Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Bump rapids-cmake version #502

Merged
merged 6 commits into from
Jun 24, 2024
Merged

Bump rapids-cmake version #502

merged 6 commits into from
Jun 24, 2024

Conversation

PointKernel
Copy link
Member

Bumps rapids-cmake version to 24.08

@PointKernel PointKernel added topic: build CMake build issue type: improvement Improvement / enhancement to an existing function labels Jun 7, 2024
Copy link
Collaborator

@sleeepyjack sleeepyjack left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are there any other relevant environment changes apart from switching to CCCL 2.5?

@PointKernel
Copy link
Member Author

Just realized the CCCL2.5 PR merged earlier today, let me rerun the CI to make sure.

@sleeepyjack
Copy link
Collaborator

It seems like most of the errors originate from distinct_count_estimator. I'll take a closer look and see if I can fix the issues.

@PointKernel
Copy link
Member Author

/home/coder/cuCollections/build/amd64-gcc12-cuda12.4/_deps/cccl-src/libcudacxx/lib/cmake/libcudacxx/../../../include/cuda/std/__ranges/access.h(138): error: function "cuda::std::ranges::__4::__begin::__cxx17::__fn::operator()(_Tp &&) const [with _Tp=cuco::distinct_count_estimator_ref<int, cuda::std::__4::thread_scope_device, cuco::detail::XXHash_64<int>> &, _LIBCUDACXX_true_=true, <unnamed>=0, <unnamed>=0]" (declared at line 126) cannot be referenced -- it is a deleted function
239
  using iterator_t = decltype(::cuda::std::ranges::__4::begin(::cuda::std::__4::declval<_Tp&>()));
240
                                                             ^
241
          detected during:
242
            instantiation of type "cuda::std::ranges::__4::iterator_t<cuco::distinct_count_estimator_ref<int, cuda::std::__4::thread_scope_device, cuco::detail::XXHash_64<int>>>" at line 181 of /home/coder/cuCollections/build/amd64-gcc12-cuda12.4/_deps/cccl-src/libcudacxx/lib/cmake/libcudacxx/../../../include/cuda/std/__ranges/concepts.h
243
            instantiation of type "cuda::std::ranges::__4::range_reference_t<cuco::distinct_count_estimator_ref<int, cuda::std::__4::thread_scope_device, cuco::detail::XXHash_64<int>>>" at line 212 of /home/coder/cuCollections/build/amd64-gcc12-cuda12.4/_deps/cccl-src/libcudacxx/lib/cmake/libcudacxx/../../../include/cuda/std/detail/libcxx/include/span
244
            instantiation of "const bool cuda::std::__4::__span_compatible_range [with _Range=cuco::distinct_count_estimator_ref<int, cuda::std::__4::thread_scope_device, cuco::detail::XXHash_64<int>>, _ElementType=std::byte]" at line 147 of /home/coder/cuCollections/examples/distinct_count_estimator/device_ref_example.cu
245

I see ranges... WOW

@sleeepyjack
Copy link
Collaborator

NVIDIA/cccl#1836 should fix the issue with span

@sleeepyjack
Copy link
Collaborator

Waiting for rapidsai/rapids-cmake#631 to get merged so we can use that version to get rid of the build errors. Once done, we can merge this PR and then bump the cuco version in `rapids-cmake' again to the HEAD of this branch.

@sleeepyjack sleeepyjack self-requested a review June 12, 2024 02:18
rapids-bot bot pushed a commit to rapidsai/rapids-cmake that referenced this pull request Jun 24, 2024
This PR updates the CCCL version to include a fix for `cuda::std::span` which is required for cuCollections to work properly with CCCL 2.5.0. 

Most of the changes between the last CCCL version bump (#607) and this one were related to doc updates and unit test fixes, so I don't expect much functional impact for RAPIDS.

After this PR we likely have to bump the cuco version again to include the new changes.

### CCCL PR:

- NVIDIA/cccl#1836

### CUCO PR:

- NVIDIA/cuCollections#502

### RAPIDS PRs:

- [x] rapidsai/cudf#15986
  - [error during docs-build](https://github.com/rapidsai/cudf/actions/runs/9406273871/job/25911619452?pr=15946#step:9:2243) but seems unrelated
- [x] rapidsai/cugraph#4483
  - some CI jobs ran into a network timeout -> rerunning
- [x] rapidsai/cuml#5924
  - successful apart from some optional conda python tests
- [x] rapidsai/raft#2358
  - This one is weird as it [shows an error](https://github.com/rapidsai/raft/actions/runs/9475150045/job/26109283214?pr=2358#step:7:824) that is similar to the one we found in cuco when bumping the CCCL version to 2.5.0. However, we thought the problem was on cuco's end and pushed a fix that resolved the issue (see [Slack thread](https://nvidia.slack.com/archives/CCP05T27R/p1718060955876199))
- [x] rapidsai/rmm#1584

Authors:
  - Daniel Jünger (https://github.com/sleeepyjack)

Approvers:
  - Bradley Dice (https://github.com/bdice)

URL: #631
@sleeepyjack
Copy link
Collaborator

Looks good. Ship it! 🛳️⛴️🚢

@PointKernel PointKernel merged commit fe372f6 into dev Jun 24, 2024
15 checks passed
@PointKernel PointKernel deleted the PointKernel-patch-1 branch June 24, 2024 19:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
topic: build CMake build issue type: improvement Improvement / enhancement to an existing function
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants