ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-01 04:07:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add gemm1 v1 to selector

Browse Source
This commit is contained in:
OscarXu
2025-05-15 16:22:45 +08:00
parent c0babbca62
commit bcb4f7e98c
2 changed files with 24 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_selector.hpp
View File

@@ -4,6 +4,7 @@
#pragma once
#include "ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_v1_moe_mx.hpp"
#include "ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_moe_mx.hpp"
namespace ck {
@@ -71,7 +72,28 @@ constexpr auto BlockGemmMXBPreshufflePipeline_Selector()
{
if constexpr(GUFusion)
{
return nullptr;
return return BlockwiseGemmXdlops_pipeline_bpreshuffle_gufusion_v1_moe_mx<
BlkGemmPipeSche,
ThreadBlockSize,
ScaleBlockSize,
ADataType,
AScaleDataType,
BDataType,
BScaleDataType,
ATileDesc,
BTileDesc,
AMmaTileDesc,
BMmaTileDesc,
ABlockTransferSrcScalarPerVector,
BBlockTransferSrcScalarPerVector,
MPerBlock,
NPerBlock,
KPerBlock,
MPerXDL,
NPerXDL,
MRepeat,
NRepeat,
KPack>{};;
}
else
{

2
include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_mx.hpp → include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_b_preshuflle_gufusion_v1_moe_mx.hpp
View File

@@ -58,7 +58,7 @@ template <index_t ThreadBlockSize,
index_t MRepeat, // MXdlPerWave
index_t NRepeat, // NXdlPerWave
index_t KPack>
struct BlockwiseGemmXdlops_pipeline_bpreshuffle_gufusion_v1_mx_tmp<BlockGemmPipelineScheduler::Intrawave,
struct BlockwiseGemmXdlops_pipeline_bpreshuffle_gufusion_v1_moe_mx<BlockGemmPipelineScheduler::Intrawave,
ThreadBlockSize,
ScaleBlockSize,
ADataType,