Skip to content

SIMD: The --cpu-baseline is not correctly configured in benchmark system. #17716

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
Closed
Qiyu8 opened this issue Nov 5, 2020 · 1 comment · Fixed by #17736
Closed

SIMD: The --cpu-baseline is not correctly configured in benchmark system. #17716

Qiyu8 opened this issue Nov 5, 2020 · 1 comment · Fixed by #17736

Comments

@Qiyu8
Copy link
Member

Qiyu8 commented Nov 5, 2020

I'm trying to use the newest benchmark system to test the performance, but found out that the option is not working. here is the command that used for benchmarking.

python runtests.py --cpu-baseline="avx2" --bench-compare master bench_linalg.Einsum.time_einsum_multiply

in this case, the macro NPY_SIMD should be 256 because my machine support avx2 intrinsics, but the actual value is 128, which is the default minimal settings.
@seiko2plus seiko2plus mentioned this issue Nov 9, 2020
Merged
@seiko2plus
Copy link
Member

seiko2plus commented Nov 9, 2020 •
edited
Loading

my bad, I confirm it. we should pass the same exact build options to setup.py build and pip wheel
otherwise, the build cache will be cleared while creating the wheel package which leads to a fully rebuild
the source without passing the provided build options.

#17284 was verified through asv --verbose by checking the CCompilerOpt final report but I didn't check
the log of pip because option -v was missing from asv's custom build_command.

#17736, should fix this issue.

@seiko2plus seiko2plus mentioned this issue Nov 9, 2020
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

BUG, Benchmark: fix passing optimization build options to asv seiko2plus/numpy
2 participants
@seiko2plus @Qiyu8