Update libcudacxx to 2.1.0 #464
Merged
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
|
Tracking CI pass/fail in these PRs:
I think it should be sufficient if these pass, since these are the only RAPIDS repos that explicitly use |
bdice
added
improvement
|
cuSpatial is a bit tricky to test with an updated libcudacxx package unless I build it locally from source. I am going to attempt that tomorrow to verify, but overall I expect the risk of breaking builds by updating libcudacxx to 2.1.0 is very low. |
|
I verified that cuSpatial builds locally. This PR should be ready to review and merge. |
robertmaynard
approved these changes
Oct 16, 2023
vyasr
approved these changes
Oct 16, 2023
|
/merge |
sleeepyjack
pushed a commit
to NVIDIA/cuCollections
that referenced
this pull request
Oct 16, 2023
This updates cuCollections to rapids-cmake 23.12. This comes with rapidsai/rapids-cmake#464, which updates libcudacxx to 2.1.0. That should unblock several cuCollections issues such as #332, #331, #289.
|
That is amazing, thanks a lot for pushing this through |
rapids-bot bot
pushed a commit
to rapidsai/cugraph
that referenced
this pull request
Dec 8, 2023
This PR adds `cuda::proclaim_return_type` to device lambdas used in `thrust::transform` and `thrust::make_transform_iterator`. This PR requires libcudacxx 2.1.0, which was provided by rapidsai/rapids-cmake#464. Closes #3863. Authors: - Seunghwa Kang (https://github.com/seunghwak) - Bradley Dice (https://github.com/bdice) Approvers: - Bradley Dice (https://github.com/bdice) - Chuck Hastings (https://github.com/ChuckHastings) URL: #3862
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Outdated suggestions cannot be applied.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.
Description
This PR separates out the libcudacxx update from #399. I am proposing to update only libcudacxx to 2.1.0, and leave thrust/cub pinned at 1.17.2 until all of RAPIDS is ready to update. Then we can move forward with #399 next.
Separating the update for libcudacxx should allow RAPIDS to use some of the new features we want while giving more time to RAPIDS libraries to migrate to CCCL 2.1.0 (particularly for breaking changes in Thrust/CUB).
Immediate benefits of bumping only libcudacxx to 2.1.0:
cuda::proclaim_return_typeinto cudf/etc. which would reduce the amount of unmerged code we're maintaining in the "testing PRs" while waiting for all RAPIDS repos to be ready for Thrust/CUB 2.1.0.cuda::mr::memory_resourceinstead of rawdevice_memory_resourcermm#1095) and quite a few planned changes for cuCollections (such as Switch to cuda::std::span once available NVIDIA/cuCollections#332, Switch to cuda::stream_ref NVIDIA/cuCollections#331, [ENHANCEMENT]: Switch to cuda::memory_resource once it is available NVIDIA/cuCollections#289)Risk Assessment:
This should be fairly low risk because libcudacxx 2.1.0 is similar to our current pinning of 1.9.1 -- the major version bump was meant to align with Thrust/CUB and isn't indicative of major breaking changes.
Checklist
cmake-format.jsonis up to date with these changes.include_guard(GLOBAL))