[DCP] Add DefaultStager example to distributed async checkpoint recipe

[niyunsheng]

Fixes #3710

Description

This PR updates the distributed_async_checkpoint_recipe to include the DefaultStager functionality introduced in PyTorch 2.9.

Motivation:
In large-scale training, even with standard async_save, the initial memory copy (Staging phase, GPU -> CPU) occurs on the main thread. This blocks the training loop. This PR introduces DefaultStager, which offloads this copy to a background thread, enabling full computation-communication overlap.

Key Changes:

1. New Section: Added "Fully Asynchronous Staging with DefaultStager" to the recipe.
2. Version Note: Added .. versionadded:: 2.9 to indicate version requirements.
3. Advanced Example: Provided a code example demonstrating how to overlap the D2H copy with the entire Forward and Backward pass:
   • Check staging_completion after backward but before optimizer.step() to ensure data consistency while maximizing parallel execution.
   • Check upload_completion before the next save to manage memory backpressure.
4. Authors: Added myself to the author list.

Checklist

• The issue that is being fixed is referred in the description (see above "Fixes #ISSUE_NUMBER")
• Only one issue is addressed in this pull request
• Labels from the issue that this PR is fixing are added to this pull request
• No unnecessary issues are included into this pull request.

cc @LucasLLC @MeetVadakkanchery @mhorowitz @pradeepfn @ekr0 @haochengsong @Saiteja64

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3711

• 📄 Preview Python docs built from this PR

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

❌ 17 Cancelled Jobs

As of commit c7a7177 with merge base f3d9750 ():

CANCELLED JOBS - The following jobs were cancelled. Please retry:

• Build tutorials / build / pytorch_tutorial_build_manager (gh)
• Build tutorials / build / pytorch_tutorial_build_worker (1, 15, linux.g5.12xlarge.nvidia.gpu) (gh)
• Build tutorials / build / pytorch_tutorial_build_worker (10, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
• Build tutorials / build / pytorch_tutorial_build_worker (11, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (12, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (13, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (14, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (15, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (2, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (3, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (4, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (5, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (6, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (7, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (8, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Build tutorials / build / pytorch_tutorial_build_worker (9, 15, linux.g5.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• Lintrunner / lintrunner (gh)
  ##[error]The operation was canceled.

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

[meta-cla]

Hi @niyunsheng!

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!

[niyunsheng]

Hi @LucasLLC, gentle ping on this.

Just to clarify context: This tutorial covers an existing feature (DefaultStager) and is independent of my other open PRs. It does not require any upstream changes.

It has been passing CI for two weeks. Could you please take a look when you have a moment? Thanks!

[AlannaBurke]

Preview link: https://docs-preview.pytorch.org/pytorch/tutorials/3711/recipes/distributed_async_checkpoint_recipe.html

[AlannaBurke]

Just a few grammar and minor changes. Otherwise this looks good to me.

[niyunsheng]

Hi @AlannaBurke , thanks for the review!

I've applied your grammar and other suggestions.

Extra Fix: While previewing the file locally (make html-noplot), I noticed another small display issue in this file that was missing from the review, so I went ahead and fixed that as well.

The file should be perfect now. Ready for another look!

[LucasLLC]

Tysm for improving the tutorial! Apologies this was in limbo for so long

[niyunsheng]

Thanks @LucasLLC for the approval!

@AlannaBurke I've pushed a fix for the remaining lint/whitespace errors. Since I'm an external contributor, could you please approve the CI workflows to run?

Once the checks are green, this should be ready to land. Thanks!

[niyunsheng]

"Hi @AlannaBurke @LucasLLC,

I've just synced this branch with main to ensure everything is up-to-date.

Could you please help approve the CI workflows to run? Once the checks pass, this should be ready to merge. Thanks!

[niyunsheng]

Hi @AlannaBurke,

Just a gentle bump on this. Since this PR has been open for a month and is now approved (thanks!), I'd really love to get it across the finish line.

The CI workflows are currently stuck in "awaiting approval" after the recent sync with main. Could you please help trigger them so we can merge?

Thanks!

[niyunsheng]

Hi @AlannaBurke and @svekars,

I've just updated the branch to resolve the out-of-date status.

Since this PR has been approved for two weeks, I'd really appreciate it if either of you could help trigger the final CI.

With the Lunar New Year holiday coming up in two weeks, I'd love to see this land beforehand so we don't carry it over.

Thanks for your help!