Steven10429's picture
llama.cpp
61b850a
raw
history blame contribute delete
137 Bytes
#include "common.cuh"
#define CUDA_IM2COL_BLOCK_SIZE 256
void ggml_cuda_op_im2col(ggml_backend_cuda_context & ctx, ggml_tensor * dst);