Hi Jason,
They are on the same line. I have used the same script with necessary
modifications to run MD in parallel and it has worked fine thusfar, I don't
know why it fails now. If there is a mismatching in MPIs above MD
calculations should also have given errors right?
Thank you very much
Sajeewa Dewage
On Fri, Feb 8, 2013 at 3:53 PM, Jason Swails <jason.swails.gmail.com> wrote:
> On Fri, Feb 8, 2013 at 3:37 PM, Sajeewa Pemasinghe <sajeewasp.gmail.com
> >wrote:
>
> > Hi all,
> >
> > I have the following script to run TREMD.
> >
> > #!/bin/bash -f
> > #PBS -q egacq
> > #PBS -l nodes=12:ppn=7
> >
> > ulimit -s unlimited
> >
> > export PATH=/wsu/arch/x86_64/mvapich-1.1/bin:/usr/local/amber10/bin:$PATH
> >
> > cd
> >
> >
> /wsu/home/ej/ej02/ej0271/new_project_5_20_2012/REMD_gb/redtemps/temp_parallel_with_pmemd/REMD_finally
> >
> > /wsu/arch/x86_64/mvapich-1.1/bin/mpirun -np 84
> > /usr/local/amber10/bin/sander.MPI -ng 42 -groupfile remd.groupfile
> >
>
> These must be on the same line. If they are on separate lines (and you do
> not end the first line with a \ to escape the newline), then it will first
> execute <nothing> with 84 cores and then execute sander.MPI with just 1
> (the default).
>
> If they _are_ both on the same line, then the only time I've ever seen this
> happen before is when the MPI from which mpirun comes from is different
> than the MPI used to compile Amber.
>
>
> > So I have allocated 84 cpus for 42 groups , so that 2 cpus get allocated
> > per group. But I get the following error message.
> >
> > Error: specified more groups ( 42 ) than the number of processors (1 ) !
>
>
> Based on the error message you got, the problem is not with sander.MPI, but
> either how you called sander.MPI or the mismatching MPIs I mentioned above.
>
> Good luck,
> Jason
>
> --
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Candidate
> 352-392-4032
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Feb 08 2013 - 13:30:04 PST
