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

clean up merge conflicts

This commit is contained in:
jiachengjason 2025-11-07 12:35:14 -05:00 committed by jiachengjason
parent 48afe04ca3
commit c770ca2812
1 changed files with 0 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/ggml-cuda/mmq.cuh
View File

@ -3024,13 +3024,9 @@ static __device__ __forceinline__ void mmq_write_back_mma(
const int i0 = (threadIdx.y / ntx) * (ntx*tile_C::I); const int i0 = (threadIdx.y / ntx) * (ntx*tile_C::I);
#if defined(TURING_MMA_AVAILABLE) || defined(AMD_MFMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE) #if defined(TURING_MMA_AVAILABLE) || defined(AMD_MFMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
static_assert(nwarps*tile_C::I == mmq_y, "nwarps*tile_C::I != mmq_y"); static_assert(nwarps*tile_C::I == mmq_y, "nwarps*tile_C::I != mmq_y");
<<<<<<< HEAD
#else #else
GGML_UNUSED(nwarps); GGML_UNUSED(nwarps);
#endif // defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE)
=======
#endif // defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE) #endif // defined(AMD_MFMA_AVAILABLE) || defined(TURING_MMA_AVAILABLE) || defined(AMD_WMMA_AVAILABLE)
>>>>>>> 1912f816 (first commit naive test to enable mmq for RDNA4)
#pragma unroll #pragma unroll
for (int j0 = 0; j0 < mmq_x; j0 += ntx*tile_C::J) { for (int j0 = 0; j0 < mmq_x; j0 += ntx*tile_C::J) {