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

1.99k stars 331 forks source link

Convert non-kernel cuda files to cpp #1322

Closed ksivaman closed 2 weeks ago

ksivaman commented 2 weeks ago

Description

There are cuda files in the paddle and pytorch extensions that neither define nor call any kernels. This PR changes them to pure c++ files.

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)
[ ] New feature (non-breaking change which adds functionality)
[ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
[x] Infra/Build change
[ ] Code refractor

Changes

Change .cu files to .cpp files wherever possible.

Checklist:

[x] I have read and followed the contributing guidelines
[x] The functionality is complete
[ ] I have commented my code, particularly in hard-to-understand areas
[ ] I have made corresponding changes to the documentation
[x] My changes generate no new warnings
[ ] 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

ksivaman commented 2 weeks ago

/te-ci

ksivaman commented 2 weeks ago

Seems reasonable once we fix the compliation error. Is there a particular bug that this is fixing?

No this isn't a bug fix. I also didn't see noticeable performance improvement in compile time, but just for better practice.