Skip to content

remove default stream sync in cuda TVL1 optical flow#3089

Merged
opencv-pushbot merged 1 commit intoopencv:masterfrom
noahstier:tlv1_cuda_optimization
Oct 26, 2021
Merged

remove default stream sync in cuda TVL1 optical flow#3089
opencv-pushbot merged 1 commit intoopencv:masterfrom
noahstier:tlv1_cuda_optimization

Conversation

@noahstier
Copy link
Copy Markdown
Contributor

@noahstier noahstier commented Oct 25, 2021 •
edited by alalek
Loading

This is an optimization in the cuda TVL1 optical flow implementation.

It moves a summation reduction off of the default cuda stream, preventing a device sync. It also eliminates a slow cudaHostAlloc and cudaHostFree.

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
  • There is reference to original bug report and related work
  • There is accuracy test, performance test and test data in opencv_extra repository, if applicable
    Patch to opencv_extra has the same branch name.
  • The feature is well documented and sample code can be built with the project CMake
force_builders=Custom
build_worker:Custom=linux-4,linux-6
build_image:Custom=ubuntu-cuda:18.04

@noahstier
Copy link
Copy Markdown
Contributor Author

noahstier commented Oct 25, 2021 •
edited
Loading

This gives me much better GPU utilization for this function. I can share nvprof data if it would be helpful.

tasks:

  • proper branch: In version 3.4 this function is in opencv, not opencv_contrib. Should I also make this change in 3.4 in opencv?

  • bug report: I didn't make one--should I?. I am attaching a screenshot from nvprof where you can see the cudaDeviceSynchronize and the extra pinned memory alloc / free.

nvprof

alalek
alalek approved these changes Oct 26, 2021
View reviewed changes
Copy link
Copy Markdown
Member

@alalek alalek left a comment

Choose a reason for hiding this comment

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

Thank you for contribution 👍

3.4 branch is in the maintenance mode.
If you need to change this branch too, then you can push changes to the main repository.

@opencv-pushbot opencv-pushbot merged commit 92987f8 into opencv:master Oct 26, 2021
@noahstier noahstier mentioned this pull request Oct 27, 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: cuda optimization

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@noahstier @alalek @opencv-pushbot