TransformerEngine
TransformerEngine copied to clipboard
Add num_splits support for FA3 backend
Description
This is a continuation of the efforts in #2357.
FA3 allows users to use the num_splits option to control the number of kernels launched for attention, which could lead to better performance. This PR integrates that support to TE's FlashAttention backend and enables it for non-CP cases.
Type of change
- [ ] Documentation change (change only to the documentation, either a fix or a new content)
- [ ] Bug fix (non-breaking change which fixes an issue)
- [x] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
- [ ] Infra/Build change
- [ ] Code refactoring
Changes
Please list the changes introduced in this PR:
- See description above.
Checklist:
- [x] I have read and followed the contributing guidelines
- [x] The functionality is complete
- [x] I have commented my code, particularly in hard-to-understand areas
- [x] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] I have added tests that prove my fix is effective or that my feature works
- [x] New and existing unit tests pass locally with my changes