AMBER Archive (2009)

Subject: RE: [AMBER] reg . parallel run

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Thu Jan 22 2009 - 09:52:55 CST

Hi Balaji

> I have installed the parallel version
> of amber in the cluster using LAM , I would like to know
> what are the other thing have to be done
> to make a parallel run
>
> like the needed files in the run directory .,
>
> and i would also like to know the restart option
> in case if there is poweroff
>
> if it is already available in any tutorial
> let me know please !

There is no tutorial for this since it is largely dependent on the way your
cluster is setup etc, what the interconnect is, if there is a queuing system
etc. In the simple case you would create a machine file that defines the
nodes you want to run on and then have lamboot access this and then when you
run mpirun it will use those nodes. In the very simple case of just running
in parallel on 1 node (say 4 cpus per node) you would just do (assuming the
correct lam installation is in your path):

lamboot
mpirun -np 4 $AMBERHOME/exe/sander.MPI -O -i mdin -o mdout -p prmtop -c
inpcrd -x mdcrd
lamhalt

With respect to running in parallel all the options - how to restart a job
etc should be identical to if you were running the job in serial. Note for
best performance in parallel you should compile pmemd and use that -
assuming you are doing MD simulations that are within it's feature set.

Good luck,
Ross

/\
\/
|\oss Walker

| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
| http://www.rosswalker.co.uk | PGP Key available on request |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber