Update and add new support for configurations with dynamic Greenland ice sheet component #6226
Conversation
Add configuration and mapping file support for MALI, 20 km res Greenland ice sheet, tri-grid configuration.
Add explicit path to Albany lib to support perlmutter builds.
Add and correct support for GIS 20km and 1-to-10km configs with r05, ne30pg2, and EC30to60E2r2 grids.
Add CIME support for new GIS 1-to-10 (r02). Confirmed that BG compset calling for related grids builds successfully.
Updated root paths for kokkos, albany, and trilinos in pm-cpu machine file. Removed no longer needed info from gnu_pum-cpu cmake file. Added support for e3sm + mali (Greenland) configurations for IG and BG cases, including: updated default output freq for mali hist files; add default PE layout for Greenland 20km IG case; add support for Greenland 20km IG test case to various testing infrastructure files.
Explicitly specify which glcmask to use for Greenland 20km test case, which allows elm namelist changes for test to be removed.
…e sheet. Archive settings switched back to mpas component defaults; change days run for ERP test.
Fix default layout to work on pm-cpu, alvarez, and chrysalis.
stephenprice
added
enhancement
mpas-albany-landice
Land
compset
Chrysalis
pm-cpu
Increase default value for h2osno_max in many-layer snowpack model. Add missing code from MALI driver so that global and regional analysis members are called in simulations with dynamic ice sheets.
…6226) Update and add new support for configurations with dynamic Greenland ice sheet component This PR updates and/or adds new support that allows for simulations, through existing IG- and BG- compsets, with a dynamic MALI component for the Greenland ice sheet. These configurations are still under development and are not yet scientifically supported. Changes include: * added support for Greenland 20km res., trigrid configuration * updated Greenland 1-to-10km res. grid and input files * add support for configs. using new 1-to-10km grid and init. cond. * updated defaults for MALI history (output) files * updated default Albany solver parameter settings * added SMS, ERS, and ERP tests for 20km res. IG configuration * updated machine file support for Albany & Trilinos on pm-cpu & Chrysalis * add default layouts for 20km & 1-to-10km configs on pm-cpu & Chrysalis * allow MALI global and regional analysis AMs to operate during e3sm runs [non-BFB] only for tests with active MALI
|
Passes:
verified all new files are in the correct format and available on the inputdata repo merged to next |
|
Re: changes to config_pesall.xml. looks like you explicitly added SGLC to all the fully coupled cases except for the one where you actually run MALI. I think thats a good way to make that distinction. |
Thanks. In fact, we could not figure out how to get the GLC-specific configs. (which are similar to some other configs in many ways) to be recognized without making this change. This was Jon W's suggested solution. |
|
requested running the new e3sm_landice_developer test suite on chrysalis and pm-cpu using the gnu compiler, SES-3111 |
|
This PR causes a build error with intel pm-cpu. I think we should revert and fix. Also, maybe ping the POC before making changes to machine files. |
| @@ -263,8 +263,9 @@ | |||
| <env name="PERL5LIB">/global/cfs/cdirs/e3sm/perl/lib/perl5-only-switch</env> | |||
| <env name="FI_CXI_RX_MATCH_MODE">software</env> | |||
| <env name="MPICH_COLL_SYNC">MPI_Bcast</env> | |||
| <env name="Albany_ROOT">$SHELL{if [ -z "$Albany_ROOT" ]; then echo /global/common/software/e3sm/mali_tpls/albany-e3sm-serial-release-gcc; else echo "$Albany_ROOT"; fi}</env> | |||
| <env name="Albany_ROOT">$SHELL{if [ -z "$Albany_ROOT" ]; then echo /global/common/software/e3sm/mali_tpls/albany-e3sm-serial-release-gcc-cmake-fix; else echo "$Albany_ROOT"; fi}</env> | |||
There was a problem hiding this comment.
Should we put these changes in the gnu environment_variables compiler like done in Chrysalis?
<environment_variables compiler="gnu">
<env name="Albany_ROOT">$SHELL{if [ -z "$Albany_ROOT" ]; then echo /global/common/software/e3sm/mali_tpls/albany-e3sm-serial-release-gcc-cmake-fix; else echo "$Albany_ROOT"; fi}</env>
<env name="Trilinos_ROOT">$SHELL{if [ -z "$Trilinos_ROOT" ]; then echo /global/common/software/e3sm/mali_tpls/trilinos-e3sm-serial-release-gcc; else echo "$Trilinos_ROOT"; fi}</env>
<env name="Kokkos_ROOT">$SHELL{if [ -z "$Kokkos_ROOT" ]; then echo /global/common/software/e3sm/mali_tpls/trilinos-e3sm-serial-release-gcc; else echo "$Kokkos_ROOT"; fi}</env>
</environment_variables>
Sorry for the inconvenience. I guess we should have put the pm-cpu changes to the machine file under the gnu environment, like we did with Chrysalis. |
|
@mperego -- can you make that fix and either add it to this PR or make a separate PR? We don't want to mess up pm-cpu testing any longer than necessary |
|
@jonbob yes, I'll add the fix to this PR. |
|
@ndkeen -- we're working to test both Mauro's recent fix for this and will also re-test one of your failing intel tests as well. Will get back to you asap. |
|
@ndkeen -- with Mauro's fix above, |
|
@stephenprice next still fails for me with |
|
@ndkeen -- I'm just re-merging to next right now. I'll let you know once it's there |
#6226) Fix environment variables in config machine file
|
Re-merged to next. @ndkeen -- we'd very much appreciate any testing you can do. With apologies |
|
I verified these two tests now build/run with next. |
|
merged to master and new tests blessed |
Move extrasnowlayers test to e3sm_landice_developer suite Since PR #6226 redefined the IGELM compset, the SMS.ne30_oECv3_gis.IGELM_MLI.elm-extrasnowlayers test has been failing. This moves that test from the e3sm_ocnice_extra_coverage suite to the e3sm_landice_developer suite, which is only run on platform/compiler combinations that support active MALI. [BFB] for all current E3SM tests
Move extrasnowlayers test to e3sm_landice_developer suite Since PR #6226 redefined the IGELM compset, the SMS.ne30_oECv3_gis.IGELM_MLI.elm-extrasnowlayers test has been failing. This moves that test from the e3sm_ocnice_extra_coverage suite to the e3sm_landice_developer suite, which is only run on platform/compiler combinations that support active MALI. [BFB] for all current E3SM tests
This PR updates and/or adds new support that allows for simulations, through
existing IG- and BG- compsets, with a dynamic MALI component for the
Greenland ice sheet. These configurations are still under development and are
not yet scientifically supported.
Changes include: