Tests: Increase robustness of context creation testing #1096
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
To ensure that no cuda calls occur prior to device selection, an always run method in the test suite previously used timing and a cutoff threshold to make sure a new context was being established. This was quite flakey, as new driver updates could improve context creation times, requiring a lower threshold. Instead, getting the current context via the driver api is used, and it should return null (or a cuda error) if no context yet exists. This should be a much more robust testing mechanism. This is still created per run of the test suite, regardless of if the specific test is requested or not, which is wasted time if running with a filter (one more context creation than needed, so extra time). We may wish to move this test to a separate ctest orcestrated test binary, to ensure it is the first to run (Google test has no way to enforce test order as a feature)
mondus
approved these changes
Jul 28, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
To ensure that no cuda calls occur prior to device selection, an always run method in the test suite previously used timing and a cutoff threshold to make sure a new context was being established. This was quite flakey, as new driver updates could improve context creation times, requiring a lower threshold.
Instead, getting the current context via the driver api is used, and it should return null (or a cuda error) if no context yet exists. This should be a much more robust testing mechanism.
This is still created per run of the test suite, regardless of if the specific test is requested or not, which is wasted time if running with a filter (one more context creation than needed, so extra time).
We may wish to move this test to a separate ctest orcestrated test binary, to ensure it is the first to run (Google test has no way to enforce test order as a feature)
Tested on linux with CUDA 11.0 and 12.2 (12.2/driver 535 broke the old test on my 3060ti as context creation was too fast), and manually added a context creating call prior to
applyConfigto ensure it will identify any errors that it is checking for.Also removed a superfluos include.