[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] use Namd with condor



Hi Erik,

Very interseting, so my next question is : how can I write the "nodelist" file after the CPUs are booked ? What's the command ? (sorry for answering late, took a week off..)

Nicolas

----------------
On Fri, 7 Jul 2006 10:23:41 -0500
Erik Paulson <epaulson@xxxxxxxxxxx> wrote:

> On Fri, Jul 07, 2006 at 03:29:31PM +0200, Nicolas GUIOT wrote:
> > Hi
> > 
> > So here is my problem : I would like to use namd with condor on several CPUs. Namd is firstly designed for this, but it's not that trivial (IMHO) to use it with condor : 
> > Here is how namd usually works : 
> > You need to first create a file (nodelist.dat) which contains all the CPUs, then namd starts on the first one (a.k.a. "master node"), and the master dispatches the jobs through the other CPUs.
> > 
> > Here is how it needs to be in condor : 
> > - I submit my job to condor
> > - the matchmaker find X CPUs available for this job
> > - Then condor "books" these CPUs, so that nobody takes them, letting time for : 
> > - Condor creates the nodelist.dat file (with the available CPUs)
> > - Condor spawns the "master node".
> > - And finally the master will spawm the "slave" jobs.
> > 
> > In these conditions, it might not be possible to register the "slave nodes" as used by condor jobs (condor_status would probably list them as "Owner"),but I don't really care of this "detail"...
> > 
> > 
> > So, do you think it is possible ?
> > 
> 
> This is exactly what the parallel universe in Condor 6.7 does. `
> 
> -Erik


----------------------------------------------------
CNRS - UPR 9080 : Laboratoire de Biochimie Theorique
Institut de Biologie Physico-Chimique
13 rue Pierre et Marie Curie
75005 PARIS - FRANCE

Tel : +33 158 41 51 70
Fax : +33 158 41 50 26
----------------------------------------------------