NVIDIA / TransformerEngine

A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit floating point (FP8) precision on Hopper and Ada GPUs, to provide better performance with lower memory utilization in both training and inference.

https://docs.nvidia.com/deeplearning/transformer-engine/user-guide/index.html

Apache License 2.0

1.85k stars 309 forks source link

[C] Allow bias support for sm80/86/89 for cuDNN 9+ #863

Closed cyanguwa closed 4 months ago

cyanguwa commented 4 months ago

Description

This PR opens up the restriction on post_scale_bias support for sm80/86/89 when cuDNN 9.0+ is installed.

Fixes QA tests for test_cp_with_fused_attention on sm80/86/89.

Type of change

[ ] Documentation change (change only to the documentation, either a fix or a new content)
[x ] 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)

Changes

Allows the use of F16_arbitrary_seqlen backend when post_scale_bias is run on sm80/86/89 with cuDNN 9.0+.

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

timmoon10 commented 4 months ago

/te-ci