search

ESMCI / cime

Common Infrastructure for Modeling the Earth
http://esmci.github.io/cime
Other
161 stars 206 forks source link

Set PIO_REARRANGER_LND to 2 for single point cases #4257

Closed ekluzek closed 1 year ago

ekluzek commented 2 years ago

We have found this needs to be set for NEON single point cases in ctsm5.1.dev095 which uses cime5.0.26. In some cases other single point cases already seem to have it set. It does seem to be required or the case will fail in some cases and in others just go really slow.

See the CTSM issue about this:

https://github.com/ESCOMP/CTSM/issues/1743

We do have single point tests in the aux_clm test suite that seem to run fine without this being set. See the ctsm5.1.dev096 tests for example on cheyenne under:

/glade/p/cgd/tss/ctsm_baselines/ctsm5.1.dev096/1x1

ekluzek commented 2 years ago

@jedwards4b do you have any thoughts on this? Does it make sense to you that this would be required for single point cases? Does this also mean this should be done for SCAM as well? Or any other single grid cell configurations?

jedwards4b commented 2 years ago

@ekluzek - this should not be required and there should be no difference in performance for a single column case.

ekluzek commented 2 years ago

OK, thanks for getting back to me. I was hoping this would make sense to you as something we'd need to do. As that's not the case, I suspect this is likely a bug in pio2_5_7 then. We started seeing this a problem in ctsm5.1.dev095 where PIO was updated from pio2_5_6 to pio2_5_7.

briandobbins commented 2 years ago

My apologies - this was likely because we were trying to do something in realtime, and Adrianna thought her run was going extremely slow. I looked at the dt in the log, it looked fast, and assumed, since this was a single-core, NetCDF run, that perhaps the problem was the same as the NEON runs. Before I could investigate it further, there was an issue, and you two digging into it, which I only saw after getting back from walking my dog.

It seems this was a wild goose chase that I erroneously and unintentionally sent you on!

On Wed, May 18, 2022 at 5:05 PM Erik Kluzek @.***> wrote:

OK, thanks for getting back to me. I was hoping this would make sense to you as something we'd need to do. As that's not the case, I suspect this is likely a bug in pio2_5_7 then. We started seeing this a problem in ctsm5.1.dev095 where PIO was updated from pio2_5_6 to pio2_5_7.

— Reply to this email directly, view it on GitHub https://github.com/ESMCI/cime/issues/4257#issuecomment-1130691984, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACL2HPJOIDB6SYPBMJZYG4TVKVZS5ANCNFSM5WJ7PIJQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>

ekluzek commented 2 years ago

@briandobbins from @jedwards4b answer it sounds like it should be NOT be required for the NEON case either. But, I do have cases where it is required for the NEON case. So I do think this indicates somethings wrong here, it's just not clear what it is. If you can investigate and figure out what this means that would be great.

github-actions[bot] commented 1 year ago

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

ekluzek commented 1 year ago

We removed the use of this in CTSM in this commit, 

commit e855ee41a58a5033fc97adc531865635abbcace0
Author: Erik Kluzek <erik@ucar.edu>
Date:   Fri Nov 4 15:05:36 2022 -0600

So there seems to be no reason for this anymore. So closing as invalid.