Home » Technology » Scientific Computing » "Poor Man's Supercomputer" -- The Buffalo Project

"Poor Man's Supercomputer" -- The Buffalo Project

© Adam Hughes

In last week's installment of this series, we looked at how the clustering of workstations as a means of "building" a supercomputer from available resources might possibly eliminate the need to shell out the funds for a more "traditional" machine. While this idea is very straightforward in concept, it's easy to think that such a simple (not easy) setup would be hard-pressed to live in the real world of computer science. In fact, however, this type of system is becoming more of a reality in university and industrial settings, with up to a hundred PC's linked as a single cluster of processors. What's more, as we'll see in this article, the clustering of workstations is not limited to run-of-the-mill PC's.

The Center for Computational Research(CCR) at the University of Buffalo has recently formed a partnership with one of the largest computer companies in the world, Sun Microsystems, to fashion a number of its powerful workstations into a viable cluster configuration. The 64 computers to be used in this project are Sun Ultra 5's with 333 MHz processors and were purchased jointly by Sun and UB. The principle investigators of this project are Russ Miller, director of the CCR, and Corky Brunskill, director of the University's Science and Engineering Node Services (SENS).

While this project admittedly won't replace the well-developed supercomputer anytime soon, it does lend itself well to certain aspects of supercomputing. The UB researchers see the cluster project as a workable alternative for applications such as date mining, web-based applications, mail programs, and even some scientific computational research. As Miller puts it, "For some applications, this type of architecture represents the future of supercomputers. It's the first step toward a whole new methodology of supercomputing."

With such grounded goals and the support of the Sun giant, this project seems to have a bright future. Who knows? Maybe someday scientists will be running their jobs not on supercomputers, but on "superclusters!" This is definitely an effort worth tracking, and more can be learned at the UB web site:

http://www.buffalo.edu/scripts/newnews.

My thanks to Ellen Goldbaum for bringing this project to my attention and permitting me to write about it.

The copyright of the article "Poor Man's Supercomputer" -- The Buffalo Project in Scientific Computing is owned by Adam Hughes. Permission to republish "Poor Man's Supercomputer" -- The Buffalo Project in print or online must be granted by the author in writing.