Merged
Conversation
…scope The cuda_prune_library_targets was not being exposed in the parent scope because it was used in a list. This commit changes the list to a set to append the targets to that CMake variable which allows us to use PARENT_SCOPE in the command.
Make pruning CUDA static libraries optional for static CUDA libraries because nvprune seems to fail for some combination of CUDA toolkits and compute capabilities.
syurkevi
approved these changes
Mar 30, 2022
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fix pruning option and set to off by default because of some failures in some combination of compute capabilities and cuda toolkits
Description
list(APPEND ...) in CMake cannot use the PARENT_SCOPE attribute. This caused an error when setting the cuda_pruned_library_targets variable which didn't propagate to the parent scope.
This PR also makes pruning optional because it fails for some compute capabilities and cuda toolkits. For example, it seemed to be working fine with 11.5 and compute capability 7.0 but not 7.5. It was failing for both computes in CUDA toolkit 10.2. It would be difficult to maintain and test all toolkits and compute combinations so I am setting it to off by default.
Changes to Users
N/A
Checklist
[ ] Functions added to unified API[ ] Functions documented