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

Fixes for C12 MOM6 #285

Closed
wants to merge 7 commits into from
Closed

Conversation

mathomp4
Copy link
Member

@mathomp4 mathomp4 commented Jan 4, 2022

This PR updates the setup scripts for the C12 MOM6 setup that @sanAkel and @yvikhlya have developed. It includes:

  • Fixes some csh logic for making the tile names
  • Remove setenv I_MPI_DAPL_UD enable. Intel MPI doesn't even recognize this flag anymore:
    I_MPI_DAPL_UD variable has been removed from the product, its value is ignored
    
  • Remove the ROMIO environment variable for Intel MPI
  • Default C12 MOM6 as 1x6 atmos/3x2 ocean. Note that C12 data ocean will still be 2x12 by default
  • Removes the symlink and binarytile.x runs for tile_hist.data. Tests by @bena-nasa and myself seem to show that MAPL_Tripolar.nc coupled with the usual History grid_label:
    geosgcm_ocn2d.template:  '%y4%m2%d2_%h2%n2z.nc4',
    geosgcm_ocn2d.archive:   '%c/Y%y4',
    geosgcm_ocn2d.mode:      'time-averaged',
    geosgcm_ocn2d.format: 'CFIO',
    geosgcm_ocn2d.frequency:  060000,
    geosgcm_ocn2d.grid_label: PC360x181-DC,
    geosgcm_ocn2d.fields:    'UW'    , 'MOM6', 'US'
                     'VW'    , 'MOM6', 'VS'
                     'TW'    , 'MOM6', 'TS'
                     'SW'    , 'MOM6', 'SS'
                        ::
    
    works. And it seems to make usable data:

US_in_stock-gcm-2022Jan04-1day-c90-MOM6 geosgcm_

NOTE1: Keeping this draft until I make sure I don't break C90 MOM6 as well as make sure all works on my laptop

NOTE2: At this point I think you can only run C12 MOM6 with my TinyBCs. I believe @yvikhlya needs to copy some files to a new /discover/nobackup/yvikhlia/coupled/Forcings/MOM6/CF0012x6C_TM0072xTM0036 directory that I can provide.

@mathomp4 mathomp4 added the 0 diff The changes in this pull request have verified to be zero-diff with the target branch. label Jan 4, 2022
@mathomp4 mathomp4 self-assigned this Jan 4, 2022
@mathomp4 mathomp4 marked this pull request as ready for review January 4, 2022 21:31
@mathomp4 mathomp4 requested a review from a team as a code owner January 4, 2022 21:31
@sanAkel
Copy link
Collaborator

sanAkel commented Feb 2, 2022

NOTE2: At this point I think you can only run C12 MOM6 with my TinyBCs. I believe @yvikhlya needs to copy some files to a new /discover/nobackup/yvikhlia/coupled/Forcings/MOM6/CF0012x6C_TM0072xTM0036 directory that I can provide.

I think this has happened.
@mathomp4 would you please check if ALL still works at your end?

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 2, 2022

@sanAkel I still do not see a CF0012x6C_TM0072xTM0036 directory in:

/discover/nobackup/yvikhlia/coupled/Forcings/MOM6/

I think until that's done, C12 MOM6 will only work with TinyBCs.

@yvikhlya
Copy link

yvikhlya commented Feb 2, 2022

@sanAkel I still do not see a CF0012x6C_TM0072xTM0036 directory in:

/discover/nobackup/yvikhlia/coupled/Forcings/MOM6/

I think until that's done, C12 MOM6 will only work with TinyBCs.

All coupled bc files are moved to /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs. I keep the old files at their old locations for now because people are running old experiments with them.

@sanAkel
Copy link
Collaborator

sanAkel commented Feb 2, 2022

@sanAkel I still do not see a CF0012x6C_TM0072xTM0036 directory in:

/discover/nobackup/yvikhlia/coupled/Forcings/MOM6/

I think until that's done, C12 MOM6 will only work with TinyBCs.

@mathomp4 I am not sure when and how @yvikhlya is handling all these ... let's please proceed with TinyBCs can @sdrabenh work with TinyBCs ?

@sdrabenh
Copy link
Collaborator

sdrabenh commented Feb 2, 2022

can @sdrabenh work with TinyBCs ?

I am not familiar with running anything at C12. Whatever you guys decide is fine.

@sanAkel
Copy link
Collaborator

sanAkel commented Feb 2, 2022

can @sdrabenh work with TinyBCs ?

I am not familiar with running anything at C12. Whatever you guys decide is fine.

@sdrabenh you always ease the pain! 😄

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 2, 2022

@sanAkel I still do not see a CF0012x6C_TM0072xTM0036 directory in:

/discover/nobackup/yvikhlia/coupled/Forcings/MOM6/

I think until that's done, C12 MOM6 will only work with TinyBCs.

All coupled bc files are moved to /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs. I keep the old files at their old locations for now because people are running old experiments with them.

@yvikhlya The problem is that I don't think:

/discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus-NLv3/MOM6/CF0012x6C_TM0072xTM0036 

has all the needed files. For example, when I try to run with that as the GRIDDIR I have five broken symlinks in the scratch/ directory:

lrwxrwxrwx 1 mathomp4 g0620  126 Feb  2 15:45 SEAWIFS_KPAR_mon_clim.data -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus-NLv3/MOM6/CF0012x6C_TM0072xTM0036/SEAWIFS_KPAR_mon_clim.72x36
lrwxrwxrwx 1 mathomp4 g0620  115 Feb  2 15:45 MAPL_Tripolar.nc -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus-NLv3/MOM6/CF0012x6C_TM0072xTM0036/MAPL_Tripolar.nc
lrwxrwxrwx 1 mathomp4 g0620  112 Feb  2 15:45 vgrid.ascii -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus-NLv3/MOM6/CF0012x6C_TM0072xTM0036/vgrid50.ascii
...
lrwxrwxrwx 1 mathomp4 g0620  116 Feb  2 15:45 kmt_cice.bin -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus-NLv3/MOM6/CF0012x6C_TM0072xTM0036/cice/kmt_cice.bin
lrwxrwxrwx 1 mathomp4 g0620  117 Feb  2 15:45 grid_cice.bin -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus-NLv3/MOM6/CF0012x6C_TM0072xTM0036/cice/grid_cice.bin

These are in the TinyBCs set that @sanAkel provided me.

I then based the TinyBCs directory structure on that of:

/discover/nobackup/yvikhlia/coupled/Forcings/MOM6/CF0090x6C_TM0360xTM0210/

where there is a cice/ directory and the other three files.

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 2, 2022

Note that this PR is still safe as long as anyone that wants to run C12 Newland MOM6, runs with my TinyBCs. It will still work fine for C90 OldLand MOM6.

@sanAkel
Copy link
Collaborator

sanAkel commented Feb 2, 2022

Note that this PR is still safe as long as anyone that wants to run C12 Newland MOM6, runs with my TinyBCs. It will still work fine for C90 OldLand MOM6.

Thanks @mathomp4 Let me think about this a bit ...

@yvikhlya
Copy link

yvikhlya commented Feb 2, 2022

@mathomp4 I split ocean and agcm bcs, since ocean bcs do not depend on agcm resolution and land model version, there is no need to copy them under every configuration. There is no GRIDDIR anymore, new locations are

/discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs
/discover/nobackup/projects/gmao/ssd/aogcm/ocean_bcs

I updated gcm_run.j and gcm_setup to work with new locations on the feature/aogcm branch, but this is not merged with develop yet.

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 2, 2022

@mathomp4 I split ocean and agcm bcs, since ocean bcs do not depend on agcm resolution and land model version, there is no need to copy them under every configuration. There is no GRIDDIR anymore, new locations are

/discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs
/discover/nobackup/projects/gmao/ssd/aogcm/ocean_bcs

I updated gcm_run.j and gcm_setup to work with new locations on the feature/aogcm branch, but this is not merged with develop yet.

@yvikhlya Ahh. Okay. If you don't mind, I can take a look at that branch and then merge the GRIDDIR to new dirs changes into my PR here?

ETA: Ooh. I also need to work with @bena-nasa to make sure we have all that data in our location at NAS!

@yvikhlya
Copy link

yvikhlya commented Feb 2, 2022

@yvikhlya Ahh. Okay. If you don't mind, I can take a look at that branch and then merge the GRIDDIR to new dirs changes into my PR here?

Oh yes, you are welcomed to do this. Instead of GRIDDIR, there are OBCS and ABCS variables now. Not sure if I did it the best way, so if you feel something needs to be tweaked, feel free to do this.

@mathomp4 mathomp4 added the Contingent - DNA Do Not Approve (DNA). These changes are contingent on other PRs label Feb 3, 2022
@mathomp4
Copy link
Member Author

mathomp4 commented Feb 3, 2022

I'm blocking this until I can get in the updates from @yvikhlya. Hopefully unblocking tomorrow

@mathomp4 mathomp4 marked this pull request as draft February 3, 2022 21:47
yvikhlya
yvikhlya previously approved these changes Feb 3, 2022
@yvikhlya
Copy link

yvikhlya commented Feb 3, 2022

I'm blocking this until I can get in the updates from @yvikhlya. Hopefully unblocking tomorrow

What updates exactly?

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 4, 2022

I'm blocking this until I can get in the updates from @yvikhlya. Hopefully unblocking tomorrow

What updates exactly?

The replacement for GRIDDIR. I need to make sure I get the merge of changes over correctly. (since feature/aogcm still uses the >>><<< style instead of the @-style for options preferred by Ricardo).

@yvikhlya
Copy link

yvikhlya commented Feb 4, 2022

The replacement for GRIDDIR.

It is split into

/discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs
/discover/nobackup/projects/gmao/ssd/aogcm/ocean_bcs

Am I getting wrong something?

EDIT: Note, that these directories are not copied to NAS currently.
EDIT2: Matt, it looks like you get everything right in your commit. What I could be missing?

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 4, 2022

The replacement for GRIDDIR.

It is split into

/discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs
/discover/nobackup/projects/gmao/ssd/aogcm/ocean_bcs

Am I getting wrong something?

EDIT: Note, that these directories are not copied to NAS currently. EDIT2: Matt, it looks like you get everything right in your commit. What I could be missing?

Yury,

I found some files are missing at NCCS. I pinged on Teams, but for history's sake:

  1. Some updates are needed to your atmosphere_bcs directory for Icarus Old Land C90 TM360. The model is expecting files like visdf_90x540.dat but they aren't there. These are all broken symlinks:
visdf.dat -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus/MOM6/CF0090x6C_TM0360xTM0210/visdf_90x540.dat
nirdf.dat -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus/MOM6/CF0090x6C_TM0360xTM0210/nirdf_90x540.dat
vegdyn.data -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus/MOM6/CF0090x6C_TM0360xTM0210/vegdyn_90x540.dat
lai.data -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus/MOM6/CF0090x6C_TM0360xTM0210/lai_clim_90x540.data
green.data -> /discover/nobackup/projects/gmao/ssd/aogcm/atmosphere_bcs/Icarus/MOM6/CF0090x6C_TM0360xTM0210/green_clim_90x540.data

I think your directory only has them without the "90x540"

  1. I think the NLv3 C12 ocean_bcs has some missing files. These two files are in my TinyBCs from Santha, but not in your directory:
/gpfsm/dnb44/mathomp4/TinyBCs-GitV10/ocean_bcs/MOM6/72x36/INPUT/grid_spec.nc
/gpfsm/dnb44/mathomp4/TinyBCs-GitV10/ocean_bcs/MOM6/72x36/INPUT/ocean_mosaic.nc

@yvikhlya
Copy link

yvikhlya commented Feb 4, 2022

@mathomp4 Could you check again? Let me know if anything is still missing.

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 4, 2022

@mathomp4 Could you check again? Let me know if anything is still missing.

@yvikhlya I think we got it. It seems to work for me. I also fixed up TinyBCs to, I think, work with the new ABCSDIR and OBCSDIR. Though testing will find that out...

@mathomp4 mathomp4 marked this pull request as ready for review February 4, 2022 18:29
@mathomp4 mathomp4 removed the Contingent - DNA Do Not Approve (DNA). These changes are contingent on other PRs label Feb 4, 2022
@mathomp4 mathomp4 requested a review from yvikhlya February 4, 2022 18:34
@mathomp4
Copy link
Member Author

mathomp4 commented Feb 4, 2022

Sigh. Or not. Dang it. I messed up something...

@mathomp4 mathomp4 added the Contingent - DNA Do Not Approve (DNA). These changes are contingent on other PRs label Feb 4, 2022
@mathomp4
Copy link
Member Author

mathomp4 commented Feb 4, 2022

@yvikhlya I think one more directory still missing. When I try to run C90 360x210 (which is my "usual" MOM6 run) I see these broken symlinks:

kmt_cice.bin -> /discover/nobackup/projects/gmao/ssd/aogcm/ocean_bcs/MOM6/360x210/cice/kmt_cice.bin
grid_cice.bin -> /discover/nobackup/projects/gmao/ssd/aogcm/ocean_bcs/MOM6/360x210/cice/grid_cice.bin

Indeed, the 360x210 case isn't in ocean_bcs:

$ ls -l /discover/nobackup/projects/gmao/ssd/aogcm/ocean_bcs/MOM6/
total 4
drwxr-xr-x 6 yvikhlia g0609 4096 Sep 16 12:12 1440x1080/
drwxr-xr-x 6 yvikhlia g0609 4096 Jan 31 13:01 1440x1080_newtopo/
drwxr-xr-x 6 yvikhlia g0609 4096 Oct 13 12:19 72x36/
-rw-r--r-- 1 yvikhlia g0609 3160 Jan 20 12:01 generate_topo.md

@yvikhlya
Copy link

yvikhlya commented Feb 4, 2022

@mathomp4 This is because MOM6 c90x360x210 does not work with new land BCs and I thought we are going to abandon it. But I can copy it to new location to use with old land BCs, hold on.

EDIT: Done.

@mathomp4
Copy link
Member Author

mathomp4 commented Feb 4, 2022

@mathomp4 This is because MOM6 c90x360x210 does not work with new land BCs and I thought we are going to abandon it. But I can copy it to new location to use with old land BCs, hold on.

EDIT: Done.

@yvikhlya Thanks. I'll retest as soon as I get nodes.

And, yes, it is sort of "abandoned", but I can only really run two cases now. C90 Oldland and C12 Newland. Probably good to keep the old girl going until I can transition to another "mid-resolution" case that you or @sanAkel can provide me. 😄

@sanAkel
Copy link
Collaborator

sanAkel commented Feb 4, 2022

@mathomp4 This is because MOM6 c90x360x210 does not work with new land BCs and I thought we are going to abandon it. But I can copy it to new location to use with old land BCs, hold on.
EDIT: Done.

@yvikhlya Thanks. I'll retest as soon as I get nodes.

And, yes, it is sort of "abandoned", but I can only really run two cases now. C90 Oldland and C12 Newland. Probably good to keep the old girl going until I can transition to another "mid-resolution" case that you or @sanAkel can provide me. 😄

@mathomp4 There are a few things already queued/piled up. Building about a 1-deg ocean case is one of them, but not at the top right now. However, the 0.25-deg case is, which is (from a non oceanographer perspective) at the "high" end of resolution spectrum! Meanwhile, enjoy the C12 x 5-deg!

@mathomp4 mathomp4 removed the Contingent - DNA Do Not Approve (DNA). These changes are contingent on other PRs label Feb 7, 2022
@mathomp4
Copy link
Member Author

mathomp4 commented Feb 7, 2022

Removing the blocker. It looks like all is well with this. C12 runs. @bena-nasa and I need to do some BCs work at NAS, but I think we are good.

@mathomp4
Copy link
Member Author

Closing in favor of #292

@mathomp4 mathomp4 closed this Feb 10, 2022
@mathomp4 mathomp4 deleted the feature/mathomp4/fix-app-for-c12-mom6 branch February 10, 2022 15:38
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
0 diff The changes in this pull request have verified to be zero-diff with the target branch.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants