Skip to content

Only use fma functions when CV_FMA3 is set.#21011

Merged
opencv-pushbot merged 1 commit intoopencv:3.4from
vrabaud:3.4
Nov 5, 2021
Merged

Only use fma functions when CV_FMA3 is set.#21011
opencv-pushbot merged 1 commit intoopencv:3.4from
vrabaud:3.4

Conversation

@vrabaud
Copy link
Copy Markdown
Contributor

@vrabaud vrabaud commented Nov 4, 2021 •
edited by alalek
Loading

In practice, processors offering AVX2/AVX512 also FMA, that is why it got unnoticed.

Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

  • I agree to contribute to the project under Apache 2 License.
  • To the best of my knowledge, the proposed patch is not based on a code under GPL or other license that is incompatible with OpenCV
  • The PR is proposed to proper branch
force_builders=linux,docs,Linux AVX2,Custom
buildworker:Custom=linux-3
build_image:Custom=ubuntu:18.04
CPU_BASELINE:Custom=AVX512_SKX
disable_ipp=ON

@vrabaud
Only use fma functions when CV_FMA3 is set.
ffd0107 
In practice, processors offering AVX2/AVX512 also FMA, that is why it got unnoticed.
@alalek
Copy link
Copy Markdown
Member

alalek commented Nov 4, 2021 •
edited
Loading

Thank you for contribution!

Please ignore CMake/ABI errors from "Linux x64" builder (it is CI-related). Resolved

Reproduced original problem (excluding FMA3 from AVX2):

cmake -GNinja <opencv> -DCPU_AVX2_IMPLIES=AVX\;FP16

@opencv-pushbot opencv-pushbot merged commit 2ce47fd into opencv:3.4 Nov 5, 2021
@alalek alalek mentioned this pull request Nov 5, 2021
Merged
@alalek alalek mentioned this pull request Dec 30, 2021
Merged
@alalek alalek mentioned this pull request Feb 22, 2022
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@alalek alalek alalek approved these changes

Assignees

@alalek alalek

Labels

category: core optimization

Projects

None yet

Milestone

3.4.17

Development

Successfully merging this pull request may close these issues.

3 participants

@vrabaud @alalek @opencv-pushbot