The fastest supercomputer in the world doesn't really exist. Or rather, it exists, but only in a virtual sense: as a network of machines scattered around the globe, linked via the Internet to three powerful servers grinding away in a lab at the University of California, Berkeley. SETI@home, as the sprawling system is known, has been online for about a year searching for signs of extraterrestrial intelligence by tapping into the spare number-crunching power of (at last count) 2.3 million PCs.
It's called community, or distributed, computing, and the phenomenal popularity of SETI@home has spawned something of a distributed-computing craze. This fall several Internet start-ups and not-for-profit groups have launched new initiatives in fields ranging from gene mapping to drug design, hoping to harness spare PC processing cycles and either give them away or sell them to the highest bidder.
Becoming part of a distributed-computing network is easy. You simply log onto a website and download free data-crunching software that doubles as a snazzy screensaver. Anytime you go online after that, the computer contacts a server that beams you a small packet of data for your PC to analyze. The calculations are performed whenever your computer is idle; the next time you log on, the results are beamed back to the server, where they are combined with those sent in by other PCs. Like any other screensaver, the system is utterly unobtrusive. Tap a key, touch a mouse, and it immediately shuts down.
Such proletarian computing systems are doing some impressively high-end work. A Stanford University lab recently launched Folding@home, a public project that uses the technology to investigate how proteins fold into the shapes that determine their function. Meanwhile, Distributed.net is working with the U.K.'s Sanger Centre to help map the human genome.
Over in the private sector, Netpreneurs are trying to tap into the same community computing power for commercial purposes, selling it to industries in dire need of supercomputer muscle. One lucrative market: biotechs and pharmaceuticals trying to analyze data from the Human Genome Project. "The genome is so huge, and it takes so long to analyze even a single protein, that there's no way to do it without resorting to some sort of distributed computing," says Vijay Pande, who runs Folding@home.
Not everybody likes the idea of someone else cashing in on their free computing cycles. "If somebody's making a dollar off my machine," notes David McNett, president of Distributed.net "I'd damn well want to be compensated for it." To sweeten the deal, many community computing start-ups let participants donate a portion of their computational cycles to pet biomedical charities. People who sign up with the San Diego-based firm Entropia, for instance, can allot part of their computer's time to the FightAIDS@home project, which uses distributed computing to help Arthur Olson's lab at the Scripps Research Institute search for new anti-AIDS drugs. Others enroll participants in raffles for free airline tickets or cash prizes. Apparently, virtue has its price, even for computers with time on their hands.
--By Unmesh Kher