llama.cpp/ggml/src/ggml-cuda/pad_reflect_1d.cuh


1
2
3
4
5

#include "common.cuh"

#define CUDA_PAD_REFLECT_1D_BLOCK_SIZE 256

void ggml_cuda_op_pad_reflect_1d(ggml_backend_cuda_context & ctx, ggml_tensor * dst);