I have parallel program written on mpi. I am able to run on openmpi environment in our Linux cluster.
I found out that it is rather slow. I have heard that it could be faster if it is run on infinipath environment where
it has lower latency.
I run my program which runs perfectly on openmpi environment first by changing the environment to
using (infinipath/core/gcc/2.1 and infinipath/ofed/2.1 in my case ) and submit my job as:
qsub -R y -pe infinipath 8 myscript.sh
where the script "myscript" contains the command:
mpirun -m $TMPDIR/machines -np $NSLOTS ./test
where test is the executable file that was produced by running my parallel program test.f90
mpif90 test.f90 -o test
After submitting my jobs with a given number of "Nslots", I got the error message as
/usr/local/Cluster-Apps/sge/default/spool/node103/job_scripts/4947620: line 37: mpirun: command not found
Any comments are welcome, thankx ? Do I need to change inside my mpi parallel code to run on infinipath environment ?
----- Original Message ----
From: Andy Georgi <Andy.Georgi_at_[hidden]>
To: Open MPI Users <users_at_[hidden]>
Sent: Sunday, April 20, 2008 9:23:19 AM
Subject: Re: [OMPI users] Deci
sion map for MPI collective operations
Thanks a lot, i got it.
George Bosilca schrieb:
> You need access to the source in order to look at the decision maps.
> They are in ompi/mca/coll/tuned/coll_tuned_decision_fixed.c file.
> On Apr 19, 2008, at 3:41 AM, Andy Georgi wrote:
>> hello all,
>> i want to measure the time needed for collective operations on a cluster
>> by using ompi 1.2.5. i read that the algorithm which is used depends on
>> msgsize and/or commsize and ompi constructs a decision map for the
>> collective on the system. now my question:
>> where can i find these decision map for my system?
>> i'm no sysadmin, but i hope there is a way to get this information
>> anyway ;-).
>> Dresden University of Technology
>> Center for Information Services
>> and High Performance Computing (ZIH)
>> D-01062 Dresden
>> Phone: (+49) 351/463-38783
>> Fax: (+49) 351/463-38245
>> e-mail: Andy.Georgi_at_[hidden]
>> WWW: http://www.tu-dresden.de/zih
>> users mailing list
> users mailing list
Dresden University of Technology
Center for Information Services
and High Performance Computing (ZIH)
Phone: (+49) 351/463-38783
Fax: (+49) 351/463-38245
users mailing list
Be a better friend, newshound, and
know-it-all with Yahoo! Mobile. Try it now. http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ