CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-05 09:36:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA/HIP: optimize mmv paths taken for HIP devices (#14324)

Browse Source 
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
This commit is contained in:
uvos
2025-06-24 01:12:56 +02:00
committed by GitHub
parent ce82bd0117
commit 0142961a2e
2 changed files with 23 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
ggml/src/ggml-cuda/common.cuh
View File

@@ -263,7 +263,11 @@ static bool fp16_mma_hardware_available(const int cc) {
}
static bool bf16_mma_hardware_available(const int cc) {
return GGML_CUDA_CC_IS_NVIDIA(cc) && cc >= GGML_CUDA_CC_AMPERE;
return (GGML_CUDA_CC_IS_NVIDIA(cc) && cc >= GGML_CUDA_CC_AMPERE) || GGML_CUDA_CC_IS_CDNA(cc) || cc >= GGML_CUDA_CC_RDNA3;
}
static bool fp32_mma_hardware_available(const int cc) {
return GGML_CUDA_CC_IS_CDNA(cc);
}
// Volta technically had FP16 tensor cores but they work very differently compared to Turing and later.