search

EarthWorksOrg / EarthWorks

Other
3 stars 3 forks source link

Update externals based on CESM tag 'cesm2_3_alpha16d' and add default files #17

Closed gdicker1 closed 1 year ago

gdicker1 commented 1 year ago

Advance the tags used in Externals.cfg to match with the versions used in ESCOMP/CESM tag 'cesm2_3_alpha16d'.

EXCEPT: cam based on cam6_3_124 and not cam6_3_125

gdicker1 commented 1 year ago

Related Pull Requests in externals:

This updates the basis of the externals we share with CESM like so:

gdicker1 commented 1 year ago

Added default file changes for mpas-o and mpas-si to this PR:

dazlich commented 1 year ago

Intel, gnu, and nvhpc compilers work for the fully coupled case at 120km if:

    • I modify the long hemco line;
    • I comment out the omp directives in seaice file src/shared/mpas_seaice_velocity_solver_constitutive_relation.F

I really want to get the second issue resolved for V2. It's not a problem with this PR - I've been commenting those lines out for a while.

I'm gonna run a couple tests at 15km, fully coupled. I expect them to run, and then I will approve this.

dazlich commented 1 year ago

Oh, and as this has my mpaso/mpassi default file updates, approval for those is implicit in approval for this PR.

gdicker1 commented 1 year ago

Intel, gnu, and nvhpc compilers work for the fully coupled case at 120km if:

    • I modify the long hemco line;
    • I comment out the omp directives in seaice file src/shared/mpas_seaice_velocity_solver_constitutive_relation.F

I really want to get the second issue resolved for V2. It's not a problem with this PR - I've been commenting those lines out for a while.

I'm gonna run a couple tests at 15km, fully coupled. I expect them to run, and then I will approve this.

Understood and noted on the Intel OpenMP issue. I agree that we want few/none "User, before building, modify these files..." instructions for the V2 release.

I want to make sure I can setup a standard FullyCoupled run on Derecho and I'm good with it too.

dazlich commented 1 year ago

You added defaults for fsurdat in clm, right? I'm having to specify it.

dazlich commented 1 year ago

Looks like what I'm testing only has cam defaults for 120km. I guess there will be future PRs for cam and clm default files?

120km looking good across the board and I approve! Thanks

gdicker1 commented 1 year ago

You added defaults for fsurdat in clm, right? I'm having to specify it.

Looks like what I'm testing only has cam defaults for 120km. I guess there will be future PRs for cam and clm default files?

Apologies, I didn't get these added this afternoon. They should be good to go now.

Added PRs:

dazlich commented 1 year ago

There’s a bug at line 273 of namelist_defaults_cam.xml At the beginning of the line should be ncdata, not data

On Sep 15, 2023, at 6:07 PM, gdicker1 @.***> wrote:

Caution: EXTERNAL Sender

You added defaults for fsurdat in clm, right? I'm having to specify it.

Looks like what I'm testing only has cam defaults for 120km. I guess there will be future PRs for cam and clm default files?

Apologies, I didn't get these added this afternoon. They should be good to go now.

Added PRs:

— Reply to this email directly, view it on GitHubhttps://github.com/EarthWorksOrg/EarthWorks/pull/17#issuecomment-1722069905, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADS3XWCZRDW6WOOF3NA3FJ3X2TUTBANCNFSM6AAAAAA4PR27D4. You are receiving this because your review was requested.Message ID: @.***>

dazlich commented 1 year ago

Not a problem with the code in the pull request, but it looks like the lnd surfdata files needed to be moved into the ctsm5.1.dev052 subdirectory so I did that and all is now good at 120km.

Don

On Sep 15, 2023, at 6:07 PM, gdicker1 @.***> wrote:

Caution: EXTERNAL Sender

You added defaults for fsurdat in clm, right? I'm having to specify it.

Looks like what I'm testing only has cam defaults for 120km. I guess there will be future PRs for cam and clm default files?

Apologies, I didn't get these added this afternoon. They should be good to go now.

Added PRs:

— Reply to this email directly, view it on GitHubhttps://github.com/EarthWorksOrg/EarthWorks/pull/17#issuecomment-1722069905, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADS3XWCZRDW6WOOF3NA3FJ3X2TUTBANCNFSM6AAAAAA4PR27D4. You are receiving this because your review was requested.Message ID: @.***>

dazlich commented 1 year ago

Is there ccs_config code involved in updating to alpha16d? If not we need to edit config_machines.xml to get pio2.6.2. It currently has 2.6.0 and cam history is going too slow at 30km.

gdicker1 commented 1 year ago

I see what you're talking about with modules. No ccs_config tag I've seen updates that module beyond 2.6.0. I know we're picking up pio/2.6.2 in our externals now, but I don't know how that is used in our builds.

sherimickelson commented 1 year ago

I noticed that on the NCAR machines, the esmf module also loads pio. You could double check the esmf version that cime is bringing in to see which version of pio is being used. I think there's also a script you can run in a cesm case directory to see which modules are being used.

dazlich commented 1 year ago

All I can say is that I had to edit config_machines.xml to get cam history write in a reasonable time. Perhaps specifying pio2.6.0 in the config file supersedes what ESMF want to do?

This is the config_machines.xml file at the moment:

  <modules DEBUG="TRUE">

    <command name="load">parallelio/2.6.0-debug</command>

    <command name="load">esmf/8.6.0b03-debug</command>

  </modules>

  <modules DEBUG="FALSE">

    <command name="load">parallelio/2.6.0</command>

    <command name="load">esmf/8.6.0b03</command>

  </modules>

On Sep 18, 2023, at 12:06 PM, sherimickelson @.***> wrote:

Caution: EXTERNAL Sender

I noticed that on the NCAR machines, the esmf module also loads pio. You could double check the esmf version that cime is bringing in to see which version of pio is being used. I think there's also a script you can run in a cesm case directory to see which modules are being used.

— Reply to this email directly, view it on GitHubhttps://github.com/EarthWorksOrg/EarthWorks/pull/17#issuecomment-1724123566, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADS3XWCVUVF5JV3MAPWTTMTX3CETFANCNFSM6AAAAAA4PR27D4. You are receiving this because your review was requested.Message ID: @.***>

gdicker1 commented 1 year ago

Yeah I'm not sure either...

For now I think I can fix this by using ncarenv/23.09 and using parallelio/2.6.2. For consistency (and maybe to share with ESMCI/ccs_configs) I'll do this in another ccs_configs PR and include that here.

gdicker1 commented 1 year ago

@dazlich I think I got PIO addressed. It was a smaller change than I thought see:

gdicker1 commented 1 year ago

I made the decision to merge all the "update/cesm2_3_alpha16d/${EXT}" branches and some of the other updates. The currently open PRs affecting this are:

I'm trying to run some tests this afternoon to make sure parallelio/2.6.2 is being picked up correctly and that the 15km files are being seen/used. I think the mpas-o and mpas-si PRs can be merged as is?

dazlich commented 1 year ago

I’m working on a 15km fully coupled test right now. I’ll keep you posted.

On Sep 19, 2023, at 3:14 PM, gdicker1 @.***> wrote:

Caution: EXTERNAL Sender

I made the decision to merge all the "update/cesm2_3_alpha16d/${EXT}" branches and some of the other updates. The currently open PRs affecting this are:

I'm trying to run some tests this afternoon to make sure parallelio/2.6.2 is being picked up correctly and that the 15km files are being seen/used. I think the mpas-o and mpas-si PRs can be merged as is?

— Reply to this email directly, view it on GitHubhttps://github.com/EarthWorksOrg/EarthWorks/pull/17#issuecomment-1726474260, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADS3XWFAEBV7EAPHLYZCQULX3IDJXANCNFSM6AAAAAA4PR27D4. You are receiving this because you were mentioned.Message ID: @.***>