ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-20 21:09:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge commit '0f8e33f81120e5734ef47a6a169ad85c6560cbd8' into develop

Browse Source
This commit is contained in:
assistant-librarian[bot]
2025-09-09 03:25:20 +00:00
parent e8724677d6
commit e702cfd808
131 changed files with 8731 additions and 5329 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops.hpp
View File

@@ -108,10 +108,8 @@ struct BlockwiseGemmXdlops_pipeline_v4
using ThisThreadBlock = ThisThreadBlock<BlockSize>;
static constexpr index_t MWaves = MPerBlock / (MRepeat * MPerXDL);
static constexpr index_t NWaves = NPerBlock / (NRepeat * NPerXDL);
static_assert(MWaves > 0);
static_assert(NWaves > 0);
static constexpr index_t MWaves = MPerBlock / (MRepeat * MPerXDL);
static constexpr index_t NWaves = NPerBlock / (NRepeat * NPerXDL);
static constexpr index_t WaveSize = BlockSize / MWaves / NWaves;
static constexpr index_t A_K0 = ATileDesc{}.GetLength(I0);

6
include/ck/tensor_operation/gpu/block/blockwise_gemm_smfmac_xdlops.hpp
View File

@@ -49,10 +49,8 @@ struct BlockwiseGemmXdlops_k0mk1_k0nk1_m0n0m1n1m2m3m4n2_v1
using ThisThreadBlock = ThisThreadBlock<BlockSize>;
static constexpr index_t MWaves = MPerBlock / (MRepeat * MPerXDL);
static constexpr index_t NWaves = NPerBlock / (NRepeat * NPerXDL);
static_assert(MWaves > 0);
static_assert(NWaves > 0);
static constexpr index_t MWaves = MPerBlock / (MRepeat * MPerXDL);
static constexpr index_t NWaves = NPerBlock / (NRepeat * NPerXDL);
static constexpr index_t WaveSize = BlockSize / MWaves / NWaves;
static constexpr index_t MPerBlock = AK0MK1BlockDesc{}.GetLength(I1);