When Sony launched the PlayStation 3 in 2006, the company touted its IBM Cell processor as one of the key features that distinguished the PS3 from other consoles. This “supercomputer on a chip,” would allow developers “for the first time can create games closer to actual intelligence instead of artificial intelligence,” Sony claimed. Although game developers are only now beginning to take advantage of the PS3’s processing ability, the United States Air Force has taken the claim literally.
When complete, the PlayStation 3 supercomputer project will cost $2 million.
Today, Stars and Stripes newspaper announced a $2 million government project to create a research supercomputer using 2,000 PS3s. The project will be headed by the Air Force Research Laboratory in Rome, New York.
According to Stars and Stripes,
Key to the whole idea is the console’s cell processor, which was designed to easily work in concert with other cell processors to combine processing power and has been critically acclaimed for its number crunching ability.
This lets the researchers leverage power toward running such applications as Back Projection Synthetic Aperture Radar Imager formation, high definition video image processing, and Neuromorphic Computing, which mimics human nervous systems.
Richard Linderman, a senior research scientist at the Air Force laboratory, explained that “this will be far and away the largest interactive high-performance computer.” Linderman also noted that the project uses Linux, an operating system that Sony has supported on the PS3 since day one.
Low cost supercomputer projects that use off-the-shelf components are not new. In 2003, I worked closely with Compaq Computer Corporation to create a case study series titled, “Compaq High Performance Computing.” By using standard components, Compaq’s supercomputer line, known as the Alpha Server SC, attempted to avoid some of the risk associated with supercomputer development.
One of the advantages the Alpha Server SC team was its ownership of the DEC Alpha microprocessor. Alpha, which was manufactured by Mitsubishi of Japan, was first developed in 1993. At the time, it was world’s best performing microprocessor for scientific applications, and became the core of DEC’s next generation servers.
A decade later, Alpha still outperformed most microprocessors on the market, including Intel’s Itanium. Launched in 2001, after seven years of development and $2 billion in R&D expenditures, Intel’s Itanium was a 64-bit processor capable of running at 800 MHz. It was also described as a “massively complex and [expensive] CPU which didn’t give the expected stellar performance.”
The architectural innovation that Compaq brought was scalable machines made by connecting commodity components. This approach reduced the development risk and moved Compaq away from the “one mistake and you die” model that characterized prior generations of supercomputers. The Air Force project is another example of how scalable designs are substituting for proprietary designs in the market, although purpose-designed supercomputers are still needed for specific applications.
Prof. 
Very nice! I wonder what kind of research they are going to conduct on it? Or if it can play Metal Gear Solid 4?
I expect these will be heavily modified with faster interconnect switches, etc. The end result will be nothing like a standard PS3.
One of the more interesting projects that will be conducted at the laboratory is neuromorphic computing. These are early attempts to create intelligence that truly mimics the complexity of the human brain.
In “Neuromorphic Computing: Teaching New Brains Old Tricks,” Brad Minnery observes that “Deep Blue may be able to mop the board with most humans in a game of chess, but pit IBM’s vaunted supercomputer against your average canine in a game of ‘fetch my slippers,’ and [the dog] wins every time.”
On the other hand, neuromorphic computing could one day provide “adaptive computer interfaces, intelligent tutoring software, personal devices that sense a user’s mood and social context, prosthetic devices for the disabled that integrate directly with the nervous system, and even intelligent household robots that do all the little things—like retrieve a pair of slippers.”
Minnery also claims that this research could have implications for the video game industry.
Of course, how much value that will have in real life remains to be seen. Currently, gamers can obtain the same results by going online and competing against any number of human opponents.
With Sony selling the PS3 at a loss and hoping to regain the losses in game & accessory sales, I wonder how Sony does financially in this situation. (I also wonder what the did with all the extra controllers…)
The idea behind selling consoles at a loss is that you make it up through software royalties. In this case, the Air Force will not be buying any licensed software to compensate for Sony’s hardware losses.
Interestingly, Compaq also did not make any money selling their supercomputers to national laboratories. Instead, they saw it as a PR opportunity that would help them sell high end servers to their corporate customers. I am not sure that Sony can claim the same type of benefit.
I’m curious as to why ARFL is doing this. The Cell as a platform has been largely abandoned by IBM for computing, plus the new PS3’s no longer support the Linux install.
Besides that, the PS3 Cell is crippled with only 6 active SPEs and 256MB of RAM. Plus, it’s the original single-precision floating point version without the pipelined double-precision engine, so any high precision numerical computing tasks will suffer a rough 10x performance penalty.
Also, forget about using the GPU, as that’s locked out on the basic Linux install. As for connectivity, the best it can do is gigabit ethernet. The PS3 sounds attractive due to it’s price and off-the-shelf nature, but honestly, I feel as though there are better solutions.
IBM created a blade, the QS22 (http://www-03.ibm.com/systems/bladecenter/hardware/servers/qs22/) for computing with the Cell. It has 2 full (8 SPE) Cells with up to 32GB of RAM and an Infiniband interconnect if one desires. Granted, it’s much more expensive, but it’s purpose built to crunch numbers.
Finally, at retail, 2000 PS3s should only cost roughly $600,000. What are they doing with the other $1.4MM?
For this type of research, throughput is less important. In performance tests, the execution time on dual Quad-Core Xeons was found to be on average 84 percent of the the PS3 Cell Processor.
When you factor in the significantly lower cost of off-the-shelf versus custom solutions, the Air Force’s decision seems to make more sense, particularly now that calls for fiscal restraint are getting louder.
Finally, government projects require significant overhead to cover installation, maintenance, and (as mentioned previously) customization. That should account for most of the cost differential that you describe.
Interestingly, about a week later news broke that Microsoft decided not to bulk sell xbox 360s to the Army, although both parties had a different purpose. The Army planned on using the 360s for military training, which could lead to bad PR for xbox. Also, I don’t think they would be buying very many games so Microsoft doesn’t really benefit financially either. I wonder what Sony gains out of this because they’re certainly losing money on just selling the ps3. This is the first time I’ve read this story so they didn’t exactly do a PR push with it.
Interesting read. Wonder how much power 2,000 PS3 consumes. Human brain in just one average person eems to run more efficient than 2,000 ps3.