vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-22 03:10:03 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

disable blackwell mma for now

Browse source
This commit is contained in:
Concedo 2026-01-03 22:45:06 +08:00
parent e4abf643fa
commit 5a505cbc62
1 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
ggml/src/ggml-cuda/common.cuh
View file

@ -256,7 +256,7 @@ static const char * cu_get_error_str(CUresult err) {
#endif // !defined(GGML_USE_HIP) && __CUDA_ARCH__ >= GGML_CUDA_CC_AMPERE
#if !defined(GGML_USE_HIP) && __CUDA_ARCH__ >= GGML_CUDA_CC_BLACKWELL && __CUDA_ARCH__ < GGML_CUDA_CC_RUBIN
# define BLACKWELL_MMA_AVAILABLE
// # define BLACKWELL_MMA_AVAILABLE // kcpp: no blackwell mma for now, requires too high cuda
#endif // !defined(GGML_USE_HIP) && __CUDA_ARCH__ >= GGML_CUDA_CC_BLACKWELL
#if !defined(GGML_USE_HIP) && __CUDA_ARCH__ >= GGML_CUDA_CC_AMPERE
@ -331,6 +331,7 @@ static bool cp_async_available(const int cc) {
}
static bool blackwell_mma_available(const int cc) {
return false; //kcpp: no blackwell mma for now, requires too high cuda
return GGML_CUDA_CC_IS_NVIDIA(cc) && ggml_cuda_highest_compiled_arch(cc) >= GGML_CUDA_CC_BLACKWELL &&
ggml_cuda_highest_compiled_arch(cc) < GGML_CUDA_CC_RUBIN;
}