leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 29 Packages Projects Releases Wiki Activity

Hopper Grouped GEMM support for FP8 Accum (#2123)

Browse Source 
* Add support for fp8accum, with profiler extension

* Update .gitignore

* contri

---------

Co-authored-by: Haicheng Wu <haichengw@nvidia.com>
This commit is contained in:
ANIKET SHIVAM
2025-02-20 21:55:26 -05:00
committed by GitHub
co-authored by Haicheng Wu
parent b84e9802d8
commit 9b3772dfa6
9 changed files with 914 additions and 71 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/library/src/grouped_gemm_operation_3x.hpp
+2 -3
View File
@@ -204,9 +204,6 @@ protected:
fusion_args.beta_ptr = nullptr;
fusion_args.alpha_ptr_array = nullptr;
fusion_args.beta_ptr_array = nullptr;
// Single alpha and beta for all groups
fusion_args.dAlpha = {cute::_0{}, cute::_0{}, 0};
fusion_args.dBeta = {cute::_0{}, cute::_0{}, 0};
return Status::kSuccess;
}
@@ -215,6 +212,8 @@ protected:
fusion_args.beta = 0;
fusion_args.alpha_ptr = static_cast<ElementCompute const*>(arguments.alpha);
fusion_args.beta_ptr = static_cast<ElementCompute const*>(arguments.beta);
fusion_args.alpha_ptr_array = nullptr;
fusion_args.beta_ptr_array = nullptr;
return Status::kSuccess;
}
else {