Re: [Condor-users] Quick n Dumb MPI Question

[Philipp Kolmann <kolmann@xxxxxxxxxxxxxxxx>]

On Mon, Apr 25, 2005 at 11:07:58PM -0500, Matt Baker wrote:
> I'm currently setting up a 10-node cluster to run MPI under Condor. This
> cluster has the latest stable ROCKS configuration, with MPI 1.2.6.
> I know I need to use 1.2.4 for Condor to run jobs, but I'm not that fluent
> in MPI yet.
me neither, but i am a step a head of you ;-)
 
> 1. Does 1.2.4 need to be installed on compute nodes, or is just mpicc needed
> on the head node?
I have the libs availeable, but they don't seem to be needed to execute. List
the needed libs:

ldd executable:
ldd xhpl-mpi-condor-icc 
        libpthread.so.0 => /lib/tls/libpthread.so.0 (0xb7fd4000)
        libm.so.6 => /lib/tls/libm.so.6 (0xb7fb1000)
        libc.so.6 => /lib/tls/libc.so.6 (0xb7e7d000)
        /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0xb7fea000)

then you see which libs are needed to execute.

If you compile with 1.2.6 you will get errors, that the mpi communication can
not be set up!
 
> 2. Are there plans for 1.2.x / MPICH2 / whatever MPI under Condor?
I hope so too...

> 3. What's the (probably obvious) problem in condor_submitting a shell script
> that calls mpirun instead of using the MPI universe?
you still need a .maschines file for the mpirun. where should you get this
from?

Thats the job of condor, to reserve the needed nodes for you and distribute
your job accordingly

HTH
Philipp kolmann
University of Technology, Vienna, Austria

-- 
If you have problems in Windows: REBOOT
If you have problems in Linux:   BE ROOT