ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-19 20:40:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add gemm universal bf16 instances (#1484)

Browse Source 
* revert ckprofiler change

* temp save

* Add test and test pass

* test pass

* Fix bug inside rotating buffer when tensor is not packed

* bug fix

* clang format

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>

[ROCm/composable_kernel commit: 5b10dae6a4]
This commit is contained in:
Haocong WANG
2024-09-05 11:58:54 +08:00
committed by GitHub
parent c41f94b84b
commit 4e4514caa8
35 changed files with 1048 additions and 75 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
profiler/src/profile_gemm_universal.cpp
View File

@@ -171,6 +171,14 @@ int profile_gemm_universal(int argc, char* argv[])
{
return profile(BF16{}, BF16{}, BF16{}, F32{}, BF16{}, Row{}, Col{}, Row{});
}
else if(data_type == GemmDataType::BF16_BF16_BF16 && layout == GemmMatrixLayout::KM_NK_MN)
{
return profile(BF16{}, BF16{}, BF16{}, F32{}, BF16{}, Col{}, Col{}, Row{});
}
else if(data_type == GemmDataType::BF16_BF16_BF16 && layout == GemmMatrixLayout::KM_KN_MN)
{
return profile(BF16{}, BF16{}, BF16{}, F32{}, BF16{}, Col{}, Row{}, Row{});
}
else if(data_type == GemmDataType::F8_F8_BF16 && layout == GemmMatrixLayout::MK_KN_MN)
{
return profile(F8{}, F8{}, F8{}, F32{}, BF16{}, Row{}, Row{}, Row{});