Re: HELP setup on>>charmrun/mpirun & NAMD via torque

If you built namd with MPI  charmrun will be a shell script and not a  
binnary (find out by running:  file charmrun)

If you built it with mpi,  then you dont need charm run,  you can  
just do:

mpirun -np 4 namd2

as long as your mpirun is configured to use $PBS_NODEFILE or tm.

Brock Palen
Center for Advanced Computing
brockp@xxxxxxxxx
(734)936-1985


On Jul 25, 2007, at 10:17 AM, <Daniel.G.Roberts@xxxxxxxxxxxxxxxxxx>  
wrote:

> Hello All
> I am trying to perform namd runs using Torque/maui and Ihave several
> questions regaring the charmrun/mpirun portion especially.
>
> When I submit the script below>
>
> qsub script.txt
>
> I see this in this running inthe queue
> [nm67371@bwdl01-m001 Test06Proc]$ qstat -n
>
> bwdl01-m001.brw.pharma.aventis.com:
>                                                                     
> Req'd
> Req'd   Elap
> Job ID               Username Queue    Jobname    SessID NDS   TSK
> Memory Time  S Time
> -------------------- -------- -------- ---------- ------ ----- ---
> ------ ----- - -----
> 69.bwdl01-m001.brw.p nm67371  ghts     TEST-Webex   9559     2   
> --    --
> 336:0 R 00:05
>    bwdl01-n014/1+bwdl01-n014/0+bwdl01-n013/1+bwdl01-n013/0
> [nm67371@bwdl01-m001 Test06Proc]$
>
> But, when I do a ps on both nodes 14 and 13 I only can find my PID on
> node 14 and not node 13.
> So any ideas on how I can properly parrellize my jobs?
> What is the relationship between mpirun and charmrun?
> Thanks for any advice!
> Dan
> PS..
> The two different scripts below have the same type of behaviour
>
>
> EXAMPLE Script #0ne
> Script.txt
>
> #!/bin/bash
> #PBS -l nodes=2:ppn=2,walltime=336:00:00
> #PBS -N TEST-Webexample
> cd /global/bwsn001/home/users1/Dan/Muzet/Test06Proc
> #
> TMPDIR=/scratch/$PBS_JOBID
> #
> PROG=/global/bwsn002/software/ia32/namd/NAMD_2.6b1_Linux-i686/namd2
> ARGS="+strategy USE_GRID"
> IFILE="/global/bwsn001/home/users1/Dan/Muzet/Test06Proc/test.conf"
> OFILE="/global/bwsn001/home/users1/Dan/Muzet/Test06Proc/test.log"
> #
> # # NP should always be: nodes*ppn from #PBS -l above
> let NP=1*2
> #
> # # Full path to charmrun
> CHARMRUN=/global/bw_2/software/ia32/namd/NAMD_2.6b1_Linux-i686/ 
> charmrun
> #
> # # Preparing nodelist file for charmrun
>   export NODES=`cat $PBS_NODEFILE`
>   export NODELIST=nodelist
>   echo group main > $NODELIST
>        for node in $NODES ; do
>            echo host $node ++shell ssh >> $NODELIST
>        done
> #
> # # Execute the job using charmrun
> $CHARMRUN $PROG $ARGS $IFILE +p$NP ++nodelist $NODELIST > $OFILE
>
> OR
> EXAMPLE Script #TWO
> #PBS -N DAN2
> #PBS -l nodes=2:ppn=2,walltime=48:00:00
>
> #change to input directory
> cd /global/bwsn001/home/users1/Dan/Muzet/Test06Proc
>
> #run namd2  (this assumes namd2 is in my $PATH)
> #mpirun -np 4 namd2 <configfile>
>
> /global/bw_2/software/ia32/namd/NAMD_2.6b1_Linux-i686/charmrun ++p 6
> ++verbose ++nodelist $PBS_NODEFILE \
> /global/bwsn002/software/ia32/namd/NAMD_2.6b1_Linux-i686/namd2\
> /global/bwsn001/home/users1/nm67371/Muzet/Test06Proc/test.conf
>
>
>
> #end
>
>
> _______________________________________________
> torqueusers mailing list
> torqueusers@xxxxxxxxxxxxxxxx
> http://www.supercluster.org/mailman/listinfo/torqueusers