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

[CK TILE][AICK-439] Fix cshuffle epilogue wave per shuffle (#3364)

Browse Source 
* [CK TILE] Fix cshufle epligoue wave per shuffle

* Align shuffle per tile with smem

* fixes

* Fixes for double smem

* fix
This commit is contained in:
Bartłomiej Kocot
2025-12-15 12:59:48 +01:00
committed by GitHub
parent 3143a5a480
commit 3b773109e5
11 changed files with 73 additions and 34 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/ck_tile/ops/gemm_quant/kernel/gemm_quant_kernel.hpp
View File

@@ -1324,7 +1324,7 @@ struct QuantGemmKernel
assert(kargs.k_batch == 1);
if constexpr(GemmPipeline::DoubleSmemBuffer == true)
{
__shared__ char smem_ptr_1[GetSmemSize()];
__shared__ char smem_ptr_1[GemmPipeline::GetSmemSize()];
RunGemm2LDS(a_ptr,
b_ptr,

2
include/ck_tile/ops/gemm_quant/kernel/grouped_gemm_quant_kernel.hpp
View File

@@ -325,7 +325,7 @@ struct QuantGroupedGemmKernel
kQuantType == QuantType::BQuantGrouped)
{
__shared__ char smem_ptr_1[GetSmemSize()];
__shared__ char smem_ptr_1[GemmPipeline::GetSmemSize()];
RunGemmWithPipelineSelection2LDS(a_ptr,
b_ptr,
aq_ptr,