The most powerful computer in the world, "Janus," sits in a building at Sandia National Laboratories in Albuquerque, N.M., under fluorescent lights and roaring refrigerator fans. It runs on 9,072 Pentium Pro processors, all wired together behind 84 tall gray cabinets. It is capable of carrying out 1.8 trillion mathematical instructions per second. Loaded with such computational horsepower, Janus has one primary job: running simulations of nuclear weapons explosions.
The world's other super computing celebrity IBM's Deep Blue plays chess.
Seems a little uninspired, doesn't it, for super computing in the Digital Age? As supercharged as Intel engineers will tell you Janus is, or as scary-smart Gary Kasparov has conceded Deep Blue to be, there's a school of engineers who just aren't impressed. They're practitioners of another brand of computing distributed, or parallel, computing. They see the ultimate "computer" already at humankind's fingertips: the vast, Internet-linked network of computers that envelops most of the planet.
Consider the numbers: Tens of millions of processors are wired to the Internet at any given time, each typically capable of executing more than 100 million instructions per second. Grouping together just a tiny fraction of the world's collective processing power for whatever computing task or problem you could devise would yield the equivalent of a global Janus at your disposal, 24 hours a day, free. The laughable irony, though, is that most of that networked "24/7" processing power is locked up in a computing task even less inspired than simulating nuke blasts or playing chess screen saving.
When Sun Micro systems co-founder Bill Joy coined the slogan "the network is the computer" more than a decade ago, it's a safe bet he wasn't thinking of flying toasters as humankind's most ubiquitous computing enterprise. But here we are. The lights may be on 24 hours a day along this great network, but is anybody home?
Enter a handful of distributed-computing buffs at the University of California at Berkeley. They think they can find ways to apply these vast (and mostly idle) computational resources to important scientific problems and keep saving those screens at the same time. Their vision? Build screensaver applications that will parcel out massive super computing tasks to thousands, maybe millions of idling computers around the planet, all linked via the Internet. Their your mission, should you choose to accept it? Easy. Find E.T.