ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 14:29:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix merge mfma_wmma (part 1) regression (#2749)

Browse Source 
root cause: a typo in GetGfx11InputBlkIdx, const ia added by mistake.
This commit is contained in:
linqunAMD
2025-08-27 13:49:34 +08:00
committed by GitHub
parent 19d5327c45
commit 95e4a4efcb
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/ck/tensor_operation/gpu/warp/xdlops_gemm.hpp
View File

@@ -1937,7 +1937,7 @@ struct XdlopsGemm
template <bool SwizzleA>
__device__ static auto GetGfx11InputBlkIdx()
{
const auto laneId = GetLaneId() % mfma_instr.num_threads_per_blk;
auto laneId = GetLaneId() % mfma_instr.num_threads_per_blk;
if constexpr(SwizzleA)
{
laneId = ((laneId & 1) << 3) | (laneId >> 1);