[CI] Prebuild E2E tests as part of build workflow #16682

aelovikov-intel · 2025-01-18T00:39:10Z

This will allow us to save time on runner setup and toolchain download/unpacking.

.github/workflows/sycl-linux-build.yml

uditagarwal97 · 2025-01-23T23:52:10Z

.github/workflows/sycl-linux-build.yml

+    - name: Source OneAPI TBB vars.sh
+      shell: bash
+      run: |
+        # https://github.com/actions/runner/issues/1964 prevents us from using
+        # the ENTRYPOINT in the image.
+        env | sort > env_before
+        if [ -e /runtimes/oneapi-tbb/env/vars.sh ]; then
+          source /runtimes/oneapi-tbb/env/vars.sh;
+        elif [ -e /opt/runtimes/oneapi-tbb/env/vars.sh ]; then
+          source /opt/runtimes/oneapi-tbb/env/vars.sh;
+        else
+          echo "no TBB vars in /opt/runtimes or /runtimes";
+        fi
+        env | sort > env_after
+        comm -13 env_before env_after >> $GITHUB_ENV
+        rm env_before env_after


Does it mean that whenever we call .github/workflows/sycl-linux-build.yml, we also build E2E tests? IIRC, we don't use pre-built test binaries in post-commit and nightly.

Yes, we do, and no, we don't. That said I imagine we can start doing so relatively soon. Alternatively, we can make this addition optional, but I thought that build-only mode of e2e tests isn't that different from check-sycl that we do unconditionally everywhere (well, if sycl is modified in pre-commit). Writing this, I can even imagine re-unification of sycl/test and sycl/test-e2e now!

I would personally be of the opinion that we make this addition optional - don't build E2E tests by default when sycl-linux-build.yml is called, unless the caller explicitly asks for it.
Once the "split-test" mode is mature enough that we use it for all workflows and other runners as well, we can build E2E tests by default, just like sycl/test.

I think it will just complicate implementation with very little benefit... Nightly/post traffic is several times smaller than pre-commit (because PR usually take several iterations). Also, unlike pre-commit, nightly/post don't skip any check-* at build stage, so having e2e built there takes less percentage of the overall duration.

i agree with udit if the implementation cost is low, buf it's high as andrei says im fine with enabling it always

I think it will just complicate implementation with very little benefit...

If it complicates the implementation, I'm fine with the current PR.

Thanks. I'll plan to work on it the coming days/weeks. Either make sure the e2e artifacts get used more or make optional, don't believe we need to delay progress by making extra changes and extra testing. Too easy to make a typo and too long to run the full CI cycle. Very much would like to merge while it's green :)

.github/workflows/sycl-linux-precommit.yml

sarnex

lgtm just minor kinda unrelated questions

sarnex · 2025-01-24T15:15:52Z

.github/workflows/sycl-linux-precommit.yml

@@ -47,7 +47,7 @@ jobs:
      build_artifact_suffix: "default"
      build_cache_suffix: "default"
      # Docker image has last nightly pre-installed and added to the PATH
-      build_image: "ghcr.io/intel/llvm/sycl_ubuntu2204_nightly:build"
+      build_image: "ghcr.io/intel/llvm/sycl_ubuntu2404_nightly:latest"


can we just delete the build image?

discussed offline, we only create *nightly:latest now, the :build is getting increasingly stale/outdated.

sarnex · 2025-01-24T15:16:46Z

.github/workflows/sycl-linux-build.yml

+        e2e_testing_mode: build-only
+        target_devices: all
+        artifact_suffix: default
+        cxx_compiler: $GITHUB_WORKSPACE/toolchain/bin/clang++


if we dont pass this, it falls back to which clang++, so i assume the built clang++ isn't on the path?

discussed offline, we need to make sure CMake sees the same CXX compiler at build/run stages.

Reduce number of input parameters and make the logic a bit cleaner (IMO). This PR also uses that updated logic to make building E2E tests optional in `sycl-linux-build.yml` and makes enabled in pre-commit only for now, effectively fixing the regression in Nightly CI introduced in #16682.

) Reduce number of input parameters and make the logic a bit cleaner (IMO). This PR also uses that updated logic to make building E2E tests optional in `sycl-linux-build.yml` and makes enabled in pre-commit only for now, effectively fixing the regression in Nightly CI introduced in #16682.

aelovikov-intel temporarily deployed to WindowsCILock January 18, 2025 00:39 — with GitHub Actions Inactive

aelovikov-intel temporarily deployed to WindowsCILock January 18, 2025 00:54 — with GitHub Actions Inactive

aelovikov-intel temporarily deployed to WindowsCILock January 18, 2025 01:24 — with GitHub Actions Inactive

aelovikov-intel force-pushed the e2e-build branch from dadb6af to 2060b2a Compare January 21, 2025 18:04

aelovikov-intel had a problem deploying to WindowsCILock January 21, 2025 18:04 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from 2060b2a to 902d98d Compare January 21, 2025 19:03

aelovikov-intel temporarily deployed to WindowsCILock January 21, 2025 19:03 — with GitHub Actions Inactive

aelovikov-intel had a problem deploying to WindowsCILock January 21, 2025 20:40 — with GitHub Actions Error

aelovikov-intel had a problem deploying to WindowsCILock January 21, 2025 20:55 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from 902d98d to c79e005 Compare January 21, 2025 21:04

aelovikov-intel had a problem deploying to WindowsCILock January 21, 2025 21:04 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from c79e005 to 52ab1c0 Compare January 21, 2025 22:11

aelovikov-intel had a problem deploying to WindowsCILock January 21, 2025 22:11 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from 52ab1c0 to 3169ec2 Compare January 23, 2025 00:47

aelovikov-intel had a problem deploying to WindowsCILock January 23, 2025 00:47 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from 3169ec2 to b5cf3bb Compare January 23, 2025 00:49

aelovikov-intel had a problem deploying to WindowsCILock January 23, 2025 00:49 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from b5cf3bb to ca90f35 Compare January 23, 2025 00:49

aelovikov-intel temporarily deployed to WindowsCILock January 23, 2025 00:50 — with GitHub Actions Inactive

aelovikov-intel temporarily deployed to WindowsCILock January 23, 2025 01:20 — with GitHub Actions Inactive

aelovikov-intel temporarily deployed to WindowsCILock January 23, 2025 01:35 — with GitHub Actions Inactive

aelovikov-intel force-pushed the e2e-build branch from ca90f35 to 6b3c47e Compare January 23, 2025 16:13

aelovikov-intel temporarily deployed to WindowsCILock January 23, 2025 16:14 — with GitHub Actions Inactive

aelovikov-intel had a problem deploying to WindowsCILock January 23, 2025 17:41 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from 77d41e2 to fb70bbf Compare January 23, 2025 19:02

aelovikov-intel had a problem deploying to WindowsCILock January 23, 2025 19:03 — with GitHub Actions Error

aelovikov-intel force-pushed the e2e-build branch from fb70bbf to 63c84c2 Compare January 23, 2025 21:06

aelovikov-intel had a problem deploying to WindowsCILock January 23, 2025 21:06 — with GitHub Actions Error

aelovikov-intel had a problem deploying to WindowsCILock January 23, 2025 21:07 — with GitHub Actions Error

[CI] Prebuild E2E tests as part of build workflow

1416929

aelovikov-intel force-pushed the e2e-build branch from 63c84c2 to 1416929 Compare January 23, 2025 21:46

aelovikov-intel temporarily deployed to WindowsCILock January 23, 2025 21:46 — with GitHub Actions Inactive

aelovikov-intel temporarily deployed to WindowsCILock January 23, 2025 22:13 — with GitHub Actions Inactive

aelovikov-intel temporarily deployed to WindowsCILock January 23, 2025 22:41 — with GitHub Actions Inactive

aelovikov-intel marked this pull request as ready for review January 23, 2025 22:49

aelovikov-intel requested a review from a team as a code owner January 23, 2025 22:49

uditagarwal97 reviewed Jan 23, 2025

View reviewed changes

aelovikov-intel requested a review from uditagarwal97 January 24, 2025 00:06

sarnex approved these changes Jan 24, 2025

View reviewed changes

uditagarwal97 approved these changes Jan 24, 2025

View reviewed changes

aelovikov-intel merged commit 5f69e3e into sycl Jan 24, 2025
25 checks passed

aelovikov-intel deleted the e2e-build branch January 24, 2025 17:03

aelovikov-intel added a commit that referenced this pull request Jan 24, 2025

[CI] Fix Nightly after #16682

d9b1a5f

aelovikov-intel added a commit that referenced this pull request Jan 24, 2025

[CI] Fix Nightly after #16682

163589c

aelovikov-intel mentioned this pull request Jan 25, 2025

[CI] Refactor build/run-only logic in run-tests/e2e/action.yml #16780

Merged

EwanC mentioned this pull request Feb 7, 2025

[SYCL][Graph] Update to UR cmd-buf enqueue rename #16747

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[CI] Prebuild E2E tests as part of build workflow #16682

[CI] Prebuild E2E tests as part of build workflow #16682

aelovikov-intel commented Jan 18, 2025 •

edited

Loading

uditagarwal97 Jan 23, 2025

aelovikov-intel Jan 24, 2025

uditagarwal97 Jan 24, 2025

aelovikov-intel Jan 24, 2025

sarnex Jan 24, 2025

uditagarwal97 Jan 24, 2025

aelovikov-intel Jan 24, 2025

sarnex left a comment

sarnex Jan 24, 2025

aelovikov-intel Jan 24, 2025

sarnex Jan 24, 2025 •

edited

Loading

aelovikov-intel Jan 24, 2025

[CI] Prebuild E2E tests as part of build workflow #16682

[CI] Prebuild E2E tests as part of build workflow #16682

Conversation

aelovikov-intel commented Jan 18, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sarnex left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sarnex Jan 24, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

aelovikov-intel commented Jan 18, 2025 •

edited

Loading

sarnex Jan 24, 2025 •

edited

Loading