ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-03 05:01:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

Shuffle fix for gfx950 (#3491)

Browse Source 
* solve compiler issue

* solve the gfx950 mfma shuffle regression

* refactor jenkinsfile to handle arch name better

* [CK TILE] set divisor to count of thread along k dimension

* fix the compiler error

* solve degradation

* Finish the multiplies fix

* fix the scales

* solve compilation error

* solve the composes

* solve the error of tile sweeper

* fix the test and example

* fix for gfx950

---------

Co-authored-by: Max Podkorytov <4273004+tenpercent@users.noreply.github.com>
Co-authored-by: illsilin_amdeng <Illia.Silin@amd.com>
Co-authored-by: Cong Ma <congma13@amd.com>
This commit is contained in:
Thomas Ning
2026-01-14 01:21:29 +08:00
committed by GitHub
parent 9908a87c31
commit 00c46785a8
33 changed files with 161 additions and 152 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
include/ck_tile/ops/fmha/pipeline/tile_fmha_shape.hpp
View File

@@ -42,9 +42,9 @@ struct TileFmhaShape
using Gemm1WarpTile = remove_cvref_t<Gemm1WarpTile_>;
static constexpr index_t NumGemm0Warps =
reduce_on_sequence(Gemm0BlockWarps{}, multiplies{}, number<1>{});
reduce_on_sequence(Gemm0BlockWarps{}, multiplies<>{}, number<1>{});
static constexpr index_t NumGemm1Warps =
reduce_on_sequence(Gemm1BlockWarps{}, multiplies{}, number<1>{});
reduce_on_sequence(Gemm1BlockWarps{}, multiplies<>{}, number<1>{});
static_assert(NumGemm1Warps % NumGemm0Warps == 0);
static constexpr index_t NumWarps = max(NumGemm0Warps, NumGemm1Warps);
@@ -95,10 +95,10 @@ struct TileFmhaBwdShape
using Gemm4WarpTile = remove_cvref_t<Gemm4WarpTile_>;
static constexpr index_t NumWarps =
reduce_on_sequence(Gemm0BlockWarps{}, multiplies{}, number<1>{});
reduce_on_sequence(Gemm0BlockWarps{}, multiplies<>{}, number<1>{});
static_assert(NumWarps == reduce_on_sequence(Gemm1BlockWarps{}, multiplies{}, number<1>{}) &&
NumWarps == reduce_on_sequence(Gemm4BlockWarps{}, multiplies{}, number<1>{}));
static_assert(NumWarps == reduce_on_sequence(Gemm1BlockWarps{}, multiplies<>{}, number<1>{}) &&
NumWarps == reduce_on_sequence(Gemm4BlockWarps{}, multiplies<>{}, number<1>{}));
static constexpr index_t kM0 = BlockTile::at(number<0>{}); // tile size along q seqlen
static constexpr index_t kN0 = BlockTile::at(number<1>{}); // tile size along k seqlen