happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

make tensor core path available for cc 7.5 and above

This commit is contained in:
bssrdf 2025-11-01 14:35:44 -04:00
parent 417cfc3cc6
commit f95664c76c
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-cuda/conv2d-implicit.cu
View File

@ -927,7 +927,7 @@ static void conv2d_implicit_cuda(const float * X_D, const T * K_D, float * Y_D,
static void conv2d_implicit_cuda_f16(ggml_backend_cuda_context & ctx, const float * X_D, const half * K_D, float * Y_D, int cc, const param_t P, cudaStream_t st) {
if (GGML_CUDA_CC_IS_NVIDIA(cc) && ampere_mma_available(cc) && P.c % 8 == 0 && (P.r > 1 || P.s > 1)) {
if (GGML_CUDA_CC_IS_NVIDIA(cc) && turing_mma_available(cc) && P.c % 8 == 0 && (P.r > 1 || P.s > 1)) {
int id = ggml_cuda_get_device();