Re: [AMBER] sander.MPI running on 8 nodes or 1??

From: Jason Swails <jason.swails.gmail.com>
Date: Tue, 8 Mar 2011 11:27:13 -0800

What command did you use to spawn all 8 threads? I've seen this happen
before when, for instance, I compile sander.MPI with OpenMPI compiler
wrappers, yet use an mpich2 mpirun/mpiexec script to launch sander.MPI.
This is bad because then you have 8 separate 1-thread MPI jobs trying to
read from and write to the same files.

Try recompiling Amber, taking care that you're using the right mpicc/mpif90
(make sure PATH is set up correctly to use the proper compilers, or modify
the compilers with absolute paths in config.h).

Hope this helps,
Jason

On Tue, Mar 8, 2011 at 1:11 AM, colvin <colvin4367.gmail.com> wrote:

> Hi all,
>
> I tried to run my simulations on a machine with 2 quad core processors (8
> in
> total). When i top, it shows 8 running sander.MPI.
>
> But when i open the .out file from md run, it stated that:
>
> "Running AMBER/MPI version on 1 nodes"
>
> and this actually explained why my simulation taken such a long time, even
> for 200 ps (about 3 days).
>
> Pls assist and comment. Thanks.
>
> Thanks and regards,
> Colvin
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>



-- 
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
Received on Tue Mar 08 2011 - 11:30:04 PST
Custom Search