ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-08 23:38:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'develop' into gptoss_sink

Browse Source
This commit is contained in:
Linjun-AMD
2025-12-25 15:11:19 +08:00
committed by GitHub
parent 4bc109fe40 e08efa551f
commit b86a860f09
20 changed files with 662 additions and 490 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
include/ck_tile/ops/gemm_quant/pipeline/gemm_bquant_pipeline_ag_bg_cr_v3.hpp
View File

@@ -422,7 +422,7 @@ struct BQuantGemmPipelineAgBgCrCompV3 : public BaseGemmPipelineAgBgCrCompV3<Prob
currIdx = (currIdx + 1) % 2;
if constexpr(is_a_col_major)
if constexpr(is_a_col_major && !is_a_load_tr_v())
{
auto a_shuffle_tmp = make_static_distributed_tensor<ADataType>(
Policy::template MakeShuffledARegTileDistribution<Problem>());
@@ -433,7 +433,7 @@ struct BQuantGemmPipelineAgBgCrCompV3 : public BaseGemmPipelineAgBgCrCompV3<Prob
{
Base::LocalPrefill(a_copy_lds_window, a_block_tile, a_element_func);
}
if constexpr(is_b_row_major)
if constexpr(is_b_row_major && !is_b_load_tr_v())
{
// Note: BDataType gets converted during loading from PkInt4
auto b_shuffle_tmp = make_static_distributed_tensor<OverrideBDataType>(