John Burkardt and Gene Cliff have given a number of Introduction to Parallel Computing in Matlab
short courses at Virginia Tech. Feel free to use the course material provided in
the links below as you begin taking advantage of this new capability in Matlab.
Notes (and code) on changing the walltime parameter for the ithaca Matlab queue :
Notes
and PBSClusterInfo.m.
One of the main thrusts at ICAM is scientific computing. Over the past decade, ICAM
has obtained a number of high-end computers to aid in computational mathematics and
scientific computing tasks. Our current computing needs are met by high-end workstations
and access to the Virginia Tech Advanced Research Computing (ARC) resources. In particular,
we take advantage of the high speed reliable networks, computational systems and support
services of ARC, www.arc.vt.edu.
ICAM also has access to smaller compute clusters for algorithm development and
testing.
ICAM Clusters
In Spring 2002, ICAM began building a Linux cluster using AFOSR and SCHEV funds.
The cluster supports LAM/MPI software for parallel computing. The server is comprised
of a two processor IBM xSeries computer which calls on 8 additional dual Athlon processor compute nodes. Currently, the cluster stands at 18 nodes and 38 processors.
Thanks to support from the National Science Foundation, an additional cluster of 24 dual G5 nodes was acquired for Large Eddy Simulation research.
This cluster (named "Mini-Me") mimics the architecture of System X, with
similarly configured compute nodes and network interconnects. Simulation
software developed on mini-me can be ported easily to the main campus
computing platform.
Traian and Jeff begin the task of assembling the Mini-Me cluster.
