Skip to content

Conversation

@maksimsab
Copy link
Contributor

@maksimsab maksimsab commented Nov 13, 2025

This patch has the following changes:

  • --param enable_new_offload_model is added to SYCL E2E LIT test configuration
  • NewOffloadModel mode run is added to sycl-linux-precommit
  • NewOffloadModel mode run is added to sycl-nightly for linux and windows.

runner: '["Windows", "gen12"]'
target_devices: level_zero:gpu
uses: ./.github/workflows/sycl-windows-run-tests.yml
permissions: read-all

Check warning

Code scanning / zizmor

overly broad permissions Warning

overly broad permissions
@maksimsab maksimsab force-pushed the enable_e2e_testing_new_offload branch from 49a3481 to b8a7ff6 Compare November 18, 2025 15:58
@maksimsab maksimsab changed the title [CI] Enable e2e testing for New Offloading Model [CI][SYCL] Enable E2E testing for New Offloading Model Nov 18, 2025
@maksimsab maksimsab marked this pull request as ready for review November 18, 2025 16:05
@maksimsab maksimsab requested review from a team as code owners November 18, 2025 16:05
@maksimsab
Copy link
Contributor Author

Hi @sarnex

Could you please help with reviewing this change or recommend somebody for that?

@sarnex
Copy link
Contributor

sarnex commented Nov 18, 2025

I can review, thanks

binaries_artifact: e2e_bin_preview
- name: New Offload Model
runner: '[Linux, "gen12"]'
target_devices: opencl:cpu
Copy link
Contributor

Choose a reason for hiding this comment

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

High level question, do we think changes that break the new offload model but nothing else are common enough to justify precommit testing? Last I heard we were only going to do nightly testing

Copy link
Contributor Author

@maksimsab maksimsab Nov 18, 2025

Choose a reason for hiding this comment

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

I was also considering that aspect. To being able to check E2E tests for changes related to new model we need either precommit functionality or some workflow other that sycl-nightly that we would run manually every time in PR. Perhaps, we could introduce some PR label "new-offload-model" that could enable this branch in precommit.

Copy link
Contributor

@sarnex sarnex Nov 18, 2025

Choose a reason for hiding this comment

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

I'm happy to have it in pre or postcommit if you guys ( @YuriPlyakhin / @intel/dpcpp-tools-reviewers ) think there is a decent likelihood that some PR passes all other CI but fails on the new offload model. If we think that is pretty rare I would recommend we do nightly-only testing, if you want to manually test new offload stuff in a PR you can run the nightly explicitly though the github UI

Copy link
Contributor

Choose a reason for hiding this comment

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

If you decide to go with the label, then

test-perf:
needs: [build, detect_changes]
permissions:
contents: write
packages: read
if: |
!cancelled()
&& needs.build.outputs.build_conclusion == 'success'
&& (contains(github.event.pull_request.labels.*.name, 'run-perf-tests')
|| contains(needs.detect_changes.outputs.filters, 'perf-tests'))

might be useful to look at. I don't remember why I decided to keep that as a separate job instead of extending regular matrix though.

if: ${{ !cancelled() && needs.build-win.outputs.build_conclusion == 'success' }}
strategy:
fail-fast: false
uses: ./.github/workflows/sycl-windows-run-tests.yml
Copy link
Contributor

Choose a reason for hiding this comment

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

permissions:
contents: write
packages: read

should work hopefully

Copy link
Contributor Author

Choose a reason for hiding this comment

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

For some reason, sycl-windows-run-tests.yml requires many permissions.

Copy link
Contributor

Choose a reason for hiding this comment

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

I think we just didn't restrict the permissions on that one, did you try restricting them and it failed?

@sarnex
Copy link
Contributor

sarnex commented Nov 18, 2025

Also you might want to rebase this PR against HEAD if you're going to run the nightly, I merged a fix today that should fix the AOT failures, they're unrelated to this PR.

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants