Ignore:
Timestamp:
2020-07-08T14:49:37+02:00 (3 months ago)
Author:
acc
Message:

Extra_Halo: revert previous bugfix in mppini.F90 and fix, instead, the ordering and treatment of processors in the north-fold communicator. Now passes all SETTE including an extra version of ORCA2_ICE_PISCES, REPRO_8_4 with land-suppression and ln_nnogather=F. run.stat differences from trunk@13218 in ORCA2_ICE_PISCES associated with calving (REPRO tests which have ln_calving=F match). Also tracer.stat differences after 42 timesteps (unexplained).

File:
1 edited

Legend:

Unmodified
Added
Removed
  • NEMO/branches/2020/dev_r12558_HPC-08_epico_Extra_Halo/src/OCE/LBC/lib_mpp.F90

    r13247 r13266  
    10981098      ! Look for how many procs on the northern boundary 
    10991099      ndim_rank_north = 0 
    1100       DO jjproc = 1, jpnij 
    1101          IF( njmppt(jjproc) == njmppmax )   ndim_rank_north = ndim_rank_north + 1 
     1100      DO jjproc = 1, jpni 
     1101         IF( nfproc(jjproc) /= -1 )   ndim_rank_north = ndim_rank_north + 1 
    11021102      END DO 
    11031103      ! 
     
    11091109      ! Note : the rank start at 0 in MPI 
    11101110      ii = 0 
    1111       DO ji = 1, jpnij 
    1112          IF ( njmppt(ji) == njmppmax   ) THEN 
     1111      DO ji = 1, jpni 
     1112         IF ( nfproc(ji) /= -1   ) THEN 
    11131113            ii=ii+1 
    1114             nrank_north(ii)=ji-1 
     1114            nrank_north(ii)=nfproc(ji) 
    11151115         END IF 
    11161116      END DO 
Note: See TracChangeset for help on using the changeset viewer.