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

HIP: fix RDNA3 FP16/BF16 matrix multiplication (#17817)

This commit is contained in:
Johannes Gäßler 2025-12-06 13:45:36 +01:00 committed by GitHub
parent a28e3c7567
commit f334b79494
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/ggml-cuda/mmf.cu
View File

@ -160,9 +160,9 @@ bool ggml_cuda_should_use_mmf(enum ggml_type type, int cc, int warp_size, const
case GGML_TYPE_F32: case GGML_TYPE_F32:
return ampere_mma_available(cc); return ampere_mma_available(cc);
case GGML_TYPE_F16: case GGML_TYPE_F16:
return volta_mma_available(cc) || turing_mma_available(cc) || amd_wmma_available(cc); return volta_mma_available(cc) || turing_mma_available(cc) || (amd_wmma_available(cc) && GGML_CUDA_CC_IS_RDNA4(cc));
case GGML_TYPE_BF16: case GGML_TYPE_BF16:
return ampere_mma_available(cc) || amd_wmma_available(cc); return ampere_mma_available(cc) || (amd_wmma_available(cc) && GGML_CUDA_CC_IS_RDNA4(cc));
default: default:
return false; return false;
} }