[mpich-discuss] dear Dr.Balaji
Pavan Balaji
balaji at mcs.anl.gov
Fri Jun 17 15:49:48 CDT 2011
Dear Deqiang,
Please send all queries to mpich-discuss at mcs instead of to me directly.
I've cc'ed the list.
I see that you are using the following command-line inside your SGE script:
$MPI_DIR/bin/mpiexec -np $NSLOTS -machinefile $TMP/machines
$SSLE_DIR/ssle < input.in
Is $TMP/machines the machinefile you created or the one that SGE
created? If the SGE created file only has one node, it means that you
only allocated one node, not 5. Or am I missing something here?
-- Pavan
On 06/17/2011 03:33 PM, Deqiang Mao wrote:
> Dear Pavan
>
> One quick question about SGE and MPICH2.
> In my machine file, I set like this
> *compute-0-0:2*
> *compute-0-1:2*
> *compute-0-2:2*
> *compute-0-3:2*
> *compute-0-4:2*
>
> For each machine, I only want to run 2 processes. When I test the set
> using mpiexec, it works well.
> *compute-0-4.local*
> *compute-0-4.local*
> *compute-0-1.local*
> *compute-0-1.local*
> *compute-0-0.local*
> *compute-0-0.local*
> *compute-0-2.local*
> *compute-0-2.local*
> *compute-0-3.local*
> *compute-0-3.local*
>
> Like the results above. But when I use sge to initiate the my own
> program using script
> *qsub -pe mpich 10 ~/app/ssle/ssle.sh*
>
> The output file shows
> *-catch_rsh
> /opt/gridengine/default/spool/compute-0-0/active_jobs/10.1/pe_hostfile*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
> *compute-0-0*
>
> Because I have 16 processes on each machine, it only call 10 processes
> on the first machine. How do I set the script to let it run 2 process on
> each machine?
> My script is like this
> *#!/bin/bash*
> *#*
> *# Runs the job from the directory of submission*
> *#$ -cwd*
> *
> *
> *# mergy outputs *
> *#$ -j y*
> *
> *
> *# specify shell to be bash*
> *#$ -S /bin/bash*
> *
> *
> *# transfer all environment variables*
> *#$ -V*
> *
> *
> *# file to send standard output*
> *# -o ssle.out*
> *
> *
> *#file to send standard error*
> *# -e ssle.err*
> *#*
> *MPI_DIR=/home/deqiang/mpich2_install/*
> *SSLE_DIR=/home/deqiang/ssle/*
> *
> *
> *$MPI_DIR/bin/mpiexec -np $NSLOTS -machinefile $TMP/machines
> $SSLE_DIR/ssle < input.in <http://input.in>*
>
> Thanks a lot!
>
> Deqiang
>
>
>
> On Thu, Jun 16, 2011 at 12:21 PM, Deqiang Mao <deqiang at email.arizona.edu
> <mailto:deqiang at email.arizona.edu>> wrote:
>
> Dear Pavan
>
> It works!
> Thanks a lot for your help!
>
> Wish you a good day!
>
> Deqiang
>
>
> On Thursday, June 16, 2011, Pavan Balaji <balaji at mcs.anl.gov
> <mailto:balaji at mcs.anl.gov>> wrote:
> >
> > Yes, deleting the installation folder is fine.
> >
> > -- Pavan
> >
> > On 06/16/2011 01:21 PM, Deqiang Mao wrote:
> >
> > Dear Pavan
> >
> > Do you know the best way to uninstall MPICH2?
> > Just delete the installation folder or something else?
> >
> > Sorry to bother you. I just don't want to mess it up.
> >
> > Deqiang
> >
> > On Thu, Jun 16, 2011 at 11:13 AM, Pavan Balaji
> <balaji at mcs.anl.gov <mailto:balaji at mcs.anl.gov>
> > <mailto:balaji at mcs.anl.gov <mailto:balaji at mcs.anl.gov>>> wrote:
> >
> >
> > On 06/16/2011 12:28 PM, Deqiang Mao wrote:
> >
> > Thanks! Should I uninstall mpich2 and install it again
> using the
> > configuration?
> >
> >
> > Yes.
> >
> >
> > -- Pavan
> >
> > --
> > Pavan Balaji
> > http://www.mcs.anl.gov/~balaji
> >
> >
> >
> >
> > --
> > /Deqiang Mao/
> > /PhD candidate
> > Department of Hydrology & Water Resources
> > The University of Arizona
> > (520)576-3111 <tel:%28520%29576-3111>/
> >
> >
> >
> > --
> > Pavan Balaji
> > http://www.mcs.anl.gov/~balaji
> >
>
> --
> *Deqiang Mao*
> *PhD candidate
> Department of Hydrology & Water Resources
> The University of Arizona
> (520)576-3111 <tel:%28520%29576-3111>*
>
>
>
>
> --
> /Deqiang Mao/
> /PhD candidate
> Department of Hydrology & Water Resources
> The University of Arizona
> (520)576-3111/
>
--
Pavan Balaji
http://www.mcs.anl.gov/~balaji
More information about the mpich-discuss
mailing list