I see. Thanks Jason.
Catherine
> Date: Mon, 20 Aug 2012 00:07:09 -0400
> From: jason.swails.gmail.com
> To: amber.ambermd.org
> Subject: Re: [AMBER] Error message in cpptraj when running at single node in supercomputer using mpirun
>
> Hello,
>
> On Sun, Aug 19, 2012 at 9:53 PM, Catein Catherine <askamber23.hotmail.com>wrote:
>
> >
> > Dear Sir/Madam,
> > I have a cpptraj input file as follows:
> > parm xxx.prmtoptrajin xxx.mdcrd.gzaverage xxx.pdb pdb
> > I then run $AMBERHOME/bin/cpptraj < cpptraj
> > It works perfectly. However, when I run the same command in mpirun and
> > queuing the job at supercomputer system, it did not work. Following are
> > some of the error message that I got.
> >
>
> By "the same command", do you mean:
>
> mpirun -np X $AMBERHOME/bin/cpptraj < cpptraj
>
> ??
>
> If so, this will not work for any of Amber's MPI programs. Every program
> in Amber that is MPI-parallelized has a separate MPI-enabled executable
> (such as cpptraj.MPI). I don't think cpptraj.MPI is ever built (not sure
> if it would even work if you could build it...)
>
> If you want parallel support for cpptraj, I think you need to do an OpenMP
> build (note, this is NOT OpenMPI). Before doing this, though, you should
> check the AmberTools manual for the specific actions that are
> parallelizable to see if it is worthwhile.
>
> Otherwise, just run cpptraj in serial.
>
> HTH,
> 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 Mon Aug 20 2012 - 00:00:03 PDT
