-
Notifications
You must be signed in to change notification settings - Fork 0
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
more optimal distribution of tiles in cubed-sphere grid tile space #41
Conversation
Nightly Tests summary: **These are in slurm ** and ntasks-per-node is 40 |
- use ldas_abort() instead of stop - clarified comments - fixed typos
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@weiyuan-jiang: After I just approved the PR I noticed one more thing: It looks like elemental function rms_cs()
isn't used. If so, we should delete it. Can you please take a quick look.
PR addresses a situation where too many processors are requested for a cube-sphere tile space simulation. In this case, a process may end up with a domain that consists of a stripe that is only one grid cell wide, which causes the re-gridder to fail.
The PR ensures that each process has a stripe that is at least 2 grid cells wide.
The PR further improves the distribution of tiles onto processors.
IMPORTANT: The PR is not zero-diff for data assimilation in cube-sphere tile space. This is because the GLOBALCS/assim test case has never been zero-diff when the layout changed.
Tests are also non-zero-diff for the AGGGLOBALCS/model case, but 0-diff is not required/expected with layout changes when using "aggressive" compilation.