ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-05 06:01:23 +00:00
Code Issues Packages Projects Releases Wiki Activity

Support Wave32 in CK_TILE - Part 1 (#2594)

Browse Source 
* Support wave32/wave64 in CK_TILE - Part 1

* remove blocksize in kernel launch

* fix build error

* fix clang format

* fix clang format 2

* fix clang format 3

* fix fmha build error

* fix fmha build 2

* fix fmha build 3

* fix build error 4

* address review comment

* update change log

* replace KernelBlockSize with kBlockSize

* fix CI fail

* fix clang format

* address review comment and rebase code.

* fix universal test fail

---------

Co-authored-by: Lin, Qun <Quentin.Lin+amdeng@amd.com>
Co-authored-by: Thomas Ning <Thomas.Ning@amd.com>
This commit is contained in:
linqunAMD
2025-08-19 01:08:31 +08:00
committed by GitHub
parent 26d3300930
commit 9fcc1ee9fd
113 changed files with 610 additions and 531 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
include/ck_tile/ops/flatmm/pipeline/flatmm_pipeline_agmem_bgmem_creg_v1_policy.hpp
View File

@@ -237,15 +237,16 @@ struct UniversalFlatmmPipelineAgBgCrPolicy
template <typename Problem>
CK_TILE_HOST_DEVICE static constexpr auto GetKBPerLoad()
{
using TileShape = typename Problem::BlockGemmShape;
using TileShape = typename Problem::BlockGemmShape;
constexpr index_t scale = get_warp_size() == 32 ? 2 : 1;
if constexpr(TileShape::WarpTile::at(I1) == 32)
{
return TileShape::WarpTile::at(I2) / 2;
return TileShape::WarpTile::at(I2) * scale / 2;
}
else
{
static_assert(TileShape::WarpTile::at(I1) == 16);
return TileShape::WarpTile::at(I2) / 4;
return TileShape::WarpTile::at(I2) * scale / 4;
}
}