Supercheap Supercomputer

Illinois researchers make US $50 000 cluster from Sony Playstations and off-the-shelf software

2 min read

The same hardware that lets millions of gamers race cars or battle giants may soon allow physicists to probe the foundations of reality. Researchers at the National Center for Supercomputing Applications and the computer science department at the University of Illinois in Urbana-Champaign have created a cluster of Sony Playstation 2 game consoles [photo] for just US $50 0000, hijacking specialized graphics hardware for scientific supercomputing.

One of the researchers, Mike Showerman, explains why the Playstation 2 had caught their eye after Sony Corp. released a Linux kit for the console last year. ”We saw a potential at such a low cost that we decided that it was just something that we should be playing around with,” Showerman told IEEE Spectrum. With the kit, users can install the popular Linux open-source operating system, turning each console into a general-purpose, network-ready computer. So, the researchers used the kit to network the consoles together and, more importantly, get access to each Playstation’s sophisticated graphics hardware. This can process as many as 66 million polygons a second, which is what allows the Playstation to render realistic 3-D worlds so well. For mathematical reasons, each polygon vertex is represented as a vector, an array of numbers with four elements that represent positional information. At the heart of the Playstation’s graphics hardware is a pair of ”vector units”—dedicated arithmetic processors designed to operate on vectors.

”You can use them for generalized matrix-matrix and matrix-vector calculations by building up software libraries” from the arithmetic performed on the four-term vectors, says Craig Steffen, another researcher on the project. Many scientific problems are tackled using vectors and matrices.

Developing these libraries is the main focus of the current research. Ultimately, it would allow specialized physics software for things like quantum chromodynamics computation to run on the cluster by replacing normal mathematical software libraries with software libraries designed for the graphics hardware. But there are a number of obstacles. One is the problem of developing a more efficient way to move data in and out of the vector units. ”We’re going to have to see what tricks we can play with the hardware,” says Steffen. Another is that the vector units do not fully comply with IEEE 754, the standard for hardware computation of floating-point arithmetic.

Ultimately, the purpose of the university’s Playstation 2 effort is to prepare for the 2005 release of the Playstation 3. The hope is that if useful scientific computation can be demonstrated on the current system, there will be a road map for the Playstation 3, which will feature even more powerful graphics hardware, still at a fraction of a PC’s cost.

This article is for IEEE members only. Join IEEE to access our full archive.

Join the world’s largest professional organization devoted to engineering and applied sciences and get access to all of Spectrum’s articles, podcasts, and special reports. Learn more →

If you're already an IEEE member, please sign in to continue reading.

Membership includes:

  • Get unlimited access to IEEE Spectrum content
  • Follow your favorite topics to create a personalized feed of IEEE Spectrum content
  • Save Spectrum articles to read later
  • Network with other technology professionals
  • Establish a professional profile
  • Create a group to share and collaborate on projects
  • Discover IEEE events and activities
  • Join and participate in discussions