ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-20 06:49:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix timing issue in CK_TILE GEMM example (#2940)

Browse Source
This commit is contained in:
Hosang
2025-09-29 18:34:04 -04:00
committed by GitHub
parent bebf0e9d15
commit 28ad8ae5d8
2 changed files with 6 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
example/ck_tile/03_gemm/gemm_splitk_two_stage_invoker.hpp
View File

@@ -231,7 +231,7 @@ struct SplitKTwoStageInvoker
preprocess = clear_gemm_output;
}
return ck_tile::launch_kernel_time_mask(
ave_time = ck_tile::launch_kernel_time_mask(
s,
preprocess,
ck_tile::make_kernel<GemmConfig::kBlockPerCu>(
@@ -245,6 +245,8 @@ struct SplitKTwoStageInvoker
ck_tile::make_tuple(args.N, 1), // Output Stride
input_tensors,
static_cast<CDataType*>(c_ptr)));
return ave_time;
};
const auto RunSplitk = [&](const auto has_hot_loop_, const auto tail_number_) {

4
example/ck_tile/03_gemm/universal_gemm_invoker.hpp
View File

@@ -174,10 +174,12 @@ struct UniversalInvoker
preprocess = clear_gemm_output;
}
return ck_tile::launch_kernel_time_mask(
ave_time = ck_tile::launch_kernel_time_mask(
s,
preprocess,
ck_tile::make_kernel<GemmConfig::kBlockPerCu>(Kernel{}, grids, blocks, 0, kargs));
return ave_time;
};
const auto RunSplitk = [&](const auto has_hot_loop_, const auto tail_number_) {