Skip to content

ASR Decoding: allow fallback to CUDA graphs without while loops#15173

Merged
artbataev merged 2 commits intomainfrom
vbataev/fallback_for_cuda_graphs
Dec 10, 2025
Merged

ASR Decoding: allow fallback to CUDA graphs without while loops#15173
artbataev merged 2 commits intomainfrom
vbataev/fallback_for_cuda_graphs

Conversation

@artbataev
Copy link
Copy Markdown
Collaborator

Important

The Update branch button must only be pressed in very rare occassions.
An outdated branch is never blocking the merge of a PR.
Please reach out to the automation team before pressing that button.

What does this PR do ?

Due to multiple failures of ASR decoding with full CUDA graphs (with while loops), allow fallback to native PyTorch CUDA graphs without while loops.
The fallback option was implemented previously, but used only when cuda-python was unavailable. This PR also enables it if full CUDA graph compilation failed.

For testing purposes, fallback is disallowed in forced mode.

E.g., see issue: #15145

Collection: [ASR]

Changelog

  • Add specific line by line info of high level changes in this PR.

Usage

  • You can potentially add a usage example below
# Add a code snippet demonstrating how to use this 

GitHub Actions CI

The Jenkins CI system has been replaced by GitHub Actions self-hosted runners.

The GitHub Actions CI will run automatically when the "Run CICD" label is added to the PR.
To re-run CI remove and add the label again.
To run CI on an untrusted fork, a NeMo user with write access must first click "Approve and run".

Before your PR is "Ready for review"

Pre checks:

  • Make sure you read and followed Contributor guidelines
  • Did you write any new necessary tests?
  • Did you add or update any necessary documentation?
  • Does the PR affect components that are optional to install? (Ex: Numba, Pynini, Apex etc)
    • Reviewer: Does the PR have correct import guards for all optional libraries?

PR Type:

  • New Feature
  • Bugfix
  • Documentation

If you haven't finished some of the above items you can still open "Draft" PR.

Who can review?

Anyone in the community is free to review the PR once the checks have passed.
Contributor guidelines contains specific people who can review PRs to various areas.

Additional Information

  • Related to # (issue)

… if full graph cannot be compiled

Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
@github-actions
Copy link
Copy Markdown
Contributor

[🤖]: Hi @artbataev 👋,

We wanted to let you know that a CICD pipeline for this PR just finished successfully.

So it might be time to merge this PR or get some approvals.

//cc @chtruong814 @ko3n1g @pablo-garay @thomasdhc

@artbataev artbataev merged commit 7fc6fd0 into main Dec 10, 2025
138 checks passed
@artbataev artbataev deleted the vbataev/fallback_for_cuda_graphs branch December 10, 2025 19:23
AkCodes23 pushed a commit to AkCodes23/NeMo that referenced this pull request Jan 28, 2026
…IA-NeMo#15173)

* ASR decoding with CUDA graphs: fallback to native PyTorch CUDA graphs if full graph cannot be compiled

Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Signed-off-by: Akhil Varanasi <akhilvaranasi23@gmail.com>
nune-tadevosyan pushed a commit to nune-tadevosyan/NeMo that referenced this pull request Mar 13, 2026
…IA-NeMo#15173)

* ASR decoding with CUDA graphs: fallback to native PyTorch CUDA graphs if full graph cannot be compiled

Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ASR core Changes to NeMo Core

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants