Add Cuda Guard for Forced Align Module

[sabilmakbar]

fixes #3910

PLEASE NOTE THAT THE TORCHAUDIO REPOSITORY IS NO LONGER ACTIVELY MONITORED. You may not get a response. For open discussions, visit https://discuss.pytorch.org/.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/audio/4115

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• Long queue for ROCM runners, also B200 and XPU queueing is observed

✅ No Failures

As of commit 0fd1a6a with merge base 87ff22e ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[meta-cla]

Hi @sabilmakbar!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at [email protected]. Thanks!

[meta-cla]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks!

[sabilmakbar]

It seems this PR is preceded by (or potentially blocked by) #4079. Idk if that PR also solves forced_align ops that doesn't work with cuda device index >0

[pearu]

I confirm that the PR fixes the issue.

We aim at porting torchaudio to stable ABI and there exists a concern that using OptionalCUDAGuard will interfere with the process. On the other hand, even it does, using OptionalCUDAGuard makes total sense and we'll need to tackle any porting issues as they raise.

@NicolasHug what do you think?

[NicolasHug]

This sounds OK but I share your concern about porting this to the stable ABI. Would it be safer and quicker to first do the ABI port and then integrate this fix?

[pearu]

This sounds OK but I share your concern about porting this to the stable ABI. Would it be safer and quicker to first do the ABI port and then integrate this fix?

I think the order does not matter as this PR is just a two-liner one.

To alleviate the stable ABI concern, I think we'll need a CI job that ensures that audio stable ABI porting is functional, that is, audio tests must pass with nightly pytorch while audio itself is built with an older version of pytorch that has sufficiently complete stable ABI support for audio.