Skip to content

Various improvements #78

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.

Sign up for GitHub

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

Jump to bottom
Open
breyerml wants to merge 215 commits into
base: develop
Choose a base branch
from
Open

Various improvements #78

breyerml wants to merge 215 commits into from

Conversation

@breyerml
Copy link
Member

@breyerml breyerml commented Dec 15, 2025 •
edited
Loading

  • Improve the performance of all backends.
  • Add support for scikit-learn 1.8.0 in our Python bindings.
  • Rename SYCL's kernel invocation types to data parallel kernels.
  • Fix nearly all clang-tidy warnings.
  • Fix some test issues.
  • Other small improvements.

breyerml added 30 commits May 24, 2025 22:07
@breyerml
Improve performance (mainly on AMD GPUs) and change implementations s…
18691a5 
…lightly such that the backends are more similar.
@breyerml
Additional performance improvement tests.
6cddbb6
@breyerml
Preliminary changes.
a185caf
@breyerml
Update CUDA implementation and update comments.
c74aca8
@breyerml
Do not use std::vector directly for the kernel matrix since it sequen…
dbc00ae 
…tially initializes all values to zero.

Instead, use a std::unique_ptr together with a C++17 conformant make_unique_for_overwrite implementation followed by an OpenMP parallel zero initialization of all values drastically reducing the overhead.
@breyerml
Improve the performance of the OpenMP cg_explicit kernel matrix assem…
10d303e 
…bly and BLAS implementation. Align names more to the ones used in the other backends.
@breyerml
Improve the performance of the OpenMP cg_implicit kernel matrix assem…
2e64193 
…bly + BLAS implementation. Align names more to the ones used in the other backends.
@breyerml
Improve the performance of the OpenMP predict implementation. Align n…
8aa1c93 
…ames more to the ones used in the other backends.
@breyerml
Improve variable names and remove some implicit conversions.
51b75b6
@breyerml
Fix tests after slight API changes.
8a570a8
@breyerml
Remove unnecessary conditions. Improve variable naming.
3025c76
@breyerml
Update variable names.
46a9558
@breyerml
Update documentation and add missing headers.
0c68206
@breyerml
Update the HIP backend kernels.
a9e8271
@breyerml
Fix Doxygen documentation.
45832e7
@breyerml
Add additional assert.
10ff3c2
@breyerml
Fix variable names.
dad55f2
@breyerml
Use typename instead of class.
e6b76f2
@breyerml
Move parallel zero memset to header function (used in multiple places).
5913b50
@breyerml
Add documentation and rearrange constant declarations.
a67751b
@breyerml
Inverse all temp indices for better consistency.
54741ff
@breyerml
Add missing doxygen documentation.
46891d9
@breyerml
Update the HPX backend kernels.
fa5cea3
@breyerml
Some small changes: where possible change remaining const to constexp…
ff89212 
…r, remove superfluous braces, add missing static_casts, and use correct THREAD_BLOCK_SIZE_uz.
@breyerml
Update comments.
b4d553a
@breyerml
Rename sv to support_vectors for better readability and consistency.
4020339
@breyerml
Update some comments.
39513f8
@breyerml
Also use trimmed names in performance tracking output.
3ef281d
@breyerml
Always use a loop for the custom powi function.
a20d76d
@breyerml
The get_default_queue now honors the default target platform.
1c55fb1
@breyerml
Fix clang-format.
a20af0a
@breyerml
Update GitHub actions.
5a4db76
@breyerml
Fix install documentation.
c6d107d
@breyerml
Revert py::native_enum changes back to py::enum_ to again support str…
2416aba 
…ing -> C++ enum conversions which is not possible with py::native_enum without worsening the public API.
@breyerml
Try using Python virtual env to fix macOS GitHub action error.
3c1bd4d
@breyerml
Try using older {fmt} version to fix MSVC GitHub action error.
9387abd
@breyerml
Reduce ctest output.
da1fbbe
@breyerml
Fix formatting.
2573017
@breyerml
Try removing pip install.
f934876
@breyerml
Add __sklearn_tags__ function which is mandatory since version 1.8.0.
86462df
@breyerml
Try reducing the amount of dots in the GitHub action output.
e8f9413
@breyerml
Add necessary pickle support to the new Tag classes.
3cd4b16
@breyerml
Fix some warnings in newer scikit-learn versions.
e4f87cb
@breyerml
Try making the ctest output quiet and instead write it to a file and …
2417369 
…upload it as artifact.
@breyerml
Try alternative macOs Python library build.
eae87f5
@breyerml
Revert recent changes and use older CMake version again (maybe fixing…
67c9c22 
… the ctest problem in the GitHub actions).
@breyerml
Try fixing the macOS runner related errors.
cbaefb4
@breyerml
Switch back to older LLVM version.
98d7a1e
@breyerml
Fix regex after test name refactor.
353ee2c
@breyerml
Remove all trailing \0 of the retrieved hostname and username (result…
de7a0fc 
…ed in failing tests since the string comparison stopped early after reading \0).
@breyerml
Add hostname and username support for Windows.
017ea87
@breyerml
Fix the macOS GitHub action.
ca7cc7c
@breyerml
Fix linker error.
5e12c3b
@breyerml
Remove Windows support for hostname and username query again due to e…
2b3611b 
…xtended compiler and linker errors in Windows internal headers.
@breyerml breyerml marked this pull request as ready for review December 22, 2025 11:46
@breyerml breyerml requested a review from vancraar December 22, 2025 12:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@vancraar vancraar Awaiting requested review from vancraar

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@breyerml