RK2 loop with GPU halo support#48
Merged
Merged
Conversation
NVIDIA was correcting for our missing uhtr allocations on the GPU in the slowest way possible. This patch adds an explicit allocation, which reduced the number of transfers dramatically.
ALso: * Declate [uv]int_cor in MOM_barotropic as loop-locals * Add h to data transfer after RK2 step
Owner
Author
|
Also, we should probably not take this in until we've added a |
The default FMS is not compatible with the new GPU-based MPI methods, so we just change it in .testing/Makefile for now. Down the road, we should probably move this into the .github config. DO_LOCALITY() now returns a ; if do concurrent locality modifiers are unsupported. This prevents errors associated with line continuations.
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## dev/gpu #48 +/- ##
========================================
Coverage 37.68% 37.68%
========================================
Files 304 304
Lines 92475 92474 -1
Branches 17741 17741
========================================
Hits 34849 34849
+ Misses 51075 51074 -1
Partials 6551 6551 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Owner
Author
|
The CI is now fixed, it now points to the GPU version of FMS. |
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Uses the new GPU halo exchange support in the FMS fork.
AFAIK this does all the halo updates on (although only tested on a reentrant 1-GPU job).
This won't go in until everyone is able to build and run with the FMS GPU dev fork.
(Also d69264c was probably supposed to be in the previous PR, but it's fairly minor.)