Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] Global Communicator
From: McCalla, Mac (macmccalla_at_[hidden])
Date: 2009-02-06 11:59:09

Hi Ted,
Does the "default mpirun command" implementation match the build
environment for quest_ompi.x ?
ie., what mpi implementation (mpich, LAM/MPI, OPENMPI, or other) was
quest_ompi.x compiled/linked with? and does that match the result of
"which mpirun"? You might try running a job using your PBS system that
simply executes the which mpirun command and see what you get.
Mac McCalla


From: users-bounces_at_[hidden] [mailto:users-bounces_at_[hidden]] On
Behalf Of Ted Yu
Sent: 06 February 2009 10:02
To: Open MPI Users; Ralph Castain
Subject: Re: [OMPI users] Global Communicator

Just to make sure, because I have to use open mpi for this program:

I'm using the default mpirun command.

When I type "man mpirun", these are the first few lines:


       orterun, mpirun, mpiexec - Execute serial and parallel jobs in

       Note: mpirun, mpiexec, and orterun are all exact synonyms for
       other. Using any of the names will result in exactly identical

--- On Fri, 2/6/09, Ralph Castain <rhc_at_[hidden]> wrote:

        From: Ralph Castain <rhc_at_[hidden]>
        Subject: Re: [OMPI users] Global Communicator
        To: tedhyu_at_[hidden], "Open MPI Users"
        Date: Friday, February 6, 2009, 7:55 AM
        Hi Ted

        From what I can tell, you are not using Open MPI, but mpich's
mpirun. You might want to ask for help on their mailing list.


        On Feb 6, 2009, at 8:49 AM, Ted Yu wrote:

Thanx for the reply.

I guess I should go back a step: I had used the openmpi version on my
system which is simply:
"mpirun -machinefile $PBS_NODEFILE -np $NPROCS ${CODE}

This did not work because I was just getting a blank output.

I tried this older version because at least i was getting an output.
"/opt/mpich- -machinefile $PBS_NODEFILE -np

$NPROCS ${CODE} >/ul/tedhyu/fuelcell/HOH/test/HH.out"
I think this older version is failing me for whatever reason. Do you
have any clue? I read somewhere that new versions of mpirun adds extra
commandline arguments to the end of the line. Therefore the newer
version of mpirun may be not be giving an output because it sees all
extra commandline arguments after my output file

This is where I'm reading that there are extra commandline arguments for
a version of mpirun:

Again, I'm new at this, and I'm just guessing. Any ideas of where to
turn would be helpful!


--- On Thu, 2/5/09, doriankrause <doriankrause_at_[hidden]> wrote:

        From: doriankrause <doriankrause_at_[hidden]>
        Subject: Re: [OMPI users] Global Communicator
        To: tedhyu_at_[hidden], "Open MPI Users"
        Date: Thursday, February 5, 2009, 11:14 PM
        Ted Yu wrote:
> I'm trying to run a job based on openmpi.
         For some reason, the
        program and the global communicator are not in sync and it reads
that there is
        only one processors, whereas, there should be 2 or more. Any
advice on where to
        look? Here is my PBS script. Thanx!
> #!/bin/sh
> ### Set the job name
> #PBS -N HH
> ### Declare myprogram non-rerunable
> #PBS -r n
> ### Combine standard error and standard out to one file.
> #PBS -j oe
> ### Have PBS mail you results
> #PBS -m ae
> #PBS -M tedhyu_at_[hidden]
> ### Set the queue name, given to you when you get a
> #PBS -q workq
> ### Specify the number of cpus for your job. This example
will run on 32
         ### using 8 nodes with 4 processes per node.
> #PBS -l nodes=1:ppn=2,walltime=70:00:00
> # Switch to the working directory; by default PBS launches
processes from
        your home directory.
         # Jobs should only be run from /home, /project, or /work; PBS
        results via NFS.
> PBS_O_WORKDIR=/temp1/tedhyu/HH
> export
> echo Working directory is $PBS_O_WORKDIR
> mkdir -p $PBS_O_WORKDIR
> rm -rf *
> cp /ul/tedhyu/fuelcell/HOH/test/ ./
> cp /ul/tedhyu/atom_pbe/* .
> echo Running on host `hostname`
> echo Time is `date`
> echo Directory is `pwd`
> echo This jobs runs on the following processors:
> echo `cat $PBS_NODEFILE`
> Number=`wc -l $PBS_NODEFILE | awk '{print $1}'`
> export Number
> echo
> # Define number of processors
> # And the number or hosts
> NHOSTS=`cat $PBS_NODEFILE|uniq|wc -l`
> echo This job has
         allocated $NPROCS cpus
> echo NHOSTS
> #mpirun -machinefile $PBS_NODEFILE ${CODE}
> #mpiexec -np 2 ${CODE} >/ul/tedhyu/fuelcell/HOH/test/HH.out
> /opt/mpich- -machinefile
        $NPROCS ${CODE} >/ul/tedhyu/fuelcell/HOH/test/HH.out
> cd ..
> rm -rf HH
        Please note, that you are mixing Open MPI (API/Library) with
        (mpirun). This is a mistake I like to make, too. If you use
        the ompi mpiexec program, it probably works.
> users mailing list
> users_at_[hidden]

                users mailing list