[cam-users] SPMD/MPI issues

[David Noone]

Cathryn,

You need to tell mpirun to use more than one processor. You should start
cam as something like:

   mpirun -np 2 cam < namelist

If you examine the batch script, you will find the line like this.
The -np number can not be odd, or certain multiples of 2.

Checking that this has been set correctly, might require working out how
your PBS works. For instance you might bae somethlike like the folliwing
in you script:

#PBS -l nodes=2:ppn=2

  cd ${PBS_O_WORKDIR};
  set nproc = `cat $PBS_NODEFILE | wc -l`

  mpirun -np $nproc $blddir/cam < namelist


Good luck,

David



On Fri, 24 Sep 2004, Cathryn Meyer wrote:

> Hello,
>
> I am trying to run CAM3.0 on a linux cluster using two nodes, and the
> model configured and compiled fine, however now every time I try to run
> it I get the error message
>
> Visible optical depth (tauback) =     0.000000000000000
> (shr_orb_print) Orbital parameters calculated for year: AD      1950
>   ------------------------------------------
>  ISCCP calcs and history IO will NOT be done
>  Problem factors: 2**            6  * 3**            0  * 5**            0
>  SPMDINIT_DYN: nprocs(            1 ) must be a multiple of 2
>  ENDRUN: called without a message string
> [0] MPI Abort by user Aborting program !
> [0] Aborting program!
> p0_25511:  p4_error: : 1
>
>
> No matter what I change in my PBS submission script (no matter how I
> change the number of nodes or processors).  Any ideas?
>
> Thanks,
> Cathy
> _______________________________________________
> cam-users mailing list
> cam-users at ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/cam-users
>

-- 
------------------------------------------------------------------------
PAOS & CIRES, University of Colorado, Boulder, UCB 311, CO 80309-0311
Ph: (303) 735-6073  Fax: (303) 492-3524
http://paos.colorado.edu/~dcn