[Wien] Running Parallel Jobs

Peter Blaha pblaha at theochem.tuwien.ac.at
Wed Oct 18 09:59:50 CEST 2006


Is this a "reasonable" case (of size > 5000 ?
Does it run with the sequential version ?

If yes, maybe your scalapack installation has a problem.

Also your script does not seem right. In your email there are 2 lines (I 
don't know if they are broken accidentally):

 > /usr/local/mpich/bin/mpirun -np 4 -machinefile $PBS_NODEFILE
 > /usr/opt/WIEN2k/run_lapw -p

The first line does not have any "command" and is not right.

For testing you could use something like (in one line, not broken!):
usr/local/mpich/bin/mpirun -np 4 -machinefile $PBS_NODEFILE 
/usr/opt/WIEN2k/lapw1 lapw1_1.def

This runs only lapw1 in parallel and should be the same as
x lapw1 -p
when you have configured the parallel run (siteconfig) correctly. (The 
WIEN2k scripts will insert the mpirun commands automatically).

A full scf cycle is executed as usual

run_lapw -p

and the selection between k-point parallel and mpi-parallel comes only 
from the corresponding generation of the .machines file.

jadhikari at clarku.edu schrieb:
> Hello,
> Thank you very much for the previous helps.
> 
> We are trying to run a job in a parallel mode for the first time.
> But it never runs past the LAPW1 step.
> Here is the PBS script-
> ________________________________________
> #!/bin/sh
> #PBS -l nodes=2:ppn=2:myrinet
> #PBS -j oe
> cd /home/lsmith/rut
> rm -f .machines
> awk '{print "1:"$1}' $PBS_NODEFILE > .machines
> echo "granularity:1" >> .machines
> /usr/local/mpich/bin/mpirun -np 4 -machinefile $PBS_NODEFILE
> /usr/opt/WIEN2k/run_lapw -p
> ________________________________________
> Error message-
> 
> Missing: program name
>  LAPW0 END
>  LAPW1 END
>  LAPW1 END
>  LAPW1 END
>  LAPW1 END
>  LAPW1 END
> 
> Also we got another error as follows-
> 
> --Cholesky INFO =           895
>  'SECLR4' - POTRF (Scalapack/LAPACK) failed.
> 
> Is it the case.in1 error or the script error? We could not figure it out.
> The nodes run for 3-4 minutes and then stop running. TESTPARA1 shows this
> error.
> 
> Regards,
> Subin Adhikari
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

-- 

                                       P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: 
http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------


More information about the Wien mailing list