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

Patch perf regression for mmq kernels in ROCm 

recover performance regression for https://github.com/ggml-org/llama.cpp/issues/17917
This commit is contained in:
Jiacheng (Jason) Chen 2025-12-28 18:38:25 -05:00 committed by GitHub
parent 382808c14b
commit b75c391276
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 7 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
ggml/src/ggml-cuda/mmq.cu
View File

@ -333,7 +333,13 @@ bool ggml_cuda_should_use_mmq(enum ggml_type type, int cc, int64_t ne11, int64_t
}
if (amd_wmma_available(cc)) {
return true;
if (ne11 <= 128 || type == GGML_TYPE_Q4_0 || type == GGML_TYPE_Q4_1 || type == GGML_TYPE_Q5_0 || type == GGML_TYPE_Q5_1) {
return true;
}
if (ne11 <= 256 && (type == GGML_TYPE_Q4_K || type == GGML_TYPE_Q5_K)) {
return true;
}
return false;
}
return (!GGML_CUDA_CC_IS_CDNA(cc)) || ne11 < MMQ_DP4A_MAX_BATCH_SIZE;