Skip to content

chore: Pin nvidia-cudnn-frontend to 1.23.0#3874

Open
chtruong814 wants to merge 6 commits into
mainfrom
chtruong/26.06-cudnn
Open

chore: Pin nvidia-cudnn-frontend to 1.23.0#3874
chtruong814 wants to merge 6 commits into
mainfrom
chtruong/26.06-cudnn

Conversation

@chtruong814
Copy link
Copy Markdown
Contributor

@chtruong814 chtruong814 commented May 18, 2026

What does this PR do ?

chore: Pin nvidia-cudnn-frontend to 1.23.0

The uv lock file was already using this version. So this is more just formalizing it so that we do not unintentionally change it.

Changelog

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

GitHub Actions CI

See the CI sectionin the Contributing doc for how to trigger the CI. A Nvidia developer will need to approve and trigger the CI for external contributors.

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?

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

Additional Information

  • Related to # (issue)

Signed-off-by: Charlie Truong <chtruong@nvidia.com>
kajalj22
kajalj22 previously approved these changes May 18, 2026
@claude
Copy link
Copy Markdown
Contributor

claude Bot commented May 18, 2026

Light Code Review

Clean change — moves nvidia-cudnn-frontend from a bare pip install in Dockerfile.fw_final (pinned at 1.17.0) into override-dependencies in pyproject.toml (pinned at 1.23.0), which matches the existing uv.lock. This is the right approach: uv-managed pins are the source of truth, and the Dockerfile pip install was a workaround that could drift.

Observations

  1. PR title missing area prefix — per CONTRIBUTING.md, the title should be [build] chore: Pin nvidia-cudnn-frontend to 1.23.0 rather than chore: Pin nvidia-cudnn-frontend to 1.23.0.
  2. Version jump from 1.17.0 to 1.23.0 — the Dockerfile was installing 1.17.0, but the lockfile already resolved 1.23.0. The PR description notes this, so it is intentional. Worth confirming there are no regressions from dropping the older Dockerfile pin (the uv-managed 1.23.0 will now be the only version installed in FW final images too).
  3. No uv.lock diff — expected, since the lockfile already had 1.23.0. Good.

No bugs, typos, or missing test coverage issues found.

Suggested test cases

No perf tests impacted.

@yaoyu-33 yaoyu-33 added area:build Dependencies, packaging, images, and environment setup ci CI, automation, test queue, or workflow infrastructure work full-test-suite waiting-on-customer Waiting on the original author to respond labels May 18, 2026
@chtruong814
Copy link
Copy Markdown
Contributor Author

/ok to test 84d020

Signed-off-by: Charlie Truong <chtruong@nvidia.com>
…ong/26.06-cudnn

Signed-off-by: Charlie Truong <chtruong@nvidia.com>
Signed-off-by: Charlie Truong <chtruong@nvidia.com>
@chtruong814
Copy link
Copy Markdown
Contributor Author

/ok to test 6c75cc8

@chtruong814
Copy link
Copy Markdown
Contributor Author

/ok to test 6c75cc8

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area:build Dependencies, packaging, images, and environment setup ci CI, automation, test queue, or workflow infrastructure work full-test-suite waiting-on-customer Waiting on the original author to respond

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants