[MITgcm-support] Problem with cal_fulldate

Jean-Michel Campin jmc at ocean.mit.edu
Mon Sep 22 11:49:09 EDT 2014


Hi Ying,

There have benn some changes in pkg/cal since checkpoint62v,
and specially pkg/cal setting (call to CAL_SET) has been moved,
in April 2012, from cal_readparms.F to cal_init_fixed.F (called at
the top of model/src/packages_init_fixed.F). And in order to make
this change safe, a STOP has been added to cal_fulldate.F
in case you try to use some pkg/cal variables that have not yet been set
(and I belive you are in this type of situation).

I would suggest to check in your RecoM pkg for early call to 
any pkg/cal subroutines, and in particular in recom_readparams.F 
(or equivalent). These calls will have to be move later in the call
sequence, for instance in recom_init_fixed.F (or equivalent).

Cheers,
Jean-Michel

On Mon, Sep 22, 2014 at 10:05:59AM +0200, Ying Ye wrote:
> Dear all,
> 
> I have updated my MITgcm from the version with the checkpoint 62v to
> 65c. As I started a test run with a package for our ecosystem model
> (RecoM, which worked well on the old version of MITgcm), I got the
> following error in STDERR file:
> 
> (PID.TID 0000.0001) *** ERROR *** CAL_FULLDATE: yymmdd=    10115 ,
> hhmmss=        0
> (PID.TID 0000.0001) *** ERROR *** CAL_FULLDATE: called too early
> (cal_setStatus= 0 )
> 
> And the output file as the job was brocken shows nothing more than
> this information:
> 
> Check F_SYSLEN: 1024
> /localtmp/tmp.0:2498.sx8-siox
> Tue Sep 16 10:35:18 DST 2014
>  STOP ABNORMAL END: S/R CAL_FULLDATE
> 
> 
>      ******  Program Information  ******
>   Real Time (sec)       :          0.218295
>   User Time (sec)       :          0.121446
>   Sys  Time (sec)       :          0.055556
>   Vector Time (sec)     :          0.002825
>   Inst. Count           :          37899144.
>   V. Inst. Count        :             74878.
>   V. Element Count      :          10551639.
>   FLOP Count            :             41196.
>   MOPS                  :        398.331806
>   MFLOPS                :          0.339212
>   VLEN                  :        140.917746
>   V. Op. Ratio (%)      :         21.811766
>  STOP ABNORMAL END: S/R CAL_FULLDATE
>   Memory Size (MB)      :        736.031250
>   MIPS                  :        312.065159
>   I-Cache (sec)         :          0.028743
>   O-Cache (sec)         :          0.009910
>   Bank (sec)            :          0.018754
> 
> 
> 
>   Start Time (date)  :  2014/09/16 10:35:18
>   End   Time (date)  :  2014/09/16 10:35:19
>      ******  Program Information  ******
>  STOP ABNORMAL END: S/R CAL_FULLDATE
> ./mitgcmuv(lang:f90): signal trap(SIGTERM: Software termination)
> ./mitgcmuv(lang:f90): signal trap(SIGTERM: Software termination)
> sx8-7: mpid: MPI process (universe = 0, rank = 1) terminated by exit(1)
> MPI process (universe = 0, rank = 2) terminated by exit(1)
> Tue Sep 16 10:35:24 DST 2014
> No match.
> No match.
> 
> Did anybody have a similar problem by updating? I saw a change with
> repect to cal_fulldate in tag-index on the checkpoint 63m. Schould I
> do any corresponding changes to the call of cal_fulldate in the
> ecosystem model package?
> 
> Many thanks.
> 
> Best regards,
> Ying
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support



More information about the MITgcm-support mailing list