github/ROCm
1
1
Fork 0
mirror of https://github.com/ROCm/ROCm.git synced 2026-04-05 03:01:17 -04:00
Code Issues Packages Projects Releases Wiki Activity

[MFMA] Refactor dot pipeline to reduce code duplication (#400)

Browse Source 
This PR:
- simplifies data types generated by `shared->mfma dot op` layout conversions. Do not pack data types in int32 or int64
- reduce code duplication between fast/normal path
- reduce code duplication between operand A and operand B

Co-authored-by: Shucai Xiao <shucai.xiao@amd.com>
Co-authored-by: Lixun Zhang <lixun.zhang@amd.com>
This commit is contained in:
Alexander Efimov
2023-12-13 22:33:02 +01:00
committed by GitHub
parent 605a90c58e
commit f2afd65e8c
6 changed files with 132 additions and 308 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
include/triton/Dialect/TritonGPU/IR/TritonGPUAttrDefs.td
View File

@@ -855,8 +855,7 @@ section 9.7.13.4.1 for more details.
int bitwidth) const;
#ifdef USE_ROCM
SmallVector<int64_t> getMFMAElemsPerInstr() const;
SmallVector<int64_t> getMFMARep(ArrayRef<int64_t> operandShape,
Type elemType) const;
SmallVector<int64_t> getMFMARep(ArrayRef<int64_t> operandShape) const;
#endif
}];
}