Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

CUDA: refactor mma data loading for AMD (#22051)

Browse Source 
* CUDA: refactor mma data loading for AMD

* fix CDNA MMQ occupancy

* fix CDNA3 mma

* fix RDNA3 compile
This commit is contained in:
Johannes Gäßler
2026-04-19 18:26:59 +02:00
committed by GitHub
parent d5b780a676
commit 4eac5b4509
4 changed files with 112 additions and 395 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cuda/common.cuh
-4
View File
@@ -269,10 +269,6 @@ static const char * cu_get_error_str(CUresult err) {
#define FLASH_ATTN_AVAILABLE
#endif // !defined(GGML_CUDA_NO_FA) && !(defined(GGML_USE_MUSA) && __MUSA_ARCH__ < 220)
#if defined(TURING_MMA_AVAILABLE)
#define LDMATRIX_TRANS_AVAILABLE
#endif // defined(TURING_MMA_AVAILABLE)
static bool fp16_available(const int cc) {
return ggml_cuda_highest_compiled_arch(cc) >= GGML_CUDA_CC_PASCAL ||
(GGML_CUDA_CC_IS_MTHREADS(cc) && cc >= GGML_CUDA_CC_PH1);