Fms2 io implementation #74

laurenchilutti · 2021-03-11T18:21:06Z

Description
This pull request implements fms2io into the GFDL_atmos_cubed_sphere code. In a coming release of FMS, the modules mpp_io_mod and fms_io_mod are being deprecated and replaced by the new fms2_io_mod. These changes go hand in hand with the changes to the physics in PR6.

Fixes # (issue)

How Has This Been Tested?
These changes have been tested with SHiELD and do not change answers. Using FMS 2021.02-beta2 tag and SHiELD_driver laurenchilutti:user/lec/fms2io
Specific tests run are:

C48 grid test with coarse grained restarts and intermediate restarts and with external_ic=true and nggps_ic=true (No nests)
C48 grid test with coarse grained restarts and intermediate restarts and warm start from RESTARTS (No nests)
C48 grid test with external_ic=true and nggps_ic=true (No nests)
C48 grid test with external_ic=true and nggps_ic=true and warm start from RESTARTS (No nests)
C48 grid test with external_ic=true and nggps_ic=true and 1 nested grid
C768 grid test with external_ic=true and nggps_ic=true (No nests)
C768 grid test with external_ic=true and nggps_ic=true and 1 nested grid
Regional 3km test with external_ic=true and nggps_ic=true (No nests)

Checklist:

Please check all whether they apply or not

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

…cing the necessary functions with fms2_io_mod functions

…name when needed and removing unneccessary code in fv_io_mod

lharris4 · 2021-03-11T19:52:10Z

Hi, Lauren. Thanks. Previously we had seen significant slowdowns in SHiELD when using FMS2io. Have these been fixed? Lucas

…

laurenchilutti · 2021-03-11T20:22:20Z

Lucas - I can run some tests and compare the timing.

bensonr · 2021-03-11T20:29:06Z

The performance degradation you are talking about was a fundamental issue initially that has since been rectified. In fact, anything you are running today with a recent release of FMS (2020.03 and newer) has the diagnostics being output via the fms2_io interfaces. If you aren't seeing runtime impacts at the end of a simulation when all diagnostic data is being pushed to storage, it is unlikely to occur with the restarts. Regardless, a few timing tests will be run to prove it.

bensonr · 2021-04-22T12:26:13Z

@laurenchilutti - This PR was opened prior to a template being in place. There is a lot of good work here, but I want to make sure we've tested to the largest extent that we can. Could you list the tests you've run with the fms2_io to give an idea of the code coverage? In particular, it would be good to list the IC type (GFS, HRRR, ECMWF, SHiELD), model type (nested, regional, etc.) and whether restarts (intermediate writing and functional test) were a part of the testing as well. Also, are there changes coming/needed within the SHiELD_driver project?

bensonr

Can you remove any references to the use association of write_version_number from fms_io_mod. I think the only place it appears is in tools/fv_diagnostics.F90 where it is also use associated via fms_mod.

laurenchilutti · 2021-04-28T16:19:27Z

@bensonr My apologies for not directly responding to the comment from 6 days ago. I am putting a list of all tests run and will update the description of this PR to include that. SHiELD_drivers will also require updates to fms2io that has not been finished yet. For now, the branch: user/lec/fms2io from my fork is needed to test fms2io with SHiELD.

I will go ahead and clean up the use of write_version_mod.

lharris4 · 2021-04-28T16:20:52Z

Hi, Lauren. Thanks for working on this. Lucas

…

On Wed, Apr 28, 2021 at 12:19 PM laurenchilutti ***@***.***> wrote: @bensonr <https://github.com/bensonr> My apologies for not directly responding to the comment from 6 days ago. I am putting a list of all tests run and will update the description of this PR to include that. SHiELD_drivers will also require updates to fms2io that has not been finished yet. For now, the branch: user/lec/fms2io <https://github.com/laurenchilutti/SHiELD_driver/tree/user/lec%2Ffms2io> from my fork is needed to test fms2io with SHiELD. I will go ahead and clean up the use of write_version_mod. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#74 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMUQRVFI4GDZVDGM6KLQ2KLTLAYSHANCNFSM4ZAY7JPQ> .

laurenchilutti · 2021-04-28T16:25:05Z

Can you remove any references to the use association of write_version_number from fms_io_mod. I think the only place it appears is in tools/fv_diagnostics.F90 where it is also use associated via fms_mod.

@bensonr I do not see any use statements from fms_io_mod anywhere in this new code. When you look at tools/fv_diagnostics.F90 it used to use write_version_number from fms_io_mod, but I removed and replaced that with fms_mod. I may be misunderstanding your request.

bensonr · 2021-04-28T16:36:20Z

@laurenchilutti thanks for confirming. I was looking at the current code and didn't realize you'd fixed it as part of this PR.

tools/external_ic.F90

bensonr · 2021-04-29T18:38:54Z

@laurenchilutti - do you feel ready for a formal review?

laurenchilutti · 2021-04-30T12:17:09Z

@bensonr Yes, I am ready for a formal review. I updated the description to include testing completed. Note that I have not been able to test a nested run warm-starting from restarts - I am working with Joseph on this.

bensonr · 2021-04-30T12:20:27Z

@bensonr <https://github.com/bensonr> Yes, I am ready for a formal review. I updated the description to include testing completed. Note that I have not been able to test a nested run warm-starting from restarts - I am working with Joseph on this.

We can add that as a note to the CHANGELOG.md for the updates.

model/fv_regional_bc.F90

GFDL_tools/fv_ada_nudge.F90

tools/external_ic.F90

tools/fv_restart.F90

tools/fv_io.F90

…, addressding Uriel's comments from PR#74

uramirez8707

Thanks @laurenchilutti. It all looks good to me.

lharris4

This looks good. I am really glad we can get rid of the INTERNAL_FILE_NML. One question: has nggps_ic been tested with these changes?

bensonr · 2021-05-06T13:52:15Z

@lharris4 - there is a "How this has been tested" section in the description of the PR (thanks to the PR template we appropriated from the NOAA-GFDL/FMS project). Please let us know if there is untested functionality that needs to be addressed.

bensonr

I believe you still have an instance of INTERNAL_FILE_NML in fv_diag_column.F90

GFDL_tools/fv_ada_nudge.F90

GFDL_tools/fv_diag_column.F90

…ter_axis in fv_ada_nudge. Reintroducing flush function and updating namelist read in fv_diag_column

laurenchilutti added 2 commits March 5, 2021 14:38

Removing use of mpp_io_mod and fms_io_mod from the dycore code. Repla…

ca1066f

…cing the necessary functions with fms2_io_mod functions

Adding a call to set_filename_appendix so that nest is added to file…

c4d0928

…name when needed and removing unneccessary code in fv_io_mod

laurenchilutti added 2 commits April 21, 2021 10:44

Updating dimension string arrays to address GNU compilation errors.

fe683c8

Merge branch 'master' into fms2_io_implementation

fb96053

laurenchilutti force-pushed the fms2_io_implementation branch from 9063a74 to fb96053 Compare April 28, 2021 13:18

bensonr requested changes Apr 28, 2021

View reviewed changes

bensonr self-requested a review April 28, 2021 16:39

bensonr reviewed Apr 28, 2021

View reviewed changes

tools/external_ic.F90 Outdated Show resolved Hide resolved

tools/external_ic.F90 Outdated Show resolved Hide resolved

tools/external_ic.F90 Outdated Show resolved Hide resolved

prepending INPUT/ to hrr_ctrl.nc and gfs_ctrl.nc in external_ic

04a6d6e

laurenchilutti mentioned this pull request Apr 30, 2021

Removing a duplicate use statement of FMS function "write_version_number" #98

Merged

6 tasks

bensonr requested review from uramirez8707 and bensonr May 3, 2021 15:48

Merge branch 'master' into fms2_io_implementation

22fa998

uramirez8707 suggested changes May 3, 2021

View reviewed changes

Adding pelist argument to some open_file calls, correcting some typos…

9746100

…, addressding Uriel's comments from PR#74

uramirez8707 approved these changes May 4, 2021

View reviewed changes

laurenchilutti requested a review from lharris4 May 5, 2021 11:58

lharris4 reviewed May 6, 2021

View reviewed changes

bensonr reviewed May 7, 2021

View reviewed changes

GFDL_tools/fv_ada_nudge.F90 Outdated Show resolved Hide resolved

GFDL_tools/fv_ada_nudge.F90 Show resolved Hide resolved

GFDL_tools/fv_diag_column.F90 Show resolved Hide resolved

laurenchilutti mentioned this pull request May 7, 2021

FMS2 io updates NOAA-GFDL/SHiELD_physics#6

Merged

correcting minor typo and using fv_io_register_axis rather than regis…

35ee073

…ter_axis in fv_ada_nudge. Reintroducing flush function and updating namelist read in fv_diag_column

bensonr approved these changes May 7, 2021

View reviewed changes

lharris4 approved these changes May 7, 2021

View reviewed changes

laurenchilutti merged commit 5d8ef5e into NOAA-GFDL:master May 10, 2021

binli2337 added a commit to binli2337/GFDL_atmos_cubed_sphere that referenced this pull request Oct 20, 2022

Fms mixed mode (NOAA-GFDL#74)

dcff74d

mathomp4 mentioned this pull request Dec 6, 2024

Help with mixed-precision FMS (in GEOS) NOAA-GFDL/FMS#1621

Closed

mathomp4 mentioned this pull request Feb 7, 2025

FMS 2025 does not support our fvdycore GEOS-ESM/GFDL_atmos_cubed_sphere#108

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fms2 io implementation #74

Fms2 io implementation #74

laurenchilutti commented Mar 11, 2021 •

edited

Loading

lharris4 commented Mar 11, 2021 via email

laurenchilutti commented Mar 11, 2021

bensonr commented Mar 11, 2021 via email

bensonr commented Apr 22, 2021

bensonr left a comment

laurenchilutti commented Apr 28, 2021

lharris4 commented Apr 28, 2021 via email

laurenchilutti commented Apr 28, 2021

bensonr commented Apr 28, 2021

bensonr commented Apr 29, 2021

laurenchilutti commented Apr 30, 2021

bensonr commented Apr 30, 2021 via email

uramirez8707 left a comment

lharris4 left a comment

bensonr commented May 6, 2021

bensonr left a comment

Fms2 io implementation #74

Fms2 io implementation #74

Conversation

laurenchilutti commented Mar 11, 2021 • edited Loading

lharris4 commented Mar 11, 2021 via email

laurenchilutti commented Mar 11, 2021

bensonr commented Mar 11, 2021 via email

bensonr commented Apr 22, 2021

bensonr left a comment

Choose a reason for hiding this comment

laurenchilutti commented Apr 28, 2021

lharris4 commented Apr 28, 2021 via email

laurenchilutti commented Apr 28, 2021

bensonr commented Apr 28, 2021

bensonr commented Apr 29, 2021

laurenchilutti commented Apr 30, 2021

bensonr commented Apr 30, 2021 via email

uramirez8707 left a comment

Choose a reason for hiding this comment

lharris4 left a comment

Choose a reason for hiding this comment

bensonr commented May 6, 2021

bensonr left a comment

Choose a reason for hiding this comment

laurenchilutti commented Mar 11, 2021 •

edited

Loading