Re: [AMBER] invoking MPI_ABORT causes Open MPI to kill all MPI processes.

From: David A Case <david.case.rutgers.edu>
Date: Wed, 28 Sep 2016 09:09:46 -0400

On Wed, Sep 28, 2016, Pooja Kesari wrote:
>
> *I m trying to run a production run*
> mpirun -np 8 sander.MPI -O -i prod.in -o prod_complex-solv.out -p
> complex-pose8.prmtop -c protein_solv-equil.rst -r protein_solv-prod.rst -x
> prod.mdcrd -ref protein_solv-equil.rst
>
> *However my program is terminating with an error*
>
> MPI_ABORT was invoked on rank 0 in communicator MPI_COMM_WORLD
> with errorcode 1.

You need to look in the output files for indications about the error messages.
The above message just says that the job failed.

With sander.MPI, try to see if the problem occurs right at the end (after the
MD is finished, and while the program is trying to print the timings on
various nodes...we've seen some problems with some MPI's on that.

If possible, run with pmemd.MPI instead of sander.MPI.

...dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Sep 28 2016 - 06:30:05 PDT
Custom Search