Advertisement

Legendary Bell Joins Microsoft : Technology: Designer will co-direct research into harnessing a company’s desktop resources into a ‘supercomputer.’

Share
TIMES STAFF WRITER

Famed computer designer Gordon Bell, inventor of the seminal VAX minicomputer, has joined Microsoft Corp., for which he will establish a new San Francisco research laboratory and pursue a novel concept in high-performance computing.

Bell has acted as a Microsoft adviser for the past four years, mostly offering guidance on future technologies. Now he will join Jim Gray, another highly regarded computer scientist, on a project called SNAP (for “scalable networks and platforms”). The idea is to harness all the computing resources of a company and, in effect, turn them into a supercomputer.

“The power for doing this is all in Redmond,” said an ebullient Bell, referring to Microsoft headquarters in Washington state. “You really couldn’t do this anywhere else.”

Advertisement

Bell, former head of research and development for Digital Equipment Corp., made his name in the 1970s as the inventor of Digital’s VAX, among the most successful computers ever created. It allowed upstart DEC to challenge IBM and created a broad technological alternative to the giant mainframe computer.

Gray, a UC Berkeley-trained computer scientist, spent a decade at an IBM research laboratory in San Jose, followed by another 10 years designing software for Tandem Computers Inc., the pioneer in so-called fault-tolerant computers.

Bell and Gray’s new project is based on the idea that high-speed computer networks will allow anywhere from a small cluster to hundreds of personal computers to work in concert within a given company, Bell said.

Previously, it had been thought that lashing together microprocessors to create one powerful machine was the way to create an affordable supercomputer. Bell had subscribed to that notion until he realized that bottlenecks--such as 10 processors trying to access one pool of memory--slow the performance of such machines. And it is difficult to manufacture that kind of computer for less than $100,000, he said.

“To build really large-scale systems, you need commodity software and hardware,” he explained.

Bell and Gray became acquainted a couple of years ago when Gray, who had read Bell’s latest computer science theories in a number of trade journals, set out to meet the legendary inventor.

Advertisement

“He said, ‘You and I agree on this and we’re the only two who do, so we should work together,’ ” Bell recalled. At the time, Gray had decided to take a year off from work.

Gray emphasized that the key to their project--indeed, the key to any new type of high-performance computing system--is software.

“It doesn’t require a behemoth [computer] in the basement,” Gray said. “All the computing power is on the desktop. But for each user, you want it to look like a single computer when you are really getting the benefit of a bunch. That takes software.”

Bell’s association with Microsoft goes back to 1991, when he helped the company establish its first research laboratory. Through Bell, Microsoft was able to lure a number of top computer scientists--including Rick Rashid of Carnegie-Mellon University, inventor of the Mach operating system, considered to be among the most advanced pieces of software ever created for the PC. Rashid is now general manager of the Advanced Technology Group.

Microsoft recently became intrigued with SNAP, which Bell and Gray based on Microsoft’s NT operating system.

Advertisement