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.

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

Fix for inefficient iceberg restarts #1460

Merged
merged 1 commit into from
May 9, 2024

Conversation

alex-huth
Copy link
Contributor

Changed registration of axes during iceberg restarts so that dimension lengths from each PE are gathered to only the I/O root PE, replacing the original approach of gathering to all PEs. The original approach was inefficient enough to stall runs with a large number of PEs until they timed out, and this PR fixes this issue.

How Has This Been Tested?
intel-classic/2022.0.2. Confirmed that the restart files produced with/without this commit are identical

Checklist:

  • My code follows the style guidelines of this project
  • I have performed a self-review of my own code
  • I have commented my code, particularly in hard-to-understand areas
  • I have made corresponding changes to the documentation
  • My changes generate no new warnings
  • Any dependent changes have been merged and published in downstream modules
  • New check tests, if applicable, are included
  • make distcheck passes

register_unlimited_compressed_axis. Here, the iceberg
dimension length for each PE was being sent to all other PEs, an
operation inefficient enough to cause runs with a large number of PEs
to appear to stall until they timed out. However,
these local dimension lengths actually only need to be gathered on the
I/O root PE. Making this change fixed the issue.
@rem1776 rem1776 merged commit 42f8506 into NOAA-GFDL:main May 9, 2024
17 of 19 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants