ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-12 10:08:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove unnecessary prefetch and block_gemm

Browse Source
This commit is contained in:
Aleksander Dudek
2025-06-17 05:22:55 -05:00
parent 6e5f8b05ca
commit 7ec46a4650
1 changed files with 0 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
include/ck_tile/ops/gemm/pipeline/gemm_pipeline_ag_bg_cr_comp_v5.hpp
View File

@@ -548,9 +548,6 @@ struct GemmPipelineAgBgCrCompV5 : public BaseGemmPipelineAgBgCrCompV5<Problem>
block_sync_lds();
block_gemm(c_block_tile, a_lds_tile, b_lds_tile);
// Local prefetch 3
PipelineImplBase::LocalPrefetch(a_lds_tile, a_lds_gemm_window_cap);
PipelineImplBase::LocalPrefetch(b_lds_tile, b_lds_gemm_window_cap);
HotLoopScheduler();
};
@@ -569,13 +566,6 @@ struct GemmPipelineAgBgCrCompV5 : public BaseGemmPipelineAgBgCrCompV5<Problem>
std::ignore = b_copy_lds_window_cap;
block_gemm(c_block_tile, a_lds_tile, b_lds_tile);
// Local prefetch 4
PipelineImplBase::LocalPrefetch(a_lds_tile, a_lds_gemm_window_cap);
PipelineImplBase::LocalPrefetch(b_lds_tile, b_lds_gemm_window_cap);
// TODO verify
block_gemm(c_block_tile, a_lds_tile, b_lds_tile);
HotLoopScheduler();
};