ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-24 06:44:36 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix the CK Tile related operators (#2356)

Browse Source 
* fix the flatmm

* Fix the pipeline

* address the comment

[ROCm/composable_kernel commit: 3c4cdfac4f]
This commit is contained in:
Thomas Ning
2025-06-16 17:38:52 -07:00
committed by GitHub
parent 456719c9bc
commit bc6af0fa49
10 changed files with 22 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
example/ck_tile/03_gemm/gemm_basic.cpp
View File

@@ -69,9 +69,12 @@ float gemm(const ck_tile::GemmHostArgs</*NumDTensor = 0*/>& args, const ck_tile:
using GemmEpilogue = ck_tile::CShuffleEpilogue<
ck_tile::CShuffleEpilogueProblem<ADataType,
BDataType,
ck_tile::tuple<>,
AccDataType,
CDataType,
ck_tile::tuple<>,
CLayout,
ck_tile::element_wise::PassThrough,
CodegenPipelineProblem::kBlockSize,
TilePartitioner::MPerBlock,
TilePartitioner::NPerBlock,

2
example/ck_tile/03_gemm/universal_gemm.cpp
View File

@@ -166,7 +166,7 @@ float gemm(const ck_tile::GemmHostArgs</*NumDTensor = 0*/>& args, const ck_tile:
// clear c mem
if(args.k_batch > 1)
hipGetErrorString(hipMemsetAsync(
args.c_ptr, 0, args.M * args.N * sizeof(CDataType), s.stream_id_));
args.e_ptr, 0, args.M * args.N * sizeof(CDataType), s.stream_id_));
};
ave_time = ck_tile::launch_kernel_preprocess(
s,

3
example/ck_tile/18_flatmm/flatmm_basic.cpp
View File

@@ -49,9 +49,12 @@ float flatmm_calc(const ck_tile::FlatmmHostArgs& args, const ck_tile::stream_con
using GemmEpilogue = ck_tile::CShuffleEpilogue<
ck_tile::CShuffleEpilogueProblem<ADataType,
BDataType,
ck_tile::tuple<>,
AccDataType,
CDataType,
ck_tile::tuple<>,
CLayout,
ck_tile::element_wise::PassThrough,
CodegenPipelineProblem::kBlockSize,
TilePartitioner::MPerBlock,
TilePartitioner::NPerBlock,