diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2025-05-12 15:09:33 +0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-05-12 15:09:33 +0300 |
| commit | 627f40643759ef28369ca6db14464299ad0e628e (patch) | |
| tree | a7be8eb6b3effeb72652c7de574f992e9e903253 | |
| parent | 1d2da7feaee3e4dd1b78fb4108988c977b47e266 (diff) | |
Fix new CUDA FA on Touring (#413)
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
| -rw-r--r-- | ggml/src/ggml-cuda/fattn-new-mma.cu | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/ggml/src/ggml-cuda/fattn-new-mma.cu b/ggml/src/ggml-cuda/fattn-new-mma.cu index 630baf33..5aeca3c4 100644 --- a/ggml/src/ggml-cuda/fattn-new-mma.cu +++ b/ggml/src/ggml-cuda/fattn-new-mma.cu @@ -273,11 +273,11 @@ struct fattn_mma_f16_config<576, 512> { } static constexpr __device__ int get_nbatch_V2_device(int ncols) { -#if __CUDA_ARCH__ == GML_CUDA_CC_TURING +#if __CUDA_ARCH__ == CC_TURING return ncols <= 16 ? 64 : 128; #else return ncols <= 16 ? 256 : 128; -#endif // __CUDA_ARCH__ == GML_CUDA_CC_TURING +#endif // __CUDA_ARCH__ == CC_TURING } static int get_nbatch_combine_host(const int /*cc*/, const int /*ncols*/) { |