Skip to content

NXP backend: Enable initial unit tests workflow #11396

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged

Conversation

robert-kalmar
Copy link
Collaborator

@robert-kalmar robert-kalmar commented Jun 5, 2025

Summary

Initial workflow definition for running unit tests for Neutron (NXP) backend.

Follow up for the #11109, our IT updated the eIQ PyPI, so the neutron-converter python package shall be reachable using the executorch's self-hosted agents too.

@digantdesai , @mergennachin as the problem was only reproduced with the self-hosted agents used by the ExecuTorch/PyTorch, I need your help to verify the IT change. Can you please let run the "trunk" workflow?

cc @digantdesai @JakeStevens

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jun 5, 2025
Copy link

pytorch-bot bot commented Jun 5, 2025

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/11396

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

❌ 1 New Failure

As of commit 92eea29 with merge base c120b35 (image):

NEW FAILURE - The following job has failed:

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

@robert-kalmar robert-kalmar added module: nxp Issues related to NXP Neutron NPU delegation and code under backends/nxp/ release notes: nxp Changes to the NXP Neutron backend delegate labels Jun 5, 2025
SCRIPT_DIR=$(dirname $(readlink -fm $0))
EXECUTORCH_DIR=$(dirname $(dirname $SCRIPT_DIR))

cd $EXECUTORCH_DIR
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't we need to install the NXP sdk here? See my PR with retries, I am on phone so can't look it up.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done. The test requirements are installed in the trunk.yaml:

# Install test requirements
pip install -r backends/nxp/requirements-tests.txt

@digantdesai
Copy link
Contributor

Looks good, left a comment, and add a trunk label, hopefully it will be all green now.

Comment on lines 697 to 700
docker-image: executorch-ubuntu-22.04-clang12
submodules: 'true'
ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
timeout: 90
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@kirklandsign mentioned it's better to have a separate docker image for a specific backend to avoid unstable CI jobs.

Copy link
Collaborator Author

@robert-kalmar robert-kalmar Jun 6, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sounds reasonable.
For curiosity, the github job runs from a clean docker container, right? It does not reuse already instantiated containers. (I don't see this explicitly from the log, but as the docker image was pulled I guess clean docker container is used)
Where I am heading if we can run the trunk workflow, after my fixes, to chase one error at a time, or we must isolate the build into its own docker container to not break containers for other jobs.

@cccclai , @kirklandsign, can you elaborate more about "unstable CI jobs". The background it to not mix the backends SDK, what can be potentially installed during the docker image creation (https://github.com/pytorch/executorch/blob/main/.ci/docker/ubuntu/Dockerfile#L85)? If I understand the script correctly currently only Android really uses this option, right?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Addressed is second PR: #11473

with:
runner: linux.2xlarge
docker-image: executorch-ubuntu-22.04-clang12
submodules: 'true'
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Must be set to "recursive" since 7e034ca.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fix duplicate bpe tokenizer base symbol (pytorch#63)
@robert-kalmar robert-kalmar force-pushed the upstream/main-nxp/neutron-unittest-ci branch from 1375a85 to 92eea29 Compare June 6, 2025 10:13
@pytorch-bot pytorch-bot bot removed the ciflow/trunk label Jun 6, 2025
@robert-kalmar
Copy link
Collaborator Author

@digantdesai , we are green.
The tests (including trunk workflow) passed. https://github.com/pytorch/executorch/actions/runs/15498620828/job/43641195770?pr=11396#step:15:11806.

The failing trunk/test-arm-cortex-m-size-test seems unrelated. It failed on slight size increase for the cmake-out/test/size_test application (https://github.com/pytorch/executorch/actions/runs/15498620828/job/43641195433?pr=11396#step:15:15498). Maybe someone from Arm to confirm?

The @cccclai request is addressed in separate PR #11473 to have a baseline here. Started the tests, if pass, we can merge the PR 11473 instead of this as it contains the NXP backend: Enable initial unit tests workflow commit too.

@digantdesai
Copy link
Contributor

Cc @GregoryComer for Arm size test.
LGTM otherwise.

@digantdesai
Copy link
Contributor

Thanks you guys. Since this is on CI trunk for now, make sure to manually apply trunk label to NXP PRs. Also going forward let's ensure new PRe include tests for the CI so we don't regress silently.

@digantdesai digantdesai merged commit 8f05c35 into pytorch:main Jun 9, 2025
198 of 200 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ciflow/trunk CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. module: nxp Issues related to NXP Neutron NPU delegation and code under backends/nxp/ release notes: nxp Changes to the NXP Neutron backend delegate
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants