ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-15 10:37:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

Use new mfma instructions for FP8 on gfx950 (#2202)

Browse Source 
* Add logic to use new mfma instructions for fp8 bf8

* Fix example_gemm_xdl_fp8_pk_i4_bpreshuffle_v3 on gfx950 and run clang format

* Update include/ck/tensor_operation/gpu/warp/xdlops_gemm.hpp

Co-authored-by: Andriy Roshchenko <107577548+andriy-ca@users.noreply.github.com>

* Fix intrin_mfma f8 calls due to merge mistake

---------

Co-authored-by: Andriy Roshchenko <107577548+andriy-ca@users.noreply.github.com>

[ROCm/composable_kernel commit: f18170064d]
This commit is contained in:
jefyang1
2025-05-19 17:29:51 -07:00
committed by GitHub
parent 8c6eb1c0b8
commit 23a8bed9af
34 changed files with 548 additions and 180 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
example/01_gemm/gemm_xdl_fp8.cpp
View File

@@ -32,6 +32,8 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemm_Xdl_CShuffle
// ######| | | | | | | | | Operation| Operation| Operation| | Stage| | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl| | | | |
// ######| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
< ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, AccDataType, CShuffleDataType, AElementOp, BElementOp, CElementOp, GemmDefault, 1, 256, 256, 128, 64, 16, 16, 32, 32, 4, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 16, 16, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<1, 64, 1, 4>, 8, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB>;
// this instance has been tested working on gfx950
// < ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, AccDataType, CShuffleDataType, AElementOp, BElementOp, CElementOp, GemmDefault, 1, 256, 256, 128, 128, 32, 32, 32, 32, 4, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 16, 16, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<1, 64, 1, 4>, 8, LoopSched, PipelineVer, ComputeTypeA, ComputeTypeB>;
// clang-format on
using ReferenceGemmInstance = ck::tensor_operation::host::