mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-03-01 17:40:25 +00:00
154e0d75fc
* Merging mainline - WIP * Merging mainline - WIP AVX2 and CUDA appear to work. CUDA performance seems slightly (~1-2%) lower as it is so often the case with llama.cpp/ggml after some "improvements" have been made. * Merging mainline - fix Metal * Remove check --------- Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
6 lines
137 B
Plaintext
6 lines
137 B
Plaintext
#include "common.cuh"
|
|
|
|
#define CUDA_POOL2D_BLOCK_SIZE 256
|
|
|
|
void ggml_cuda_op_pool2d(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
|