Skip to content

Avoid installing CCCL headers in wheels#7662

Merged
rapids-bot[bot] merged 12 commits intorapidsai:mainfrom
divyegala:wheels-no-cccl-headers
Jan 14, 2026
Merged

Avoid installing CCCL headers in wheels#7662
rapids-bot[bot] merged 12 commits intorapidsai:mainfrom
divyegala:wheels-no-cccl-headers

Conversation

@divyegala
Copy link
Copy Markdown
Member

@divyegala divyegala commented Jan 12, 2026

Uncompressed size of CCCL headers: 19M
No of headers per CCCL library:

cudax/include:
208
cub/cub:
172
libcudacxx/include/cuda:
950
thrust/thrust:
729

Before this PR:

  • compressed size: 0.232G
  • uncompressed size: 0.273G

After this PR:

  • compressed size: 0.229G
  • uncompressed size: 0.258G

@divyegala divyegala self-assigned this Jan 12, 2026
@divyegala divyegala requested review from a team as code owners January 12, 2026 20:05
@divyegala divyegala added the improvement Improvement / enhancement to an existing function label Jan 12, 2026
@divyegala divyegala requested a review from vyasr January 12, 2026 20:05
@divyegala divyegala added the non-breaking Non-breaking change label Jan 12, 2026
@divyegala divyegala requested review from aamijar and csadorf January 12, 2026 20:05
@divyegala divyegala requested a review from a team as a code owner January 12, 2026 20:11
@github-actions github-actions Bot added conda conda issue Cython / Python Cython or Python issue labels Jan 12, 2026
Comment thread conda/recipes/libcuml/recipe.yaml
Comment thread cpp/CMakeLists.txt Outdated
Comment thread cpp/CMakeLists.txt
Comment thread cpp/src/arima/arima_common.cu
bdice
bdice previously requested changes Jan 13, 2026
Comment thread cpp/CMakeLists.txt Outdated
Copy link
Copy Markdown
Member

@jameslamb jameslamb left a comment

Choose a reason for hiding this comment

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

Agree with other reviewers' comments but I don't have anything to add. Changes make sense to me.

Comment thread cpp/CMakeLists.txt Outdated
Comment thread conda/recipes/libcuml/recipe.yaml
divyegala and others added 2 commits January 14, 2026 20:11
Closes rapidsai#7506.

This PR mirrors the input mem type when given `build_algo=nn_descent` in HDBSCAN. 
This is because
- given data on host
  - nn descent internally allocates GPU memory anyways so we keep the data on CPU
- given data on device
  - nn descent still internally allocates additional GPU memory (for processing in fp16) so no reason to copy data to CPU.

Authors:
  - Jinsol Park (https://github.com/jinsolp)

Approvers:
  - Anupam (https://github.com/aamijar)
  - Victor Lafargue (https://github.com/viclafargue)
  - Simon Adorf (https://github.com/csadorf)

URL: rapidsai#7614
@divyegala divyegala requested a review from a team as a code owner January 14, 2026 20:11
@divyegala divyegala requested a review from jcrist January 14, 2026 20:11
@divyegala divyegala removed request for a team and jcrist January 14, 2026 20:12
@divyegala
Copy link
Copy Markdown
Member Author

/merge

@rapids-bot rapids-bot Bot merged commit f9726a2 into rapidsai:main Jan 14, 2026
120 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CMake conda conda issue CUDA/C++ Cython / Python Cython or Python issue improvement Improvement / enhancement to an existing function non-breaking Non-breaking change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants