub]
# Source node to ATen node mapping:
# energy_new => sub_3
# Graph fragment:
# %sub_3 : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%expand_2, %bmm_4), kwargs = {})
triton_poi_fused_sub_4 = async_compile.triton('triton_poi_fused_sub_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sub_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sub_4(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (x2), xmask)
tmp2 = triton_helpers.maximum(tmp0, tmp1)
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp6 = triton_helpers.maximum(tmp4, tmp5)
tmp8 = tmp6 - tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/j7/cj7mv2k5l2kigfluq2rwwpouckm4oow7jia7wwvjogp3qlr23xwv.py
# Topologically Sorted Source Nodes: [attention_2], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attention_2 => amax_2, exp_2, sub_4
# Graph fragment:
# %amax_2 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%sub_3, [-1], True), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub_3, %amax_2), kwargs = {})
# %exp_2 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_4,), kwargs = {})
triton_poi_fused__softmax_5 = async_compile.triton('triton_poi_fused__softmax_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_5(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/b4/cb4vdtviarmy2ckmxjdkc3dnwp4gssl3dh4u4col3s2illvdpvql.py
# Topologically Sorted Source Nodes: [attention_2], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attention_2 => div_4, sum_5
# Graph fragment:
# %sum_5 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_2, [-1], True), kwargs = {})
# %div_4 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_2, %sum_5), kwargs = {})
triton_poi_fused__softmax_6 = async_compile.triton('triton_poi_fused__softmax_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_6(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/ps/cpsrn3vzizx4kiw2ao7vdniku66p6jnqllxcgsuag4wnndz4qytq.py
# Topologically Sorted Source Nodes: [attention], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attention => amax, div_2, exp, sub, sum_3
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%bmm, [-1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%bmm, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_2 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_3), kwargs = {})
triton_per_fused__softmax_7 = async_compile.triton('triton_per_fused__softmax_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[64, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__softmax_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 2, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__softmax_7(in_ptr0, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 64
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (16*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, float("-inf"))
tmp4 = triton_helpers.max2(tmp3, 1)[:, None]
tmp5 = tmp0 - tmp4
tmp6 = tl_math.exp(tmp5)
tmp7 = tl.broadcast_to(tmp6, [XBLOCK, RBLOCK])
tmp9 = tl.where(xmask, tmp7, 0)
tmp10 = tl.sum(tmp9, 1)[:, None]
tmp11 = tmp6 / tmp10
tl.store(out_ptr2 + (r1 + (16*x0)), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_0/inductor_cache/2r/c2rthu53scnzgesjbiparrgvjr27o2yifa3oqcdcu5hbzqwmf3wx.py
# Topologically Sorted Source Nodes: [mul_2, out_8, mul_3, out_11, ca_loss, mul, out_2, mul_1, out_5, sa_loss, add_4, add_20, mul_6, out_20, mul_7, out_23, ca_loss_1, mul_4, out_14, mul_5, out_17, sa_loss_1, add_9, add_21, mul_10, out_32, mul_11, out_35, ca_loss_2, mul_8, out_26, mul_9, out_29, sa_loss_2, add_14, add_22, mul_14, out_44, mul_15, out_47, ca_loss_3, mul_12, out_38, mul_13, out_41, sa_loss_3, add_19, add_23], Original ATen: [aten.mul, aten.add, aten.sub, aten.abs, aten.mean]
# Source node to ATen node mapping:
# add_14 => add_14
# add_19 => add_19
# add_20 => add_20
# add_21 => add_21
# add_22 => add_22
# add_23 => add_23
# add_4 => add_4
# add_9 => add_9
# ca_loss => abs_2, mean_1, sub_7
# ca_loss_1 => abs_4, mean_3, sub_15
# ca_loss_2 => abs_6, mean_5, sub_23
# ca_loss_3 => abs_8, mean_7, sub_31
# mul => mul
# mul_1 => mul_1
# mul_10 => mul_10
# mul_11 => mul_11
# mul_12 => mul_12
# mul_13 => mul_13
# mul_14 => mul_14
# mul_15 => mul_15
# mul_2 => mul_2
# mul_3 => mul_3
# mul_4 => mul_4
# mul_5 => mul_5
# mul_6 => mul_6
# mul_7 => mul_7
# mul_8 => mul_8
# mul_9 => mul_9
# out_11 => add_3
# out_14 => add_5
# out_17 => add_6
# out_2 => add
# out_20 => add_7
# out_23 => add_8
# out_26 => add_10
# out_29 => add_11
# out_32 => add_12
# out_35 => add_13
# out_38 => add_15
# out_41 => add_16
# out_44 => add_17
# out_47 => add_18
# out_5 => add_1
# out_8 => add_2
# sa_loss => abs_1, mean, sub_2
# sa_loss_1 => abs_3, mean_2, sub_10
# sa_loss_2 => abs_5, mean_4, sub_18
# sa_loss_3 => abs_7, mean_6, sub_26
# Graph fragment:
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_11, 0.1), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_2, %div), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_15, 0.1), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_3, %div_1), kwargs = {})
# %sub_7 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_2, %add_3), kwargs = {})
# %abs_2 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_7,), kwargs = {})
# %mean_1 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_2,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_3, 0.1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %div), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_7, 0.1), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %div_1), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %add_1), kwargs = {})
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_2,), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_1,), kwargs = {})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean_1, %mean), kwargs = {})
# %add_20 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_4, 0), kwargs = {})
# %mul_6 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_27, 0.1), kwargs = {})
# %add_7 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_6, %div_6), kwargs = {})
# %mul_7 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_31, 0.1), kwargs = {})
# %add_8 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_7, %div_7), kwargs = {})
# %sub_15 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_7, %add_8), kwargs = {})
# %abs_4 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_15,), kwargs = {})
# %mean_3 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_4,), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_19, 0.1), kwargs = {})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_4, %div_6), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_23, 0.1), kwargs = {})
# %add_6 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_5, %div_7), kwargs = {})
# %sub_10 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_5, %add_6), kwargs = {})
# %abs_3 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_10,), kwargs = {})
# %mean_2 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_3,), kwargs = {})
# %add_9 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean_3, %mean_2), kwargs = {})
# %add_21 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_20, %add_9), kwargs = {})
# %mul_10 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_43, 0.1), kwargs = {})
# %add_12 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_10, %div_12), kwargs = {})
# %mul_11 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_47, 0.1), kwargs = {})
# %add_13 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_11, %div_13), kwargs = {})
# %sub_23 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_12, %add_13), kwargs = {})
# %abs_6 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_23,), kwargs = {})
# %mean_5 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_6,), kwargs = {})
# %mul_8 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_35, 0.1), kwargs = {})
# %add_10 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_8, %div_12), kwargs = {})
# %mul_9 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_39, 0.1), kwargs = {})
# %add_11 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_9, %div_13), kwargs = {})
# %sub_18 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_10, %add_11), kwargs = {})
# %abs_5 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_18,), kwargs = {})
# %mean_4 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_5,), kwargs = {})
# %add_14 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean_5, %mean_4), kwargs = {})
# %add_22 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_21, %add_14), kwargs = {})
# %mul_14 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_59, 0.1), kwargs = {})
# %add_17 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_14, %div_18), kwargs = {})
# %mul_15 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_63, 0.1), kwargs = {})
# %add_18 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_15, %div_19), kwargs = {})
# %sub_31 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_17, %add_18), kwargs = {})
# %abs_8 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_31,), kwargs = {})
# %mean_7 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_8,), kwargs = {})
# %mul_12 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_51, 0.1), kwargs = {})
# %add_15 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_12, %div_18), kwargs = {})
# %mul_13 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_55, 0.1), kwargs = {})
# %add_16 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_13, %div_19), kwargs = {})
# %sub_26 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_15, %add_16), kwargs = {})
# %abs_7 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%sub_26,), kwargs = {})
# %mean_6 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_7,), kwargs = {})
# %add_19 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean_7, %mean_6), kwargs = {})
# %add_23 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_22, %add_19), kwargs = {})
triton_per_fused_abs_add_mean_mul_sub_8 = async_compile.triton('triton_per_fused_abs_add_mean_mul_sub_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: '*fp32', 8: '*fp32', 9: '*fp32', 10: '*fp32', 11: '*fp32', 12: '*fp32', 13: '*fp32', 14: '*fp32', 15: '*fp32', 16: '*fp32', 17: '*fp32', 18: '*fp32', 19: '*fp32', 20: '*fp32', 21: '*fp32', 22: '*fp32', 23: '*fp32', 24: '*fp32', 25: 'i32', 26: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {25: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 26), equal_to_1=(25,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_abs_add_mean_mul_sub_8', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 24, 'num_reduction': 8, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_abs_add_mean_mul_sub_8(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, in_ptr8, in_ptr9, in_ptr10, in_ptr11, in_ptr12, in_ptr13, in_ptr14, in_ptr15, in_ptr16, in_ptr17, in_ptr18, in_ptr19, in_ptr20, in_ptr21, in_ptr22, in_ptr23, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp3 = tl.load(in_ptr1 + (r0), None)
tmp5 = tl.load(in_ptr2 + (r0), None)
tmp7 = tl.load(in_ptr3 + (r0), None)
tmp14 = tl.load(in_ptr4 + (r0), None)
tmp17 = tl.load(in_ptr5 + (r0), None)
tmp25 = tl.load(in_ptr6 + (r0), None)
tmp27 = tl.load(in_ptr7 + (r0), None)
tmp29 = tl.load(in_ptr8 + (r0), None)
tmp31 = tl.load(in_ptr9 + (r0), None)
tmp38 = tl.load(in_ptr10 + (r0), None)
tmp41 = tl.load(in_ptr11 + (r0), None)
tmp49 = tl.load(in_ptr12 + (r0), None)
tmp51 = tl.load(in_ptr13 + (r0), None)
tmp53 = tl.load(in_ptr14 + (r0), None)
tmp55 = tl.load(in_ptr15 + (r0), None)
tmp62 = tl.load(in_ptr16 + (r0), None)
tmp65 = tl.load(in_ptr17 + (r0), None)
tmp73 = tl.load(in_ptr18 + (r0), None)
tmp75 = tl.load(in_ptr19 + (r0), None)
tmp77 = tl.load(in_ptr20 + (r0), None)
tmp79 = tl.load(in_ptr21 + (r0), None)
tmp86 = tl.load(in_ptr22 + (r0), None)
tmp89 = tl.load(in_ptr23 + (r0), None)
tmp1 = 0.1
tmp2 = tmp0 * tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp5 * tmp1
tmp8 = tmp6 + tmp7
tmp9 = tmp4 - tmp8
tmp10 = tl_math.abs(tmp9)
tmp11 = tl.broadcast_to(tmp10, [RBLOCK])
tmp13 = triton_helpers.promote_to_tensor(tl.sum(tmp11, 0))
tmp15 = tmp14 * tmp1
tmp16 = tmp15 + tmp3
tmp18 = tmp17 * tmp1
tmp19 = tmp18 + tmp7
tmp20 = tmp16 - tmp19
tmp21 = tl_math.abs(tmp20)
tmp22 = tl.broadcast_to(tmp21, [RBLOCK])
tmp24 = triton_helpers.promote_to_tensor(tl.sum(tmp22, 0))
tmp26 = tmp25 * tmp1
tmp28 = tmp26 + tmp27
tmp30 = tmp29 * tmp1
tmp32 = tmp30 + tmp31
tmp33 = tmp28 - tmp32
tmp34 = tl_math.abs(tmp33)
tmp35 = tl.broadcast_to(tmp34, [RBLOCK])
tmp37 = triton_helpers.promote_to_tensor(tl.sum(tmp35, 0))
tmp39 = tmp38 * tmp1
tmp40 = tmp39 + tmp27
tmp42 = tmp41 * tmp1
tmp43 = tmp42 + tmp31
tmp44 = tmp40 - tmp43
tmp45 = tl_math.abs(tmp44)
tmp46 = tl.broadcast_to(tmp45, [RBLOCK])
tmp48 = triton_helpers.promote_to_tensor(tl.sum(tmp46, 0))
tmp50 = tmp49 * tmp1
tmp52 = tmp50 + tmp51
tmp54 = tmp53 * tmp1
tmp56 = tmp54 + tmp55
tmp57 = tmp52 - tmp56
tmp58 = tl_math.abs(tmp57)
tmp59 = tl.broadcast_to(tmp58, [RBLOCK])
tmp61 = triton_helpers.promote_to_tensor(tl.sum(tmp59, 0))
tmp63 = tmp62 * tmp1
tmp64 = tmp63 + tmp51
tmp66 = tmp65 * tmp1
tmp67 = tmp66 + tmp55
tmp68 = tmp64 - tmp67
tmp69 = tl_math.abs(tmp68)
tmp70 = tl.broadcast_to(tmp69, [RBLOCK])
tmp72 = triton_helpers.promote_to_tensor(tl.sum(tmp70, 0))
tmp74 = tmp73 * tmp1
tmp76 = tmp74 + tmp75
tmp78 = tmp77 * tmp1
tmp80 = tmp78 + tmp79
tmp81 = tmp76 - tmp80
tmp82 = tl_math.abs(tmp81)
tmp83 = tl.broadcast_to(tmp82, [RBLOCK])
tmp85 = triton_helpers.promote_to_tensor(tl.sum(tmp83, 0))
tmp87 = tmp86 * tmp1
tmp88 = tmp87 + tmp75
tmp90 = tmp89 * tmp1
tmp91 = tmp90 + tmp79
tmp92 = tmp88 - tmp91
tmp93 = tl_math.abs(tmp92)
tmp94 = tl.broadcast_to(tmp93, [RBLOCK])
tmp96 = triton_helpers.promote_to_tensor(tl.sum(tmp94, 0))
tmp97 = 256.0
tmp98 = tmp13 / tmp97
tmp99 = tmp24 / tmp97
tmp100 = tmp98 + tmp99
tmp101 = 0.0
tmp102 = tmp100 + tmp101
tmp103 = tmp37 / tmp97
tmp104 = tmp48 / tmp97
tmp105 = tmp103 + tmp104
tmp106 = tmp102 + tmp105
tmp107 = tmp61 / tmp97
tmp108 = tmp72 / tmp97
tmp109 = tmp107 + tmp108
tmp110 = tmp106 + tmp109
tmp111 = tmp85 / tmp97
tmp112 = tmp96 / tmp97
tmp113 = tmp111 + tmp112
tmp114 = tmp110 + tmp113
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp114, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_s_1], Original ATen: [aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_div_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_2], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf0, (4, 16, 4), (64, 1, 16), 0), out=buf1)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_t_1], Original ATen: [aten.div]
triton_poi_fused_div_0.run(arg1_1, buf2, 256, grid=grid(256), stream=stream0)
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_3], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf2, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf2, (4, 16, 4), (64, 1, 16), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_s_3], Original ATen: [aten.div]
triton_poi_fused_div_1.run(arg0_1, buf4, 256, grid=grid(256), stream=stream0)
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_6], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf4, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf4, (4, 16, 4), (64, 1, 16), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_t_3], Original ATen: [aten.div]
triton_poi_fused_div_1.run(arg1_1, buf6, 256, grid=grid(256), stream=stream0)
buf7 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_7], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf6, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf6, (4, 16, 4), (64, 1, 16), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_s_5], Original ATen: [aten.div]
triton_poi_fused_div_2.run(arg0_1, buf8, 256, grid=grid(256), stream=stream0)
buf9 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_10], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf8, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf8, (4, 16, 4), (64, 1, 16), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_t_5], Original ATen: [aten.div]
triton_poi_fused_div_2.run(arg1_1, buf10, 256, grid=grid(256), stream=stream0)
buf11 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_11], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf10, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf10, (4, 16, 4), (64, 1, 16), 0), out=buf11)
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_s_7], Original ATen: [aten.div]
triton_poi_fused_div_3.run(arg0_1, buf12, 256, grid=grid(256), stream=stream0)
del arg0_1
buf13 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_14], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf12, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf12, (4, 16, 4), (64, 1, 16), 0), out=buf13)
buf14 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_t_7], Original ATen: [aten.div]
triton_poi_fused_div_3.run(arg1_1, buf14, 256, grid=grid(256), stream=stream0)
del arg1_1
buf15 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_15], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf14, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf14, (4, 16, 4), (64, 1, 16), 0), out=buf15)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy_new], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf1, buf16, 64, grid=grid(64), stream=stream0)
buf17 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [attention_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf16, buf17, 64, grid=grid(64), stream=stream0)
buf18 = buf16; del buf16 # reuse
# Topologically Sorted Source Nodes: [attention_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf17, buf18, 64, grid=grid(64), stream=stream0)
del buf17
buf19 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_2, out_6], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf18, reinterpret_tensor(buf0, (4, 4, 16), (64, 16, 1), 0), out=buf19)
buf20 = buf18; del buf18 # reuse
# Topologically Sorted Source Nodes: [energy_new_1], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf3, buf20, 64, grid=grid(64), stream=stream0)
buf21 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [attention_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf20, buf21, 64, grid=grid(64), stream=stream0)
buf22 = buf20; del buf20 # reuse
# Topologically Sorted Source Nodes: [attention_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf21, buf22, 64, grid=grid(64), stream=stream0)
del buf21
buf23 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_3, out_9], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf22, reinterpret_tensor(buf2, (4, 4, 16), (64, 16, 1), 0), out=buf23)
buf25 = empty_strided_cuda((4, 16, 16), (256, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [energy], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf0, (4, 4, 16), (64, 16, 1), 0), out=buf25)
buf28 = empty_strided_cuda((4, 16, 16), (256, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf25, buf28, 64, 16, grid=grid(64), stream=stream0)
buf29 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf28, (4, 16, 16), (256, 1, 16), 0), out=buf29)
buf30 = buf28; del buf28 # reuse
# Topologically Sorted Source Nodes: [energy_1], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf2, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf2, (4, 4, 16), (64, 16, 1), 0), out=buf30)
buf33 = buf25; del buf25 # reuse
# Topologically Sorted Source Nodes: [attention_1], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf30, buf33, 64, 16, grid=grid(64), stream=stream0)
buf34 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_3], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf2, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf33, (4, 16, 16), (256, 1, 16), 0), out=buf34)
buf36 = buf22; del buf22 # reuse
# Topologically Sorted Source Nodes: [energy_new_2], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf5, buf36, 64, grid=grid(64), stream=stream0)
buf37 = buf5; del buf5 # reuse
# Topologically Sorted Source Nodes: [attention_6], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf36, buf37, 64, grid=grid(64), stream=stream0)
buf38 = buf36; del buf36 # reuse
# Topologically Sorted Source Nodes: [attention_6], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf37, buf38, 64, grid=grid(64), stream=stream0)
del buf37
buf39 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_6, out_18], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf38, reinterpret_tensor(buf4, (4, 4, 16), (64, 16, 1), 0), out=buf39)
buf40 = buf38; del buf38 # reuse
# Topologically Sorted Source Nodes: [energy_new_3], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf7, buf40, 64, grid=grid(64), stream=stream0)
buf41 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [attention_7], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf40, buf41, 64, grid=grid(64), stream=stream0)
buf42 = buf40; del buf40 # reuse
# Topologically Sorted Source Nodes: [attention_7], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf41, buf42, 64, grid=grid(64), stream=stream0)
del buf41
buf43 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_7, out_21], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf42, reinterpret_tensor(buf6, (4, 4, 16), (64, 16, 1), 0), out=buf43)
buf45 = buf33; del buf33 # reuse
# Topologically Sorted Source Nodes: [energy_4], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf4, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf4, (4, 4, 16), (64, 16, 1), 0), out=buf45)
buf48 = buf30; del buf30 # reuse
# Topologically Sorted Source Nodes: [attention_4], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf45, buf48, 64, 16, grid=grid(64), stream=stream0)
buf49 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_12], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf4, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf48, (4, 16, 16), (256, 1, 16), 0), out=buf49)
buf50 = buf48; del buf48 # reuse
# Topologically Sorted Source Nodes: [energy_5], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf6, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf6, (4, 4, 16), (64, 16, 1), 0), out=buf50)
buf53 = buf45; del buf45 # reuse
# Topologically Sorted Source Nodes: [attention_5], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf50, buf53, 64, 16, grid=grid(64), stream=stream0)
buf54 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_15], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf6, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf53, (4, 16, 16), (256, 1, 16), 0), out=buf54)
buf56 = buf42; del buf42 # reuse
# Topologically Sorted Source Nodes: [energy_new_4], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf9, buf56, 64, grid=grid(64), stream=stream0)
buf57 = buf9; del buf9 # reuse
# Topologically Sorted Source Nodes: [attention_10], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf56, buf57, 64, grid=grid(64), stream=stream0)
buf58 = buf56; del buf56 # reuse
# Topologically Sorted Source Nodes: [attention_10], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf57, buf58, 64, grid=grid(64), stream=stream0)
del buf57
buf59 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_10, out_30], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf58, reinterpret_tensor(buf8, (4, 4, 16), (64, 16, 1), 0), out=buf59)
buf60 = buf58; del buf58 # reuse
# Topologically Sorted Source Nodes: [energy_new_5], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf11, buf60, 64, grid=grid(64), stream=stream0)
buf61 = buf11; del buf11 # reuse
# Topologically Sorted Source Nodes: [attention_11], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf60, buf61, 64, grid=grid(64), stream=stream0)
buf62 = buf60; del buf60 # reuse
# Topologically Sorted Source Nodes: [attention_11], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf61, buf62, 64, grid=grid(64), stream=stream0)
del buf61
buf63 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_11, out_33], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf62, reinterpret_tensor(buf10, (4, 4, 16), (64, 16, 1), 0), out=buf63)
buf65 = buf53; del buf53 # reuse
# Topologically Sorted Source Nodes: [energy_8], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf8, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf8, (4, 4, 16), (64, 16, 1), 0), out=buf65)
buf68 = buf50; del buf50 # reuse
# Topologically Sorted Source Nodes: [attention_8], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf65, buf68, 64, 16, grid=grid(64), stream=stream0)
buf69 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_24], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf8, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf68, (4, 16, 16), (256, 1, 16), 0), out=buf69)
buf70 = buf68; del buf68 # reuse
# Topologically Sorted Source Nodes: [energy_9], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf10, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf10, (4, 4, 16), (64, 16, 1), 0), out=buf70)
buf73 = buf65; del buf65 # reuse
# Topologically Sorted Source Nodes: [attention_9], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf70, buf73, 64, 16, grid=grid(64), stream=stream0)
buf74 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_27], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf10, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf73, (4, 16, 16), (256, 1, 16), 0), out=buf74)
buf76 = buf62; del buf62 # reuse
# Topologically Sorted Source Nodes: [energy_new_6], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf13, buf76, 64, grid=grid(64), stream=stream0)
buf77 = buf13; del buf13 # reuse
# Topologically Sorted Source Nodes: [attention_14], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf76, buf77, 64, grid=grid(64), stream=stream0)
buf78 = buf76; del buf76 # reuse
# Topologically Sorted Source Nodes: [attention_14], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf77, buf78, 64, grid=grid(64), stream=stream0)
del buf77
buf79 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_14, out_42], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf78, reinterpret_tensor(buf12, (4, 4, 16), (64, 16, 1), 0), out=buf79)
buf80 = buf78; del buf78 # reuse
# Topologically Sorted Source Nodes: [energy_new_7], Original ATen: [aten.sub]
triton_poi_fused_sub_4.run(buf15, buf80, 64, grid=grid(64), stream=stream0)
buf81 = buf15; del buf15 # reuse
# Topologically Sorted Source Nodes: [attention_15], Original ATen: [aten._softmax]
triton_poi_fused__softmax_5.run(buf80, buf81, 64, grid=grid(64), stream=stream0)
buf82 = buf80; del buf80 # reuse
# Topologically Sorted Source Nodes: [attention_15], Original ATen: [aten._softmax]
triton_poi_fused__softmax_6.run(buf81, buf82, 64, grid=grid(64), stream=stream0)
del buf81
buf83 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_15, out_45], Original ATen: [aten._softmax, aten.bmm]
extern_kernels.bmm(buf82, reinterpret_tensor(buf14, (4, 4, 16), (64, 16, 1), 0), out=buf83)
del buf82
buf85 = buf73; del buf73 # reuse
# Topologically Sorted Source Nodes: [energy_12], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf12, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf12, (4, 4, 16), (64, 16, 1), 0), out=buf85)
buf88 = buf70; del buf70 # reuse
# Topologically Sorted Source Nodes: [attention_12], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf85, buf88, 64, 16, grid=grid(64), stream=stream0)
buf89 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_36], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf12, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf88, (4, 16, 16), (256, 1, 16), 0), out=buf89)
buf90 = buf88; del buf88 # reuse
# Topologically Sorted Source Nodes: [energy_13], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf14, (4, 16, 4), (64, 1, 16), 0), reinterpret_tensor(buf14, (4, 4, 16), (64, 16, 1), 0), out=buf90)
buf93 = buf85; del buf85 # reuse
# Topologically Sorted Source Nodes: [attention_13], Original ATen: [aten._softmax]
triton_per_fused__softmax_7.run(buf90, buf93, 64, 16, grid=grid(64), stream=stream0)
del buf90
buf94 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_39], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf14, (4, 4, 16), (64, 16, 1), 0), reinterpret_tensor(buf93, (4, 16, 16), (256, 1, 16), 0), out=buf94)
del buf93
buf24 = empty_strided_cuda((), (), torch.float32)
buf96 = buf24; del buf24 # reuse
# Topologically Sorted Source Nodes: [mul_2, out_8, mul_3, out_11, ca_loss, mul, out_2, mul_1, out_5, sa_loss, add_4, add_20, mul_6, out_20, mul_7, out_23, ca_loss_1, mul_4, out_14, mul_5, out_17, sa_loss_1, add_9, add_21, mul_10, out_32, mul_11, out_35, ca_loss_2, mul_8, out_26, mul_9, out_29, sa_loss_2, add_14, add_22, mul_14, out_44, mul_15, out_47, ca_loss_3, mul_12, out_38, mul_13, out_41, sa_loss_3, add_19, add_23], Original ATen: [aten.mul, aten.add, aten.sub, aten.abs, aten.mean]
triton_per_fused_abs_add_mean_mul_sub_8.run(buf96, buf19, buf0, buf23, buf2, buf29, buf34, buf39, buf4, buf43, buf6, buf49, buf54, buf59, buf8, buf63, buf10, buf69, buf74, buf79, buf12, buf83, buf14, buf89, buf94, 1, 256, grid=grid(1), stream=stream0)
del buf0
del buf10
del buf12
del buf14
del buf19
del buf2
del buf23
del buf29
del buf34
del buf39
del buf4
del buf43
del buf49
del buf54
del buf59
del buf6
del buf63
del buf69
del buf74
del buf79
del buf8
del buf83
del buf89
del buf94
return (buf96, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import torch.nn.functional as F
import torch.nn as nn
class SCRM(nn.Module):
"""
spatial & channel wise relation loss
"""
def __init__(self, gamma=0.1):
super(SCRM, self).__init__()
self.softmax = nn.Softmax(dim=-1)
self.gamma = gamma
def spatial_wise(self, x):
m_batchsize, C, height, width = x.size()
proj_query = x.view(m_batchsize, -1, width * height).permute(0, 2, 1)
proj_key = x.view(m_batchsize, -1, width * height)
energy = torch.bmm(proj_query, proj_key)
attention = self.softmax(energy)
proj_value = x.view(m_batchsize, -1, width * height)
out = torch.bmm(proj_value, attention.permute(0, 2, 1))
out = out.view(m_batchsize, C, height, width)
out = self.gamma * out + x
return out
def channel_wise(self, x):
m_batchsize, C, height, width = x.size()
proj_query = x.view(m_batchsize, C, -1)
proj_key = x.view(m_batchsize, C, -1).permute(0, 2, 1)
energy = torch.bmm(proj_query, proj_key)
energy_new = torch.max(energy, -1, keepdim=True)[0].expand_as(energy
) - energy
attention = self.softmax(energy_new)
proj_value = x.view(m_batchsize, C, -1)
out = torch.bmm(attention, proj_value)
out = out.view(m_batchsize, C, height, width)
out = self.gamma * out + x
return out
def cal_loss(self, f_s, f_t):
f_s = F.normalize(f_s, dim=1)
f_t = F.normalize(f_t, dim=1)
sa_loss = F.l1_loss(self.spatial_wise(f_s), self.spatial_wise(f_t))
ca_loss = F.l1_loss(self.channel_wise(f_s), self.channel_wise(f_t))
return ca_loss + sa_loss
def forward(self, g_s, g_t):
return sum(self.cal_loss(f_s, f_t) for f_s, f_t in zip(g_s, g_t))
def get_inputs():
return [torch.rand([4, 4, 4, 4, 4]), torch.rand([4, 4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn.functional as F
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + x3, tmp15, xmask)
@triton.jit
def triton_poi_fused_div_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + (256 + x3), xmask)
tmp1 = tl.load(in_ptr0 + (256 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (272 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (288 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (304 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + x3, tmp15, xmask)
@triton.jit
def triton_poi_fused_div_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + (512 + x3), xmask)
tmp1 = tl.load(in_ptr0 + (512 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (528 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (544 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (560 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + x3, tmp15, xmask)
@triton.jit
def triton_poi_fused_div_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + (768 + x3), xmask)
tmp1 = tl.load(in_ptr0 + (768 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (784 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (800 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (816 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = libdevice.sqrt(tmp11)
tmp13 = 1e-12
tmp14 = triton_helpers.maximum(tmp12, tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + x3, tmp15, xmask)
@triton.jit
def triton_poi_fused_sub_4(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + x2, xmask)
tmp2 = triton_helpers.maximum(tmp0, tmp1)
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp6 = triton_helpers.maximum(tmp4, tmp5)
tmp8 = tmp6 - tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused__softmax_5(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_6(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_per_fused__softmax_7(in_ptr0, out_ptr2, xnumel, rnumel, XBLOCK:
tl.constexpr):
xnumel = 64
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 16 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, float('-inf'))
tmp4 = triton_helpers.max2(tmp3, 1)[:, None]
tmp5 = tmp0 - tmp4
tmp6 = tl_math.exp(tmp5)
tmp7 = tl.broadcast_to(tmp6, [XBLOCK, RBLOCK])
tmp9 = tl.where(xmask, tmp7, 0)
tmp10 = tl.sum(tmp9, 1)[:, None]
tmp11 = tmp6 / tmp10
tl.store(out_ptr2 + (r1 + 16 * x0), tmp11, xmask)
@triton.jit
def triton_per_fused_abs_add_mean_mul_sub_8(in_out_ptr0, in_ptr0, in_ptr1,
in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, in_ptr8, in_ptr9,
in_ptr10, in_ptr11, in_ptr12, in_ptr13, in_ptr14, in_ptr15, in_ptr16,
in_ptr17, in_ptr18, in_ptr19, in_ptr20, in_ptr21, in_ptr22, in_ptr23,
xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp3 = tl.load(in_ptr1 + r0, None)
tmp5 = tl.load(in_ptr2 + r0, None)
tmp7 = tl.load(in_ptr3 + r0, None)
tmp14 = tl.load(in_ptr4 + r0, None)
tmp17 = tl.load(in_ptr5 + r0, None)
tmp25 = tl.load(in_ptr6 + r0, None)
tmp27 = tl.load(in_ptr7 + r0, None)
tmp29 = tl.load(in_ptr8 + r0, None)
tmp31 = tl.load(in_ptr9 + r0, None)
tmp38 = tl.load(in_ptr10 + r0, None)
tmp41 = tl.load(in_ptr11 + r0, None)
tmp49 = tl.load(in_ptr12 + r0, None)
tmp51 = tl.load(in_ptr13 + r0, None)
tmp53 = tl.load(in_ptr14 + r0, None)
tmp55 = tl.load(in_ptr15 + r0, None)
tmp62 = tl.load(in_ptr16 + r0, None)
tmp65 = tl.load(in_ptr17 + r0, None)
tmp73 = tl.load(in_ptr18 + r0, None)
tmp75 = tl.load(in_ptr19 + r0, None)
tmp77 = tl.load(in_ptr20 + r0, None)
tmp79 = tl.load(in_ptr21 + r0, None)
tmp86 = tl.load(in_ptr22 + r0, None)
tmp89 = tl.load(in_ptr23 + r0, None)
tmp1 = 0.1
tmp2 = tmp0 * tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp5 * tmp1
tmp8 = tmp6 + tmp7
tmp9 = tmp4 - tmp8
tmp10 = tl_math.abs(tmp9)
tmp11 = tl.broadcast_to(tmp10, [RBLOCK])
tmp13 = triton_helpers.promote_to_tensor(tl.sum(tmp11, 0))
tmp15 = tmp14 * tmp1
tmp16 = tmp15 + tmp3
tmp18 = tmp17 * tmp1
tmp19 = tmp18 + tmp7
tmp20 = tmp16 - tmp19
tmp21 = tl_math.abs(tmp20)
tmp22 = tl.broadcast_to(tmp21, [RBLOCK])
tmp24 = triton_helpers.promote_to_tensor(tl.sum(tmp22, 0))
tmp26 = tmp25 * tmp1
tmp28 = tmp26 + tmp27
tmp30 = tmp29 * tmp1
tmp32 = tmp30 + tmp31
tmp33 = tmp28 - tmp32
tmp34 = tl_math.abs(tmp33)
tmp35 = tl.broadcast_to(tmp34, [RBLOCK])
tmp37 = triton_helpers.promote_to_tensor(tl.sum(tmp35, 0))
tmp39 = tmp38 * tmp1
tmp40 = tmp39 + tmp27
tmp42 = tmp41 * tmp1
tmp43 = tmp42 + tmp31
tmp44 = tmp40 - tmp43
tmp45 = tl_math.abs(tmp44)
tmp46 = tl.broadcast_to(tmp45, [RBLOCK])
tmp48 = triton_helpers.promote_to_tensor(tl.sum(tmp46, 0))
tmp50 = tmp49 * tmp1
tmp52 = tmp50 + tmp51
tmp54 = tmp53 * tmp1
tmp56 = tmp54 + tmp55
tmp57 = tmp52 - tmp56
tmp58 = tl_math.abs(tmp57)
tmp59 = tl.broadcast_to(tmp58, [RBLOCK])
tmp61 = triton_helpers.promote_to_tensor(tl.sum(tmp59, 0))
tmp63 = tmp62 * tmp1
tmp64 = tmp63 + tmp51
tmp66 = tmp65 * tmp1
tmp67 = tmp66 + tmp55
tmp68 = tmp64 - tmp67
tmp69 = tl_math.abs(tmp68)
tmp70 = tl.broadcast_to(tmp69, [RBLOCK])
tmp72 = triton_helpers.promote_to_tensor(tl.sum(tmp70, 0))
tmp74 = tmp73 * tmp1
tmp76 = tmp74 + tmp75
tmp78 = tmp77 * tmp1
tmp80 = tmp78 + tmp79
tmp81 = tmp76 - tmp80
tmp82 = tl_math.abs(tmp81)
tmp83 = tl.broadcast_to(tmp82, [RBLOCK])
tmp85 = triton_helpers.promote_to_tensor(tl.sum(tmp83, 0))
tmp87 = tmp86 * tmp1
tmp88 = tmp87 + tmp75
tmp90 = tmp89 * tmp1
tmp91 = tmp90 + tmp79
tmp92 = tmp88 - tmp91
tmp93 = tl_math.abs(tmp92)
tmp94 = tl.broadcast_to(tmp93, [RBLOCK])
tmp96 = triton_helpers.promote_to_tensor(tl.sum(tmp94, 0))
tmp97 = 256.0
tmp98 = tmp13 / tmp97
tmp99 = tmp24 / tmp97
tmp100 = tmp98 + tmp99
tmp101 = 0.0
tmp102 = tmp100 + tmp101
tmp103 = tmp37 / tmp97
tmp104 = tmp48 / tmp97
tmp105 = tmp103 + tmp104
tmp106 = tmp102 + tmp105
tmp107 = tmp61 / tmp97
tmp108 = tmp72 / tmp97
tmp109 = tmp107 + tmp108
tmp110 = tmp106 + tmp109
tmp111 = tmp85 / tmp97
tmp112 = tmp96 / tmp97
tmp113 = tmp111 + tmp112
tmp114 = tmp110 + tmp113
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp114, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4, 4), (256, 64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_div_0[grid(256)](arg0_1, buf0, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf0, (4, 16, 4), (64, 1, 16), 0), out=buf1)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_div_0[grid(256)](arg1_1, buf2, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf2, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf2, (4, 16, 4), (64, 1, 16), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_div_1[grid(256)](arg0_1, buf4, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf4, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf4, (4, 16, 4), (64, 1, 16), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_div_1[grid(256)](arg1_1, buf6, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf7 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf6, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf6, (4, 16, 4), (64, 1, 16), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_div_2[grid(256)](arg0_1, buf8, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf9 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf8, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf8, (4, 16, 4), (64, 1, 16), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_div_2[grid(256)](arg1_1, buf10, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf11 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf10, (4, 4, 16), (64, 16, 1
), 0), reinterpret_tensor(buf10, (4, 16, 4), (64, 1, 16), 0),
out=buf11)
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_div_3[grid(256)](arg0_1, buf12, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg0_1
buf13 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf12, (4, 4, 16), (64, 16, 1
), 0), reinterpret_tensor(buf12, (4, 16, 4), (64, 1, 16), 0),
out=buf13)
buf14 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_div_3[grid(256)](arg1_1, buf14, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg1_1
buf15 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf14, (4, 4, 16), (64, 16, 1
), 0), reinterpret_tensor(buf14, (4, 16, 4), (64, 1, 16), 0),
out=buf15)
buf16 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_sub_4[grid(64)](buf1, buf16, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf17 = buf1
del buf1
triton_poi_fused__softmax_5[grid(64)](buf16, buf17, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf18 = buf16
del buf16
triton_poi_fused__softmax_6[grid(64)](buf17, buf18, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf17
buf19 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf18, reinterpret_tensor(buf0, (4, 4, 16), (64,
16, 1), 0), out=buf19)
buf20 = buf18
del buf18
triton_poi_fused_sub_4[grid(64)](buf3, buf20, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf21 = buf3
del buf3
triton_poi_fused__softmax_5[grid(64)](buf20, buf21, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf22 = buf20
del buf20
triton_poi_fused__softmax_6[grid(64)](buf21, buf22, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf21
buf23 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf22, reinterpret_tensor(buf2, (4, 4, 16), (64,
16, 1), 0), out=buf23)
buf25 = empty_strided_cuda((4, 16, 16), (256, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 16, 4), (64, 1, 16),
0), reinterpret_tensor(buf0, (4, 4, 16), (64, 16, 1), 0), out=buf25
)
buf28 = empty_strided_cuda((4, 16, 16), (256, 16, 1), torch.float32)
triton_per_fused__softmax_7[grid(64)](buf25, buf28, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
buf29 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf28, (4, 16, 16), (256, 1, 16), 0),
out=buf29)
buf30 = buf28
del buf28
extern_kernels.bmm(reinterpret_tensor(buf2, (4, 16, 4), (64, 1, 16),
0), reinterpret_tensor(buf2, (4, 4, 16), (64, 16, 1), 0), out=buf30
)
buf33 = buf25
del buf25
triton_per_fused__softmax_7[grid(64)](buf30, buf33, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
buf34 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf2, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf33, (4, 16, 16), (256, 1, 16), 0),
out=buf34)
buf36 = buf22
del buf22
triton_poi_fused_sub_4[grid(64)](buf5, buf36, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf37 = buf5
del buf5
triton_poi_fused__softmax_5[grid(64)](buf36, buf37, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf38 = buf36
del buf36
triton_poi_fused__softmax_6[grid(64)](buf37, buf38, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf37
buf39 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf38, reinterpret_tensor(buf4, (4, 4, 16), (64,
16, 1), 0), out=buf39)
buf40 = buf38
del buf38
triton_poi_fused_sub_4[grid(64)](buf7, buf40, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf41 = buf7
del buf7
triton_poi_fused__softmax_5[grid(64)](buf40, buf41, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf42 = buf40
del buf40
triton_poi_fused__softmax_6[grid(64)](buf41, buf42, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf41
buf43 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf42, reinterpret_tensor(buf6, (4, 4, 16), (64,
16, 1), 0), out=buf43)
buf45 = buf33
del buf33
extern_kernels.bmm(reinterpret_tensor(buf4, (4, 16, 4), (64, 1, 16),
0), reinterpret_tensor(buf4, (4, 4, 16), (64, 16, 1), 0), out=buf45
)
buf48 = buf30
del buf30
triton_per_fused__softmax_7[grid(64)](buf45, buf48, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
buf49 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf4, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf48, (4, 16, 16), (256, 1, 16), 0),
out=buf49)
buf50 = buf48
del buf48
extern_kernels.bmm(reinterpret_tensor(buf6, (4, 16, 4), (64, 1, 16),
0), reinterpret_tensor(buf6, (4, 4, 16), (64, 16, 1), 0), out=buf50
)
buf53 = buf45
del buf45
triton_per_fused__softmax_7[grid(64)](buf50, buf53, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
buf54 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf6, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf53, (4, 16, 16), (256, 1, 16), 0),
out=buf54)
buf56 = buf42
del buf42
triton_poi_fused_sub_4[grid(64)](buf9, buf56, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf57 = buf9
del buf9
triton_poi_fused__softmax_5[grid(64)](buf56, buf57, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf58 = buf56
del buf56
triton_poi_fused__softmax_6[grid(64)](buf57, buf58, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf57
buf59 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf58, reinterpret_tensor(buf8, (4, 4, 16), (64,
16, 1), 0), out=buf59)
buf60 = buf58
del buf58
triton_poi_fused_sub_4[grid(64)](buf11, buf60, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf61 = buf11
del buf11
triton_poi_fused__softmax_5[grid(64)](buf60, buf61, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf62 = buf60
del buf60
triton_poi_fused__softmax_6[grid(64)](buf61, buf62, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf61
buf63 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf62, reinterpret_tensor(buf10, (4, 4, 16), (64,
16, 1), 0), out=buf63)
buf65 = buf53
del buf53
extern_kernels.bmm(reinterpret_tensor(buf8, (4, 16, 4), (64, 1, 16),
0), reinterpret_tensor(buf8, (4, 4, 16), (64, 16, 1), 0), out=buf65
)
buf68 = buf50
del buf50
triton_per_fused__softmax_7[grid(64)](buf65, buf68, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
buf69 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf8, (4, 4, 16), (64, 16, 1),
0), reinterpret_tensor(buf68, (4, 16, 16), (256, 1, 16), 0),
out=buf69)
buf70 = buf68
del buf68
extern_kernels.bmm(reinterpret_tensor(buf10, (4, 16, 4), (64, 1, 16
), 0), reinterpret_tensor(buf10, (4, 4, 16), (64, 16, 1), 0),
out=buf70)
buf73 = buf65
del buf65
triton_per_fused__softmax_7[grid(64)](buf70, buf73, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
buf74 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf10, (4, 4, 16), (64, 16, 1
), 0), reinterpret_tensor(buf73, (4, 16, 16), (256, 1, 16), 0),
out=buf74)
buf76 = buf62
del buf62
triton_poi_fused_sub_4[grid(64)](buf13, buf76, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf77 = buf13
del buf13
triton_poi_fused__softmax_5[grid(64)](buf76, buf77, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf78 = buf76
del buf76
triton_poi_fused__softmax_6[grid(64)](buf77, buf78, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf77
buf79 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf78, reinterpret_tensor(buf12, (4, 4, 16), (64,
16, 1), 0), out=buf79)
buf80 = buf78
del buf78
triton_poi_fused_sub_4[grid(64)](buf15, buf80, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf81 = buf15
del buf15
triton_poi_fused__softmax_5[grid(64)](buf80, buf81, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf82 = buf80
del buf80
triton_poi_fused__softmax_6[grid(64)](buf81, buf82, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf81
buf83 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(buf82, reinterpret_tensor(buf14, (4, 4, 16), (64,
16, 1), 0), out=buf83)
del buf82
buf85 = buf73
del buf73
extern_kernels.bmm(reinterpret_tensor(buf12, (4, 16, 4), (64, 1, 16
), 0), reinterpret_tensor(buf12, (4, 4, 16), (64, 16, 1), 0),
out=buf85)
buf88 = buf70
del buf70
triton_per_fused__softmax_7[grid(64)](buf85, buf88, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
buf89 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf12, (4, 4, 16), (64, 16, 1
), 0), reinterpret_tensor(buf88, (4, 16, 16), (256, 1, 16), 0),
out=buf89)
buf90 = buf88
del buf88
extern_kernels.bmm(reinterpret_tensor(buf14, (4, 16, 4), (64, 1, 16
), 0), reinterpret_tensor(buf14, (4, 4, 16), (64, 16, 1), 0),
out=buf90)
buf93 = buf85
del buf85
triton_per_fused__softmax_7[grid(64)](buf90, buf93, 64, 16, XBLOCK=
8, num_warps=2, num_stages=1)
del buf90
buf94 = empty_strided_cuda((4, 4, 16), (64, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf14, (4, 4, 16), (64, 16, 1
), 0), reinterpret_tensor(buf93, (4, 16, 16), (256, 1, 16), 0),
out=buf94)
del buf93
buf24 = empty_strided_cuda((), (), torch.float32)
buf96 = buf24
del buf24
triton_per_fused_abs_add_mean_mul_sub_8[grid(1)](buf96, buf19, buf0,
buf23, buf2, buf29, buf34, buf39, buf4, buf43, buf6, buf49,
buf54, buf59, buf8, buf63, buf10, buf69, buf74, buf79, buf12,
buf83, buf14, buf89, buf94, 1, 256, num_warps=2, num_stages=1)
del buf0
del buf10
del buf12
del buf14
del buf19
del buf2
del buf23
del buf29
del buf34
del buf39
del buf4
del buf43
del buf49
del buf54
del buf59
del buf6
del buf63
del buf69
del buf74
del buf79
del buf8
del buf83
del buf89
del buf94
return buf96,
class SCRMNew(nn.Module):
"""
spatial & channel wise relation loss
"""
def __init__(self, gamma=0.1):
super(SCRMNew, self).__init__()
self.softmax = nn.Softmax(dim=-1)
self.gamma = gamma
def spatial_wise(self, x):
m_batchsize, C, height, width = x.size()
proj_query = x.view(m_batchsize, -1, width * height).permute(0, 2, 1)
proj_key = x.view(m_batchsize, -1, width * height)
energy = torch.bmm(proj_query, proj_key)
attention = self.softmax(energy)
proj_value = x.view(m_batchsize, -1, width * height)
out = torch.bmm(proj_value, attention.permute(0, 2, 1))
out = out.view(m_batchsize, C, height, width)
out = self.gamma * out + x
return out
def channel_wise(self, x):
m_batchsize, C, height, width = x.size()
proj_query = x.view(m_batchsize, C, -1)
proj_key = x.view(m_batchsize, C, -1).permute(0, 2, 1)
energy = torch.bmm(proj_query, proj_key)
energy_new = torch.max(energy, -1, keepdim=True)[0].expand_as(energy
) - energy
attention = self.softmax(energy_new)
proj_value = x.view(m_batchsize, C, -1)
out = torch.bmm(attention, proj_value)
out = out.view(m_batchsize, C, height, width)
out = self.gamma * out + x
return out
def cal_loss(self, f_s, f_t):
f_s = F.normalize(f_s, dim=1)
f_t = F.normalize(f_t, dim=1)
sa_loss = F.l1_loss(self.spatial_wise(f_s), self.spatial_wise(f_t))
ca_loss = F.l1_loss(self.channel_wise(f_s), self.channel_wise(f_t))
return ca_loss + sa_loss
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
|
Tiamat-Tech/ZAQ-code
|
SCRM
| false | 14,538 |
[
"MIT"
] | 55 |
e7e9f55791e36c6784d58c356d3ced76a7583369
|
https://github.com/Tiamat-Tech/ZAQ-code/tree/e7e9f55791e36c6784d58c356d3ced76a7583369
|
LFF
|
# AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/fv/cfvs5hzhw7h4caarcomu5bs4odfeh464zxuws45ipatw3qjtn3dn.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.sin]
# Source node to ATen node mapping:
# x => convolution
# x_1 => sin
# Graph fragment:
# %convolution : [num_users=2] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %sin : [num_users=1] = call_function[target=torch.ops.aten.sin.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_sin_0 = async_compile.triton('triton_poi_fused_convolution_sin_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[65536],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_sin_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_sin_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 65536
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 4096) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl_math.sin(tmp2)
tl.store(in_out_ptr0 + (x3), tmp2, None)
tl.store(out_ptr0 + (x3), tmp3, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 2, 1, 1), (2, 1, 1, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 2, 64, 64), (8192, 4096, 64, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 64, 64), (16384, 4096, 64, 1))
buf1 = buf0; del buf0 # reuse
buf2 = empty_strided_cuda((4, 4, 64, 64), (16384, 4096, 64, 1), torch.float32)
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.sin]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_sin_0.run(buf1, primals_2, buf2, 65536, grid=grid(65536), stream=stream0)
del primals_2
return (buf2, primals_1, primals_3, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 2, 1, 1), (2, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 2, 64, 64), (8192, 4096, 64, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
import numpy as np
from torch import nn
class SinActivation(nn.Module):
def __init__(self):
super(SinActivation, self).__init__()
def forward(self, x):
return torch.sin(x)
class ConLinear(nn.Module):
def __init__(self, ch_in, ch_out, is_first=False, bias=True):
super(ConLinear, self).__init__()
self.conv = nn.Conv2d(ch_in, ch_out, kernel_size=1, padding=0, bias
=bias)
if is_first:
nn.init.uniform_(self.conv.weight, -np.sqrt(9 / ch_in), np.sqrt
(9 / ch_in))
else:
nn.init.uniform_(self.conv.weight, -np.sqrt(3 / ch_in), np.sqrt
(3 / ch_in))
def forward(self, x):
return self.conv(x)
class LFF(nn.Module):
def __init__(self, hidden_size):
super(LFF, self).__init__()
self.ffm = ConLinear(2, hidden_size, is_first=True)
self.activation = SinActivation()
def forward(self, x):
x = self.ffm(x)
x = self.activation(x)
return x
def get_inputs():
return [torch.rand([4, 2, 64, 64])]
def get_init_inputs():
return [[], {'hidden_size': 4}]
|
import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import numpy as np
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_convolution_sin_0(in_out_ptr0, in_ptr0, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 4096 % 4
tmp0 = tl.load(in_out_ptr0 + x3, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl_math.sin(tmp2)
tl.store(in_out_ptr0 + x3, tmp2, None)
tl.store(out_ptr0 + x3, tmp3, None)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 2, 1, 1), (2, 1, 1, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 2, 64, 64), (8192, 4096, 64, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 64, 64), (16384, 4096, 64, 1))
buf1 = buf0
del buf0
buf2 = empty_strided_cuda((4, 4, 64, 64), (16384, 4096, 64, 1),
torch.float32)
get_raw_stream(0)
triton_poi_fused_convolution_sin_0[grid(65536)](buf1, primals_2,
buf2, 65536, XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
return buf2, primals_1, primals_3, buf1
class SinActivation(nn.Module):
def __init__(self):
super(SinActivation, self).__init__()
def forward(self, x):
return torch.sin(x)
class ConLinear(nn.Module):
def __init__(self, ch_in, ch_out, is_first=False, bias=True):
super(ConLinear, self).__init__()
self.conv = nn.Conv2d(ch_in, ch_out, kernel_size=1, padding=0, bias
=bias)
if is_first:
nn.init.uniform_(self.conv.weight, -np.sqrt(9 / ch_in), np.sqrt
(9 / ch_in))
else:
nn.init.uniform_(self.conv.weight, -np.sqrt(3 / ch_in), np.sqrt
(3 / ch_in))
def forward(self, x):
return self.conv(x)
class LFFNew(nn.Module):
def __init__(self, hidden_size):
super(LFFNew, self).__init__()
self.ffm = ConLinear(2, hidden_size, is_first=True)
self.activation = SinActivation()
def forward(self, input_0):
primals_1 = self.ffm.conv.weight
primals_2 = self.ffm.conv.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
|
Ugness/CIPS_SR
|
LFF
| false | 14,539 |
[
"MIT"
] | 172 |
abce872f5bc1b84afb9634a7dd1991e8c74d7616
|
https://github.com/Ugness/CIPS_SR/tree/abce872f5bc1b84afb9634a7dd1991e8c74d7616
|
Clipping
|
# AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_0/inductor_cache/n5/cn5nq6vguoqj455clqkqg7qbojsmjkdk3n4drzf7gkjvvztbyszr.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.minimum, aten.maximum]
# Source node to ATen node mapping:
# x => minimum
# x_1 => maximum
# Graph fragment:
# %minimum : [num_users=1] = call_function[target=torch.ops.aten.minimum.default](args = (%arg1_1, %arg0_1), kwargs = {})
# %maximum : [num_users=1] = call_function[target=torch.ops.aten.maximum.default](args = (%minimum, %arg2_1), kwargs = {})
triton_poi_fused_maximum_minimum_0 = async_compile.triton('triton_poi_fused_maximum_minimum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_maximum_minimum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_maximum_minimum_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask)
tmp3 = tl.load(in_ptr2 + (x0), xmask)
tmp2 = triton_helpers.minimum(tmp0, tmp1)
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.minimum, aten.maximum]
stream0 = get_raw_stream(0)
triton_poi_fused_maximum_minimum_0.run(arg1_1, arg0_1, arg2_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
del arg1_1
del arg2_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
|
import torch
from torch import nn
class Clipping(nn.Module):
def __init__(self):
"""
This module perform element-wise clipping.
"""
super(Clipping, self).__init__()
def forward(self, x, max_value, min_value):
"""
The forward function of the clipping module
:param x: Input tensor to be clipped
:param max_value: The maximal value of the tensor after clipping
:param min_value: The minimal value of the tensor after clipping
:return: A clipped tensor
"""
x = torch.min(x, max_value)
x = torch.max(x, min_value)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
|
import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_maximum_minimum_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask)
tmp3 = tl.load(in_ptr2 + x0, xmask)
tmp2 = triton_helpers.minimum(tmp0, tmp1)
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tl.store(out_ptr0 + x0, tmp4, xmask)
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_maximum_minimum_0[grid(256)](arg1_1, arg0_1,
arg2_1, buf0, 256, XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
del arg1_1
del arg2_1
return buf0,
class ClippingNew(nn.Module):
def __init__(self):
"""
This module perform element-wise clipping.
"""
super(ClippingNew, self).__init__()
def forward(self, input_0, input_1, input_2):
arg0_1 = input_0
arg1_1 = input_1
arg2_1 = input_2
output = call([arg0_1, arg1_1, arg2_1])
return output[0]
|
UniSerj/ai-research
|
Clipping
| false | 14,540 |
[
"Apache-2.0"
] | 46 |
79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
https://github.com/UniSerj/ai-research/tree/79f0093c93408cc5dd7d3f56aafd7dc1f901421c
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.