A $US1,000 supercomputer?

Within 18 months, you may be able to put the equivalent of today's supercomputer on your desktop -- for about $US1,000. The CPU (central processor unit), long the heart of all PCs, may be an unnecessary component in tomorrow's high-performance computers.

The new computer will be able to process 100 billion instructions per second, according to Kent Gilson, chief technical officer of Star Bridge Systems. Company representatives discussed their plans for a high-end PC this week while announcing HAL-300GrW1, a "hypercomputer" that is said to be 60,000 times as fast as a 350-MHz Pentium, and many times as fast as IBM's supercomputer Pacific Blue. (The test used to measure the HAL's performance was different from the measure used for Pacific Blue, so exact comparisons are difficult.) The new $1,000 computer will be "three orders of magnitude different in price-performance (ratio)" from today's PCs, Gilson claims. It will fill many of the roles of a supercomputer, such as voice recognition, natural language processing, and holographic displays, he says. What's more, Gilson says, this super-PC will "run PC applications in emulation mode, in a manner similar to how the Digital Alpha runs NT, but it will run it a lot faster."

HAL comes first

Although Gilson claims the hardware for such a PC is ready now, and that Star Bridge Systems has completed the programming language, called Viva, the company's initial focus is on its high-end hypercomputer line, HAL. The HAL-300GrW1 has a price tag of about $26 million, so it doesn't take a hypercomputer to understand why Star Bridge Systems has chosen to direct its attention to the HAL line first.

"We're a small company. If we came out with a PC, we wouldn't be able to sell enough (to fund the company), but we can sell hundreds a year of the high-end ones, so it just makes sense," Gilson says.

In today's computing terms, the architecture Star Bridge Systems has developed is a "massively parallel, ultratightly coupled, asymmetrical multiprocessor." It is based on a processor called a field programmable gate array (FPGA), Gilson says. FPGAs can be programmed on the fly, so their configuration can be changed to perform the particular task at hand most efficiently.

FPGAs can be changed thousands of times per second. So in essence, an FPGA can become a specially designed CPU tailored to perform a required task right when you need the new processing architecture.

The traditional CPU, by contrast, has a fixed instruction set that is burnt into silicon. Programming instructions are written to work with the instruction set, and are limited by the capabilities built into it.

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

More about IBM AustraliaStar Bridge Systems

Show Comments