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

move some mmvf to mmf

This commit is contained in:
zhang hui 2025-11-22 15:31:32 +08:00
parent 698c9f2418
commit 99b92bd665
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
ggml/src/ggml-cuda/mmvf.cu
View File

@ -765,7 +765,7 @@ bool ggml_cuda_should_use_mmvf(enum ggml_type type, int cc, const int64_t * src0
return ne11 <= 8;
} else if (GGML_CUDA_CC_IS_AMD(cc)) {
if (fp16_mma_hardware_available(cc)) {
if (GGML_CUDA_CC_IS_RDNA3(cc) || GGML_CUDA_CC_IS_RDNA4(cc)) {
if (GGML_CUDA_CC_IS_RDNA3(cc)) {
return ne11 <= 5;
}
return ne11 <= 2;
@ -788,6 +788,9 @@ bool ggml_cuda_should_use_mmvf(enum ggml_type type, int cc, const int64_t * src0
return ne11 <= 8;
} else if (GGML_CUDA_CC_IS_AMD(cc)) {
if (bf16_mma_hardware_available(cc)) {
if (GGML_CUDA_CC_IS_RDNA4(cc)) {
return ne11 <= 2;
}
return ne11 <= 3;
}
return ne11 <= 8;