Parallel Octave possibilities?

[E. Joshua Rigler]

I will soon have access to a 16 node, 32-processor cluster (Linux
AMD64), and am hoping to experiment a little with Octave in parallel.  I
have a few quick questions whose answers I couldn't find in my own email
list archives (the octave.org search engine seems to be down):


1a) What exactly is the "--with-MPI" option in Octave's configure script
for?  Is this an evolved version of Octave-MPI (by Alex Verstak)?  Is
there an extant web page or some other documentation for this (I could
only find this currently non-responding link
http://corto.icg.to.infn.it/andy/octave-mpi/)?  

1b) I seem to have the LAM MPI libraries on my system, but Octave's
configure gets a bunch of "undefined references" when it tests "-lmpi". 
The libraries are version 6.5.9 (default on Redhat Enterprise 3).  Are
these just too old, or do I need other MPI libs?

2) How does this compare with Javier Fernández Baldomero's MPI tool box
(MPITB) fo Octave (http://atc.ugr.es/javier-bin/mpitb)?  Are they even
mutually exclusive?

3) I assume both of these are designed for passing information amongst
clustered Octave instances.  Has anyone actually had luck linking Octave
to a PBLAS and/or ScaLAPACK to simply get scalable linear algebra
functionality?


Other than a little experience compiling mutli-threaded ATLAS libraries
and linking them with Octave, I'm am pretty much a novice here, so any
feedback would be greatly appreciated.  Thanks.

-EJR



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------