T62_tn14 has bad pe-layout for DATM

[mvertens]

the following tests which use T62 for DATM provide only one processor for the datm

ERP_Ld3_P256.T62_tn14.NOINYOC.betzy_gnu
ERP_Ld3_P256.T62_tn14.NOINYOC.betzy_intel
ERS_Ld3.T62_tn14.N2NOINYOC.betzy_intel
ERS_Ld3.T62_tn14.NOINYOC.betzy_intel
ERS_Ld3.T62_tn14_wtn14nw.NOINY_WW3.betzy_gnu.blom-wavice
ERS_Ld3.T62_tn14_wtn14nw.NOINY_WW3.betzy_intel.blom-wavice
ERS_Ld3.T62_tn21.NOINY.betzy_intel
ERS_Ld5.T62_tn14.NOINY.betzy_intel.blom-woa_nuopc
SMS_D_Ld1.T62_tn14.NOIIA.betzy_intel
SMS_D_Ld1.T62_tn14.NOIIAJRAOC1850.betzy_intel
SMS_D_Ld1.T62_tn14.NOIIAJRAOC20TR.betzy_intel
SMS_D_Ld1.T62_tn14.NOIIAOC.betzy_intel
SMS_D_Ld1.T62_tn14.NOINYOC.betzy_gnu
SMS_D_Ld1.T62_tn14.NOINYOC.betzy_intel
SMS_D_Ld1.T62_tn14_wtn14nw.NOINYOC_WW3.betzy_gnu.blom-wavice
SMS_D_Ld1.T62_tn14_wtn14nw.NOINYOC_WW3.betzy_intel.blom-wavice

Comp  NTASKS  NTHRDS  ROOTPE PSTRIDE
CPL :     81/     1;      0      1
ATM :      1/     1;     80      1
LND :      1/     1;     80      1
ICE :     80/     1;      0      1
OCN :    431/     1;     81      1
ROF :      1/     1;     80      1
GLC :      1/     1;      0      1
WAV :     80/     1;      0      1
ESP :      1/     1;      0      1

I suggest changing this to the same pe-layout as for TL319:

Comp  NTASKS  NTHRDS  ROOTPE PSTRIDE
CPL :    128/     1;      0      1
ATM :    128/     1;      0      1
LND :    128/     1;      0      1
ICE :    128/     1;      0      1
OCN :    374/     1;    128      1
ROF :    128/     1;      0      1
GLC :    128/     1;      0      1
WAV :    128/     1;      0      1
ESP :      1/     1;      0      1

This caused problems when using the new ndep forcing data.
In addition - the following tests -

SMS_D_Ld1.T62_tn14.NOIIAJRAOC1850.betzy_intel
SMS_D_Ld1.T62_tn14.NOIIAJRAOC20TR.betzy_intel

should be changed to

SMS_D_Ld1.TL319_tn14.NOIIAJRAOC1850.betzy_intel
SMS_D_Ld1.TL319_tn14.NOIIAJRAOC20TR.betzy_intel

[mvertens]

@TomasTorsvik - I'll try to put a quick PR in place for these fixes.
SMS_D_Ld1.TL319_tn14.NOIIAJRAOC1850.betzy_intel
SMS_D_Ld1.TL319_tn14.NOIIAJRAOC20TR.betzy_intel
passed with no problem using the PR NorESMhub/CDEPS#28

[JorgSchwinger]

The old layout with 431 OCN / 80 ICE PEs is much more efficient than than the new layout. It used to work fine with DATM for the NOINYOC forcing (which is much less heavy than the JRA forcing) before the changes to the N-Deposition.

How many PEs are necessary for DATM with the new N-doposition? Could 4 be enough?

[mvertens]

@JorgSchwinger - the old layout was only being used with T62. I thought all of our current runs were with TL319 - and in that case 4 PEs for DATM seems too little. Can you please clarify when you want to use the 431 OCN / 80 ICE layout?

[JorgSchwinger]

The CORE-NYF (normal year) and the CORE-IAF (inter-annually varying) are T62. These are supported compsets in BLOM (and I think they still should be for a while). The compset names in BLOM are NOINYOC and NOIIANOC.

[TomasTorsvik]

@JorgSchwinger , @mvertens - Sorry, I was too quick to merge this. Should I roll back the latest merge?

[mvertens]

@TomasTorsvik @JorgSchwinger - my tests with the new nitrogen deposition at T62 was filing with just 1 PE for the atm but just for
SMS_D_Ld1.T62_tn14.NOIIAJRAOC1850.betzy_intel
SMS_D_Ld1.TL62_tn14.NOIIAJRAOC20TR.betzy_intel
Backing out the pe-layout change and just moving these tests to
SMS_D_Ld1.TL319_tn14.NOIIAJRAOC1850.betzy_intel
SMS_D_Ld1.TL319_tn14.NOIIAJRAOC20TR.betzy_intel
will fix the problem.

[TomasTorsvik]

@TomasTorsvik @JorgSchwinger - my tests with the new nitrogen deposition at T62 was filing with just 1 PE for the atm but just for SMS_D_Ld1.T62_tn14.NOIIAJRAOC1850.betzy_intel SMS_D_Ld1.TL62_tn14.NOIIAJRAOC20TR.betzy_intel Backing out the pe-layout change and just moving these tests to SMS_D_Ld1.TL319_tn14.NOIIAJRAOC1850.betzy_intel SMS_D_Ld1.TL319_tn14.NOIIAJRAOC20TR.betzy_intel will fix the problem.

Thanks for the update, @mvertens . I will make a new PR that backs out the pe-layout change.