implement flexible moist grid comp DT #527

weiyuan-jiang · 2022-01-28T16:18:47Z

Initial check in to address issue #526 .

weiyuan-jiang · 2022-01-28T16:22:34Z

I am not sure what is this alarm for. Maybe we should consider how to combine these two alarms. @atrayano

GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSmoist_GridComp/GEOS_MoistGridComp.F90

Lines 5422 to 5436 in fb1d28f

    
           call MAPL_Get( STATE, IM=IM, JM=JM, LM=LM,   & 
        
                RUNALARM = ALARM,             & 
        
                CF       = CF,                & 
        
                LONS     = LONS,              & 
        
                LATS     = LATS,              & 
        
                INTERNAL_ESMF_STATE=INTERNAL, & 
        
                RC=STATUS ) 
        
           VERIFY_(STATUS) 
        
           ! If its time, calculate convective tendencies 
        
           ! -------------------------------------------- 
        
           if ( ESMF_AlarmIsRinging( ALARM, RC=status) ) then

atrayano · 2022-01-28T16:33:15Z

@weiyuan-jiang My intention is noty to modify MAPL's original alarm (and we cannot use the MAPL's alarm since it ring at the last time step of run_dt). Convention in GEOS is that every componet in the system is called every time step during the main time loop. The new alarm will be used to decide whether to run or to skip calculations. It is almost identical to the alarm in MAPL except for the offset (RingTime). I cannot see how we could combine the two alarms

weiyuan-jiang · 2022-02-01T17:37:28Z

@atrayano It is strange that this PR crashes when I set MOIST_DT as double of HEARTBEAT. I am debugging...

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
GEOSgcm.x 000000000338847A Unknown Unknown Unknown
libpthread-2.22.s 00002AB5BDC29CE0 Unknown Unknown Unknown
GEOSgcm.x 0000000002A1D837 geos_utilsmod_mp_ 555 GEOS_Utilities.F90
GEOSgcm.x 0000000002A1D53E geos_utilsmod_mp_ 612 GEOS_Utilities.F90
GEOSgcm.x 0000000001E9E24F geos_lakegridcomp 1753 GEOS_LakeGridComp.F90
GEOSgcm.x 0000000001E98B03 geos_lakegridcomp 1441 GEOS_LakeGridComp.F90
libesmf.so 00002AB5AEE82514 _ZN5ESMCI6FTable1 Unknown Unknown
libesmf.so 00002AB5AEE8605F ESMCI_FTableCallE Unknown Unknown
libesmf.so 00002AB5AF421A8A _ZN5ESMCI3VMK5ent Unknown Unknown
libesmf.so 00002AB5AF40F285 ZN5ESMCI2VM5ente Unknown Unknown
libesmf.so 00002AB5AEE8396A c_esmc_ftablecall Unknown Unknown
libesmf.so 00002AB5AF6B686D esmf_compmod_mp_e Unknown Unknown
libesmf.so 00002AB5AF9C5C51 esmf_gridcompmod Unknown Unknown
libMAPL.generic.s 00002AB5A9B5526B mapl_genericmod_m 1794 MAPL_Generic.F90
libesmf.so 00002AB5AEE82514 _ZN5ESMCI6FTable1 Unknown Unknown
libesmf.so 00002AB5AEE8605F ESMCI_FTableCallE Unknown Unknown
libesmf.so 00002AB5AF421A8A _ZN5ESMCI3VMK5ent Unknown Unknown
libesmf.so 00002AB5AF40F285 ZN5ESMCI2VM5ente Unknown Unknown
libesmf.so 00002AB5AEE8396A c_esmc_ftablecall Unknown Unknown
libesmf.so 00002AB5AF6B686D esmf_compmod_mp_e Unknown Unknown
libesmf.so 00002AB5AF9C5C51 esmf_gridcompmod Unknown Unknown
GEOSgcm.x 0000000001E2260E geos_surfacegridc 8617 GEOS_SurfaceGridComp.F90
GEOSgcm.x 0000000001E06903 geos_surfacegridc 6754 GEOS_SurfaceGridComp.F90
libesmf.so 00002AB5AEE82514 _ZN5ESMCI6FTable1 Unknown Unknown
libesmf.so 00002AB5AEE8605F ESMCI_FTableCallE Unknown Unknown
libesmf.so 00002AB5AF421A8A _ZN5ESMCI3VMK5ent Unknown Unknown
libesmf.so 00002AB5AF40F285 ZN5ESMCI2VM5ente Unknown Unknown
libesmf.so 00002AB5AEE8396A c_esmc_ftablecall Unknown Unknown
libesmf.so 00002AB5AF6B686D esmf_compmod_mp_e Unknown Unknown
libesmf.so 00002AB5AF9C5C51 esmf_gridcompmod Unknown Unknown
libMAPL.generic.s 00002AB5A9B5526B mapl_genericmod_m 1794 MAPL_Generic.F90
libesmf.so 00002AB5AEE82514 _ZN5ESMCI6FTable1 Unknown Unknown
libesmf.so 00002AB5AEE8605F ESMCI_FTableCallE Unknown Unknown
libesmf.so 00002AB5AF421A8A _ZN5ESMCI3VMK5ent Unknown Unknown
libesmf.so 00002AB5AF40F285 ZN5ESMCI2VM5ente Unknown Unknown
libesmf.so 00002AB5AEE8396A c_esmc_ftablecall Unknown Unknown
libesmf.so 00002AB5AF6B686D esmf_compmod_mp_e Unknown Unknown
libesmf.so 00002AB5AF9C5C51 esmf_gridcompmod Unknown Unknown
GEOSgcm.x 000000000068BE7C geos_physicsgridc 2620 GEOS_PhysicsGridComp.F90

atrayano · 2022-02-01T19:28:01Z

@weiyuan-jiang When you create the new alarm (iMoistAlarm n Initialize), you need to explicitly set it to ringing. Although your ringTime is the same as currTime, the state of alarm will be changed only during the call ESMF_ClockAdvance (in Cap). So, just add this call as soon as the alarm is created
call ESMF_AlarmSet(MoistAlarm, ringing=.true., RC)
Otherwise the very first step the alarm is not ringing, and then as a result, moist skips the run method, all exports are not touched, they stay 0, and the code dies in the QSAT calculation in one of the children of Physics (lake in your case)

weiyuan-jiang · 2022-02-01T21:10:36Z

@weiyuan-jiang When you create the new alarm (iMoistAlarm n Initialize), you need to explicitly set it to ringing. Although your ringTime is the same as currTime, the state of alarm will be changed only during the call ESMF_ClockAdvance (in Cap). So, just add this call as soon as the alarm is created
call ESMF_AlarmSet(MoistAlarm, ringing=.true., RC)
Otherwise the very first step the alarm is not ringing, and then as a result, moist skips the run method, all exports are not touched, they stay 0, and the code dies in the QSAT calculation in one of the children of Physics (lake in your case)

call ESMF_AlarmSet(MoistAlarm, ringing=.true., RC)

Even without above call, the moist run at the first time. After I added, there is no difference and it still crashed at the same point.

weiyuan-jiang · 2022-02-02T16:34:54Z

@atrayano I know why it crashes. When I set MOIST_DT = HEARTBEAT, it runs and the old alarm rings. If I set MOIST_DT = 2*HEARTBEAT, at the first time, the old alarm doesn't ring. So there is interference of the MoistAlarm and Alarm .

GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSmoist_GridComp/GEOS_MoistGridComp.F90

Lines 5436 to 5444 in fb1d28f

    
           if ( ESMF_AlarmIsRinging( ALARM, RC=status) ) then 
        
              VERIFY_(STATUS) 
        
              call ESMF_AlarmRingerOff(ALARM, RC=STATUS) 
        
              VERIFY_(STATUS) 
        
              call MAPL_TimerOn(STATE,"DRIVER") 
        
              call MOIST_DRIVER(IM,JM,LM, RC=STATUS) 
        
              VERIFY_(STATUS) 
        
              call MAPL_TimerOff(STATE,"DRIVER") 
        
           endif

weiyuan-jiang · 2022-02-03T13:54:52Z

After talking to @atrayano , this PR now runs as expected. ( runs at time 0 and every Moist_DT). But I am still confused why the new MoistAlarm affects the ringing time of the old ALARM in 5414 ( commented out now) @bena-nasa @tclune

GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSmoist_GridComp/GEOS_MoistGridComp.F90

Line 5422 in e933dfb

call ESMF_ClockGetAlarm(CLOCK, 'MoistAlarm', MoistAlarm, rc=status)

weiyuan-jiang · 2022-02-03T17:22:39Z

After @bena-nasa walked me through the codes, I now understand why my new MoistAlarm affects the old generic ALARM. It is the parameter name "MOIST_DT" that gets into the generic call and changes the generic alarm's interval. ( it checks "DT", and eventually checks "MOIST_DT") . That really eases my mind and I am confident that this MoistAlarm will not change other components' alarms. @atrayano @tclune

weiyuan-jiang · 2022-02-09T19:19:18Z

@wmputman , This is a branch you can try with longer time step for Moisture grid comp. Just add parameter "MOIST_DT:" to AGCM.rc. The run alarm will ring at time 0 and every moist_dt. If you are happy with this approach, we can move onto other grid comps. @atrayano @tclune

sdrabenh · 2022-03-16T14:48:45Z

@weiyuan-jiang is this PR finalized? I'm unsure since this still has the DNA label.

weiyuan-jiang · 2022-03-17T19:25:17Z

We can use this PR [https://github.com/GEOS-ESM/MAPL/pull/1369] in MAPL to achieve the same function @atrayano @tclune

sdrabenh · 2022-05-03T13:48:15Z

@weiyuan-jiang there haven't been any updates to this PR in quite some time. Is this ready for merging? If not, could you please convert this to a draft PR or close it if no longer being pursued.

weiyuan-jiang · 2022-05-03T13:57:12Z

Do you have opinion on the PR in MAPL ? ( GEOS-ESM/MAPL#1369 ) ? We need to choose one. @tclune @bena-nasa @atrayano

tclune · 2022-05-03T13:59:11Z

I have not looked closely at this ticket. If there are choices, I would largely defer to @atrayano as to the preferred path.

sdrabenh · 2022-09-08T17:26:59Z

@weiyuan-jiang can we close this now? There has been no movement for quite some time and things looks stale. I would like to clear out stagnant PRs that are just cluttering the queue.

weiyuan-jiang · 2022-09-08T17:43:35Z

OK, let me close it for now. I will reopen it when we need it.

implement flexible moist grid comp DT

fb1d28f

weiyuan-jiang added 0 diff The changes in this pull request have verified to be zero-diff with the target branch. Contingent - DNA These changes are contingent on other PRs (DNA=do not approve) labels Jan 28, 2022

weiyuan-jiang requested a review from a team as a code owner January 28, 2022 16:18

fix typo

1a2591c

replace Alarm with new MoistAlarm

e933dfb

sdrabenh added 2 commits February 10, 2022 17:03

Merge branch 'develop' into feature/wjiang/flex_moist_dt

d84234c

Merge branch 'develop' into feature/wjiang/flex_moist_dt

6b93883

weiyuan-jiang closed this Sep 8, 2022

implement flexible moist grid comp DT #527

implement flexible moist grid comp DT #527

Uh oh!

Conversation

weiyuan-jiang commented Jan 28, 2022

Uh oh!

weiyuan-jiang commented Jan 28, 2022

Uh oh!

atrayano commented Jan 28, 2022

Uh oh!

weiyuan-jiang commented Feb 1, 2022

Uh oh!

atrayano commented Feb 1, 2022

Uh oh!

weiyuan-jiang commented Feb 1, 2022

Uh oh!

weiyuan-jiang commented Feb 2, 2022

Uh oh!

weiyuan-jiang commented Feb 3, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

weiyuan-jiang commented Feb 3, 2022

Uh oh!

weiyuan-jiang commented Feb 9, 2022

Uh oh!

sdrabenh commented Mar 16, 2022

Uh oh!

weiyuan-jiang commented Mar 17, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sdrabenh commented May 3, 2022

Uh oh!

weiyuan-jiang commented May 3, 2022

Uh oh!

tclune commented May 3, 2022

Uh oh!

sdrabenh commented Sep 8, 2022

Uh oh!

weiyuan-jiang commented Sep 8, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

weiyuan-jiang commented Feb 3, 2022 •

edited

Loading

weiyuan-jiang commented Mar 17, 2022 •

edited

Loading