[MITgcm-devel] ecco_cost_init_barfiles.F

[Matthew Mazloff]

Hi Patrick,

My model was occasionally stalling (for occasionally ~20 minutes) when  
initializing the 2D files, e.g. smrareabar, sst, and psbar, their ad*  
counterparts, etc.

I think I fixed this (no stalls as of recently) by putting the  
following in ecco_cost_init_barfiles



#ifdef ALLOW_SSH_COST_CONTRIBUTION
c--   Save psbar on file.
       ilps=ilnblnk( psbarfile )
       write(fnamepsbar,'(2a,i10.10)')
      &     psbarfile(1:ilps), '.', optimcycle
       write(adfnamepsbar,'(3a,i10.10)')
      &     yadprefix, psbarfile(1:ilps), '.', optimcycle

CMM 
(---------------------------------------------------------------------------
       MRec = ndaysrec/NR
       do irec = 1, MRec
#ifndef ALLOW_TANGENTLINEAR_RUN
         call active_write_xyz( adfnamepsbar, tmpfld3d, irec,
      &                        optimcycle, mythid, xx_psbar_mean_dummy )
#endif
       enddo
       do irec =  (MRec*NR),ndaysrec
CMM      do irec = 1, ndaysrec
CMM 
)---------------------------------------------------------------------------

c        call active_write_xy( fnamepsbar, tmpfld2d, irec,
c     &                        optimcycle, mythid, xx_psbar_mean_dummy )
#ifndef ALLOW_TANGENTLINEAR_RUN
         call active_write_xy( adfnamepsbar, tmpfld2d, irec,
      &                         optimcycle,mythid, xx_psbar_mean_dummy )
#endif
       enddo
#endif /* ALLOW_SSH_COST_CONTRIBUTION */


I also did this for the  ALLOW_DAILYSST_COST_CONTRIBUTION and  
ALLOW_SEAICE_COST_SMR_AREA parts

I would recommend checking this in.   Though will there be an issue  
when ndaysrec < NR  ... maybe something else is needed -- maybe an if  
statement -- if this issue makes it too much of a hassle I understand.

Anyway, food for thought,
Matt