ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-17 11:30:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge commit 'c7dce2ac29136939b6fe6aabadd026e53dcf35c9' into develop

Browse Source
This commit is contained in:
assistant-librarian[bot]
2025-11-26 02:44:11 +00:00
parent 6d42c1d821
commit 9eb4b35ef6
3 changed files with 74 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
include/ck_tile/ops/flatmm/pipeline/flatmm_pipeline_agmem_bgmem_creg_v1_policy.hpp Executable file → Normal file
View File

@@ -229,13 +229,13 @@ struct UniversalFlatmmPipelineAgBgCrPolicy
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr index_t GetSmemPackA()
{
using A = remove_cvref_t<typename Problem::ADataType>;
using BlockGemm = remove_cvref_t<decltype(GetBlockGemm<Problem>())>;
using A = remove_cvref_t<typename Problem::ADataType>;
using BlockFlatmm = remove_cvref_t<decltype(GetBlockFlatmm<Problem>())>;
constexpr index_t KPack = static_cast<index_t>(BlockGemm::Traits::KPack);
constexpr index_t VecElems = static_cast<index_t>(Problem::VectorLoadSize / sizeof(A));
constexpr index_t KPack = BlockFlatmm::BlockPolicy::WarpGemm::kKPerThread;
constexpr index_t VecElems = Problem::VectorLoadSize / sizeof(A);
return (KPack < VecElems) ? KPack : VecElems;
return min(KPack, VecElems);
}
template <typename Problem>