Python ops support improvements and test fixes #595

Antonyvance · 2025-10-31T04:44:26Z

Improvements

Add Xe12 and 20 support for cpp gen and remove hardcoded values (from PVC 11)
More combinations for generator.py (covers void for elementC)
Python test fixes

Test done:

Python tests
Compilation of generated files
Torch generated file test

Note about cpp file changes:

Gemm_operation3x.hpp (cast to ElementCompute done to avoid compiler errors)
xe_epilogue.cpp . This will be removed from PR if Fix for void ElementC in epilogue. #590 is merged earlier

… into pythonsupport

Pull request Templates

… into pythonsupport

Merge with main

… into pythonsupport

include/cutlass/epilogue/collective/xe_epilogue.hpp

test/python/cutlass/gemm/gemm_bf16_xe20.py

include/cutlass/epilogue/collective/xe_epilogue.hpp

python/cutlass_cppgen/backend/gemm_operation.py

python/cutlass_cppgen/backend/library.py

python/cutlass_cppgen/backend/utils/device.py

python/cutlass_cppgen/library_defaults.py

python/cutlass_library/gemm_operation.py

python/cutlass_library/generator.py

python/cutlass_library/library.py

test/python/cutlass/gemm/gemm_bf16_xe20.py

test/python/cutlass/gemm/gemm_testbed.py

Antonyvance added 30 commits October 15, 2025 22:34

Initial Python op support

f746d14

Unified implementation for PVC and Xe

a824836

Support new arch tags

e4c5b6d

Support for fp8 and int8, added guide

35bf129

Merge branch 'intel:main' into pythonsupport

51716e0

minor fixes

a456c71

Merge branch 'pythonsupport' of https://github.com/Antonyvance/sycl-tla…

97ca0e4

… into pythonsupport

make constants for arch

cc896a4

Fix link issues

1f85328

Fix link issues

f82e742

Pull request Templates

c699f99

Merge pull request #3 from Antonyvance/templates

592cdaf

Pull request Templates

Update INTEL_XE_LIBRARY_GUIDE.md

10f6f0a

examples for cutlass_library

155b766

fix examples names

7fa8d38

Merge branch 'pythonsupport' of https://github.com/Antonyvance/sycl-tla…

0657843

… into pythonsupport

Documentation for cutlass_library

b395409

Copyright changes

7284363

Copyright and documentation changes

ea11aeb

build fixes

bcbb867

Merge pull request #4 from intel/main

55f2f4e

Merge with main

build fixes

27d4045

Merge branch 'pythonsupport' of https://github.com/Antonyvance/sycl-tla…

ac3636d

… into pythonsupport

build fix experiments

3fe3c49

add MKL dependencies

1136e09

add oneMKL to library

506cf13

Fix warnings

657689d

example creates shared library. Making host compilation fixes

c8cd78c

Code clean up

50be956

fix build issues

bf7a04b

Antonyvance added 4 commits October 31, 2025 03:41

Merge branch 'main' of https://github.com/Antonyvance/sycl-tla

1ff2168

Merge branch 'main' into pythonsupport

742bd6c

clean up for merge

377520c

clean up

175f3d7

Antonyvance requested review from jiyang1011 and tdeng5 October 31, 2025 06:02

tdeng5 reviewed Oct 31, 2025

View reviewed changes

include/cutlass/epilogue/collective/xe_epilogue.hpp Show resolved Hide resolved

tdeng5 reviewed Oct 31, 2025

View reviewed changes

test/python/cutlass/gemm/gemm_bf16_xe20.py Show resolved Hide resolved

Copyright update

10eb76f

rolandschulz approved these changes Oct 31, 2025

View reviewed changes

include/cutlass/epilogue/collective/xe_epilogue.hpp Show resolved Hide resolved

include/cutlass/epilogue/collective/xe_epilogue.hpp Outdated Show resolved Hide resolved

include/cutlass/epilogue/collective/xe_epilogue.hpp Outdated Show resolved Hide resolved