The Glade 4.0

WASHINGTON, D.C. -- What do you get when you string 1,760 Sony PlayStation 3 video gaming consoles together?

You get the Condor Cluster -- the biggest, fastest interactive computer the Defense Department has, according to the Air Force. After its formal unveiling this week, it can be used to solve image-matching problems and assist in surveillance situations, using radar enhancement and pattern recognition capabilities.

What, you were expecting a Grand Theft Auto or Pac-Man Championship Edition marathon?

Researchers under the command of Wright Patterson Air Force Base near Dayton have harnessed massive computing power with off-the-shelf PlayStation 3 consoles, or PS3s, linked to more traditional graphical processing computer components. They'll cut the ribbon on the Condor on Wednesday, in a ceremony at the Air Force Research Laboratory in Rome, N.Y., where work on the computer was conducted.

Total cost is about $2 million. That's ten to 20 times cheaper that what traditional computing equipment might cost, according to a posting on Wright Patterson's public affairs website.

Officials directly involved in the project were not available for interviews Monday. But an Air Force public affairs officer pointed out an article in Stars and Stripes, a newspaper that covers the military, that said earlier this year that the Sony consoles will create a supercomputer nearly 100,000 times faster than high-end computer processors sold today.

To lay people and even ordinary video gamers, the harnessing of video gaming technology for super computing may seem unusual. But "unusual is a relative term," said Larry Merkle, assistant chairman of the Department of Computer Science and Engineering at Wright State University.

Video game consoles were developed with cutting-edge graphics capabilities, able to handle extensive numerical computations, Merkle said. That makes them well suited for certain tasks.

According to Airman Magazine Online, the use of PS3s with other aerial surveillance technology will enable scientists to monitor a 25-kilometer area (roughly 16 miles) in real time.

This does not mean that if everyone on your block buys a PS3 for Christmas, you'll be able to create a block-wide supercomputer.

"You would need some basic networking hardware and the know-how to put the networking hardware together," Merkle explained. You'd also need to know how to partition the problem you're trying to solve, breaking it down so the multiple computations could be handled separately and then linked together.

"Certainly your average home does not have these networks based on PS3s," Merkle said.

In other words, go ahead and buy a PS3 -- if you want to play Gran Turismo 5, Madden NFL 11 and other games. Leave the computing to the Pentagon.

In comic strips the person on the left always speaks first. - George Carlin

The fastest cluster, Folding@home, reported over 7.8 petaflops of processing power as of December 2009. Of this, 2.3 petaflops of this processing power is contributed by clients running on NVIDIA GeForce GPUs, AMD GPUs, PlayStation 3 systems and another 5.1 petaflops is contributed by their newly released GPU2 client.[11]

The PlayStation 3 Gravity Grid [18] uses a network of 16 machines, and exploits the Cell processor for the intended application, which is performing astrophysical simulations of large supermassive black holes capturing smaller compact objects. The Cell processor has a main CPU and 6 floating-point vector processors, giving the machine a net of 16 general-purpose machines and 96 vector processors. This cluster was built in 2007 by Dr. Gaurav Khanna, a professor in the Physics Department of the University of Massachusetts Dartmouth with support from Sony Computer Entertainment and is the first PS3 cluster that generated numerical results that were published the scientific research literature.

The 14th TOP500 list was published at the SC99 Conference in Portalnd, OR (13-19, November 1999).

A surprise outcome for this edition of the report is the fact that the duo SGI/Cray is no longer number one in the total number of systems installed. This is the first time this has happened. IBM has taken this spot with 141 systems ahead of SGI/Cray with 133 systems and Sun with 113. With respect to the installed performance SGI/Cray is still in the lead with 19.5 TFlop/s ahead of IBM with 13.7 TFlop/s and Sun with 4.8 TFlop/s.

As of 2010, the fastest PC processors six-core has a theoretical peak performance of 107.55 GFLOPS (Intel Core i7 980 XE) in double precision calculations. GPUs are considerably more powerful. For example, NVIDIA Tesla C2050 GPU computing processors perform around 515 GFLOPS[14] in double precision calculations while AMD FireStream 9270 peaks at 240 GFLOPS.[15] In single precision performance, NVIDIA Tesla C2050 computing processors perform around 1.03 TFLOPS while AMD FireStream 9270 cards peak at 1.2 TFLOPS. Both NVIDIA and AMD's consumer gaming GPUs may reach higher FLOPS. For example, AMD’s HemlockXT 5970[15] reaches 928 GFLOPS in double precision calculations with two GPUs on board while NVIDIA GTX480 reaches 672 GFLOPS[14] with one GPU on board.

The difference is cost, Lex. PS3s cost < $300, and that cost includes the networking interface, RAM, etc. Compare to your "latest processor," and the cost for that processor alone is going to surpass the PS3's cost. That's why the PS3 is being used.