Skip to content
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

Fix performance for CUDA >= 9.2 (GT v1.0-branch) #1326

Merged
merged 17 commits into from
Jul 8, 2019
Merged

Fix performance for CUDA >= 9.2 (GT v1.0-branch) #1326

merged 17 commits into from
Jul 8, 2019

Conversation

lukasm91
Copy link
Contributor

@lukasm91 lukasm91 commented Jul 1, 2019 •
edited
Loading

  • sets GT_CONSTEXPR to constexpr for nvcc
  • introduces const_ref to fix performance for CUDA >= 9.2 (T for small data types, T const& for large data types)

Backport to GridTools 1.0. See also #1326 and 9bb2d64

@lukasm91 lukasm91 changed the title Fix performance for CUDA >= 9.2 Fix performance for CUDA >= 9.2 (GT v1.0-branch) Jul 1, 2019
@lukasm91 lukasm91 mentioned this pull request Jul 4, 2019
Merged
@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 5, 2019

launch strgrid

@lukasm91 lukasm91 changed the title Fix performance for CUDA >= 9.2 (GT v1.0-branch) Fix performance for CUDA >= 9.2 (GT v1.0-branch) [WIP] Jul 5, 2019
@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 5, 2019

launch strgrid

@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 5, 2019

launch icgrid

@lukasm91 lukasm91 requested a review from anstaf July 5, 2019 06:29
@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 5, 2019

launch strgrid

@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 5, 2019

dycore performance currently ok

Lukas Mosimann and others added 2 commits July 5, 2019 13:19
@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 5, 2019

launch strgrid

@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 8, 2019

launch strgrid

@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 8, 2019

launch strgrid

@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 8, 2019

launch strgrid

@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 8, 2019

launch icgrid

@lukasm91 lukasm91 changed the title Fix performance for CUDA >= 9.2 (GT v1.0-branch) [WIP] Fix performance for CUDA >= 9.2 (GT v1.0-branch) Jul 8, 2019
@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 8, 2019

Verified performance with cosmo.

@havogt
Copy link
Contributor

havogt commented Jul 8, 2019

Does it need the review from @anstaf or are the changes the same as in master?

@lukasm91
Copy link
Contributor Author

lukasm91 commented Jul 8, 2019

Maybe he can look at const_ref, because of the CUDA 8 workaround. I already wrote him on slack earlier this day

@anstaf anstaf merged commit f97ee7d into GridTools:release_v1.0 Jul 8, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@anstaf anstaf anstaf approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@lukasm91 @havogt @anstaf