From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Fri Oct 02 2009 - 14:57:43 CDT
On Fri, 2009-10-02 at 11:50 -0700, BIN ZHANG wrote:
>
>
> So you mean if I include the following loop in the PBS script:
> for (( i = 0; i<=100; i++ ))
> do
> namd2 namd_input_$i
> done
> , then the jobs will simply run one by one?
>
yes. of course. why not?
pbs has absolutely no idea about what you are doing.
it just reserves a number of nodes or rather slots
and executes the script you give it on the head node/slot
and waits until the script is done to release the
reserved resources. there is no magic there.
if you reserve multiple nodes to be used with MPI,
you can even submit different parallel jobs on subsets
of the reserved block of nodes/slots by backgrounding
them and then put just a "wait" into the script to
wait for all of them to finish so that the script
only terminates if all of them are done.
cheers,
axel.
>
> That will be great!
>
>
> Thanks,
> Bin
>
>
>
>
> On Oct 2, 2009, at 11:30 AM, Axel Kohlmeyer wrote:
>
> > On Fri, 2009-10-02 at 11:23 -0700, BIN ZHANG wrote:
> > > Dear Axel:
> > >
> > > Thanks for your reply.
> > >
> > > You will be absolutely right in an ideal world. However, I'm
> > > planning
> > > on running a CG simulation approaching 1 micro S timescale. Also,
> > > I
> > > will be running my jobs on a queuing system, so too frequent
> > > restarting will increase the waiting time on the queue.
> >
> > ...and in a real world you would just put multiple namd
> > executions into a single script file that you submit
> > to the batch system and then you are fine again.
> >
> > other than that, the NAMD source code is available
> > for download, so feel free to implement this hack yourself.
> >
> > cheers,
> > axel.
> >
> >
> > > Bin
> > >
> > > On Oct 2, 2009, at 11:09 AM, Axel Kohlmeyer wrote:
> > >
> > > > On Thu, 2009-10-01 at 23:15 -0700, BIN ZHANG wrote:
> > > > > Dear All:
> > > >
> > > > dear bin,
> > > >
> > > > > Is it possible to change the atom type of a particle on the
> > > > > fly
> > > > > without stopping the simulation? Say, after one ns simulation,
> > > > > I want
> > > > > to change the parameter for a particular particle (alternating
> > > > > the
> > > > > parameter for LJ balls).
> > > > >
> > > > > Any suggestions is appreciated.
> > > >
> > > > if you run for one nanosecond, the time for startup
> > > > should be negligible compared to runtime. so why bother?
> > > >
> > > > just prepare the necessary topology/parameter files
> > > > ahead of time and you should be fine and run them
> > > > one after the other.
> > > >
> > > > cheers,
> > > > axel.
> > > > >
> > > > > Thanks,
> > > > > Bin
> > > > >
> > > > >
> > > > --
> > > > Dr. Axel Kohlmeyer akohlmey_at_gmail.com
> > > > Institute for Computational Molecular Science
> > > > College of Science and Technology
> > > > Temple University, Philadelphia PA, USA.
> > > >
> > >
> > --
> > Dr. Axel Kohlmeyer akohlmey_at_gmail.com
> > Institute for Computational Molecular Science
> > College of Science and Technology
> > Temple University, Philadelphia PA, USA.
> >
> >
>
-- Dr. Axel Kohlmeyer akohlmey_at_gmail.com Institute for Computational Molecular Science College of Science and Technology Temple University, Philadelphia PA, USA.
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:53:20 CST