driazati commented on a change in pull request #9733:
URL: https://github.com/apache/tvm/pull/9733#discussion_r779238411
##########
File path: Jenkinsfile
##########
@@ -397,10 +390,10 @@ stage('Unit Test') {
}
}
} else {
- Utils.markStageSkippedForConditional('python3: GPU')
+ Utils.markStageSkippedForConditional('unittest: GPU')
}
},
- 'python3: CPU': {
Review comment:
sorry should have included this, but this is how I did the accounting:
<details>
```
WHEN name like '%Build and run C++ tests' THEN 'build'
WHEN name like '%Run cmake build' THEN 'build'
WHEN name like '%Run microTVM tests' THEN 'build'
WHEN name like '%integration tests' THEN 'integration'
WHEN name like '%frontend tests' THEN 'integration'
WHEN name like '%unit tests' THEN 'unit tests'
WHEN name like '%Sphinx warnings in docs' THEN 'lint'
WHEN name like '%Run lint' THEN 'lint'
WHEN name like '%executor node info' THEN 'infra'
WHEN name like '%Check out from version control' THEN 'infra'
WHEN name like '%JUnit-formatted test results' THEN 'infra'
WHEN name like '%Docker image names' THEN 'infra'
WHEN name like '%files previously stashed' THEN 'infra'
WHEN name like '%Rust build and test' THEN 'build'
```
</details>
Isn't the concern the case where unit tests that would have failed and then
caused frontend (integration) tests to not run, so the number to look at is
unit test failures (since with this PR those would no longer be gating frontend
failures)? Frontend failures would mean that the build got all the way there
anyways so the capacity requirements of those would be the same
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
