Octave on clusters/HPCs

[Daniel Grunberg]

I was wondering if anyone had experience running Octave on clusters (either white-box Linux clusters or HPC machines like SGI’s Altix), either using MPI or simpler embarrassingly parallel-type computations?

 

At Interactive Supercomputing (www.interactivesupercomputing.com) we produce Star P, which is a product that allows users of interactive mathematical programs (like MATLAB or Octave) to transparently run their calculations in a distributed fashion on a cluster (to handle **very** large matrix calculations). 

 

We are thinking about providing an option to allow the user to easily run an Octave script simultaneously across clusters, with the data distributed automatically, and the answers collated directly onto the client (desktop) machine.  You would then have the power of Octave (or Matlab) to do further operations on your results.

 

I would appreciate any comments from users of Octave on clusters, in particular on really large problems, which is what we are targeting.  Does it work well?  Any issues people have encountered?

If you want, feel free to contact me directly (I can provide a phone # if you wish).

 

 

-Dan

Daniel Grunberg

Interactive Supercomputing

dgrunberg /AT/ chainwave.com

Visit the MIT Museum:  http://web.mit.edu/museum