mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-20 21:09:08 +00:00
Merge commit 'bedade257241fef37a28c6e540e73f1c056d27b9' into develop
This commit is contained in:
@@ -789,12 +789,12 @@ struct store_C_col_major<CType, CFragT, 32, 32>
|
||||
CScalarFragT chunks[vectorSize(CFragT{}) / VW];
|
||||
} fragC{cFrag}; // Initialize with input fragment
|
||||
|
||||
*(reinterpret_cast<CScalarFragT*>(output + startOffset)) = fragC.chunks[0];
|
||||
*(reinterpret_cast<CScalarFragT*>(output + startOffset + kMajorOffset)) = fragC.chunks[1];
|
||||
*(reinterpret_cast<CScalarFragT*>(output + startOffset + 2 * kMajorOffset)) =
|
||||
fragC.chunks[2];
|
||||
*(reinterpret_cast<CScalarFragT*>(output + startOffset + 3 * kMajorOffset)) =
|
||||
fragC.chunks[3];
|
||||
CScalarFragT* fragPtr;
|
||||
for(uint32_t idx = 0; idx < vectorSize(CFragT{}) / VW; ++idx)
|
||||
{
|
||||
fragPtr = reinterpret_cast<CScalarFragT*>(output + startOffset + idx * kMajorOffset);
|
||||
*fragPtr = fragC.chunks[idx];
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
|
||||
Reference in New Issue
Block a user