make GC deterministic in distributed #821
Merged
Conversation
This file contains 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
|
Testing out here https://buildkite.com/clima/climaatmos-scaling/builds/244 |
simonbyrne
force-pushed
the
sb/gc
branch
2 times, most recently
from
75610ea to
8dac62c
Compare
simonbyrne
force-pushed
the
sb/gc
branch
2 times, most recently
from
4ae134e to
8fd22df
Compare
|
I've added a small check so that GC is only called if free memory falls below a certain threshold |
|
It appears to boost scaling considerably, I think it's ready to go. Here are the low-res scaling benchmarks Before: After: |
|
bors r+ |
|
Build succeeded: |
bors bot
added a commit
that referenced
this pull request
Oct 5, 2022
891: Revert "make GC deterministic in distributed" r=szy21 a=szy21 revert #821, which may cause the failure in mpi jobs. Co-authored-by: jiahe23 <[email protected]>
bors bot
added a commit
that referenced
this pull request
Oct 5, 2022
891: Revert "make GC deterministic in distributed" r=szy21 a=szy21 revert #821, which may cause the failure in mpi jobs. Co-authored-by: jiahe23 <[email protected]>
bors bot
added a commit
that referenced
this pull request
Oct 5, 2022
891: Revert "make GC deterministic in distributed" r=szy21 a=szy21 revert #821, which may cause the failure in mpi jobs. Co-authored-by: jiahe23 <[email protected]>
Merged
simonbyrne
added a commit
that referenced
this pull request
Oct 14, 2022
Adds GC callbacks. Behavior can be controlled by an environment variable CLIMAATMOS_GC_NSTEPS (default = every 1000 steps). Unlike #821, this doesn't disable the GC, but by manually calling it we should be able to avoid the runtime needing to call it.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
PULL REQUEST
Purpose and Content
This should reduce MPI Waitall time by manually triggering the GC across all processes at the same time.
Benefits and Risks
The number of steps will require some tuning to avoid out-of-memory errors
Linked Issues
PR Checklist