config_pes.xml updates #215
Conversation
remove unnecessary blocks: no need to distinguish T62 and TL319 increase NTASKS_OCN for cases with WW3. '
- Introduce MOM6_WW3_CPL_METHOD xml variable controlling which wave coupling method gets used: most, legacy, or none - For B cases, set MIN_SALINITY to 1.0e-6 - For t232, set Laplacian to False - For t232, remove MEKE_ALPHA_RHINES, MEKE_ALPHA_DEFORM, MEKE_ALPHA_FRICT, MEKE_ALPHA_GRID, MEKE_CT
There was a problem hiding this comment.
Overall this looks great, I think combining the TL319 and T62 decompositions cuts down the file size and makes this much easier to read. I only had one comment about a specific change you made, but also have a general comment: in the fully coupled PE layouts, Jim uses the <comment> tags to provide expected throughput. For example, lines 280 and 281 in cime_config/config_pes.xml are
<pes pesize="M" compset="CAM.*%LT.*MOM6(?!.*%MARBL-BIO)">
<comment>10 ypd/ 5500 pe-hrs/simyr expected</comment>Could we do something similar? If you do the common derecho layouts (C/G, SWAV / WW3), I can update the MARBL compsets in a future PR. (Or we can create a separate issue for this and do it all later.)
cime_config/config_pes.xml
Outdated
| @@ -184,7 +184,7 @@ | |||
| <ntasks_rof>128</ntasks_rof> | |||
| <ntasks_cpl>128</ntasks_cpl> | |||
| <ntasks_ice>128</ntasks_ice> | |||
| <ntasks_ocn>384</ntasks_ocn> | |||
| <ntasks_ocn>1024</ntasks_ocn> | |||
There was a problem hiding this comment.
I'm a little surprised that we give MOM more tasks (1024) when running with WW3 than we do with SWAV (896, see line 121). Is this intentional?
There was a problem hiding this comment.
Agreed. I just set NTASKS_OCN to 896 instead of 1024.
Partially fixes: #211
(We still need to update PEs for fully coupled cases)