ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-30 11:47:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix performance bug of bpreshuffle f8 gemm

Browse Source
This commit is contained in:
aska-0096
2025-05-29 10:02:46 +00:00
parent c3d52993c4
commit d563dac424
2 changed files with 4 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
example/65_gemm_multiply_multiply/gemm_multiply_multiply_xdl_fp8_bpreshuffle.cpp
View File

@@ -139,13 +139,13 @@ using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMultiD_Xdl_CShu
// clang-format off
< Row, Col, DsLayout, ELayout, A0DataType, B0DataType, DsDataType, EDataType, AccDataType, CShuffleDataType,
AElementOp, BElementOp, CDEElementOp, GemmSpec, 256,
128, 128, 128,
256, 256, 128,
16, 16,
16, 16,
8, 2,
16, 4,
S<8, 32, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 16, 16, 0,
S<8, 32, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 16, 16, 0,
1, 1, S<1, 32, 1, 8>, S<8, 8, 1>,
2, 1, S<1, 32, 1, 8>, S<8, 8, 1>,
ck::BlockGemmPipelineScheduler::Intrawave, ck::BlockGemmPipelineVersion::v3, FP8>;
// clang-format on