[CIG-SEISMO] running on cluster

[Dimitri Komatitsch]

Dear Andri,

It seems that your input file (called "Par_file" in SPECFEM) contains 
NPROC = 1, i.e. it expects to run on a single processor, but in your 
batch script you put bsub -n 4 to request four processors. Thus if you 
change NPROC to 4 in your Par_file you should be fine.

The other option is that you configured and compiled the code without 
support for MPI; if so, just type "configure --with-mpi" instead of just 
"configure", then "make clean ; make all" and the problem should be fixed.

Best regards,
Dimitri.

On 24/08/2016 12:47, A. HENDRIYANA wrote:
> Dear all,
>
> I am PhD student and got interested to use specfem3D cartesian. Since
> two days ago I have striven to run specfem on cluster. I follow some
> script from specfem3D/utils/cluster and also some demos particularly
> from specfem3d/EXAMPLES/meshfem3D_examples/simple_model. The two results
> after typing following command:
>
> bsub -a mvapich -o %J.log -n 4 mpirun.lsf ./bin/xmeshfem3D
>
> 1. status retriev from bjobs:
> JOBID   USER    STAT  QUEUE      FROM_HOST   EXEC_HOST   JOB_NAME
> SUBMIT_TIME
> 419351  andri   RUN   par32      glic2       4*qdr_16_gr *meshfem3D Aug
> 24 11:59
>
> and
>
> 2. from output file (%J.log)
>
> Error: must have NPROC_XI = NPROC_ETA = 1 for a serial run
> Error: must have NPROC_XI = NPROC_ETA = 1 for a serial run
> Error: must have NPROC_XI = NPROC_ETA = 1 for a serial run
> Error: must have NPROC_XI = NPROC_ETA = 1 for a serial run
>
> I could not figure out what actually happens. Any clues would be
> appreciated.
>
> Thanks in advance.
>
> Regards,
> Andri Hendriyana
> Bandung Institute of Technology, Indonesia
>
>
>
>
> _______________________________________________
> CIG-SEISMO mailing list
> CIG-SEISMO at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-seismo
>

-- 
Dimitri Komatitsch
CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics,
UPR 7051, Marseille, France    http://komatitsch.free.fr