Reclusive Inventor Honored

WASHINGTON -- Seymour Cray enjoys the solitude of his lab, where for the past four decades he has labored to build the world's fastest computers. This man of practiced eccentricity almost never sits for interviews. "He prefers to do what he enjoys doing best, and that's designing and pushing the frontiers," said Terry Wilcom, president of Cray's company, Cray Computer Corp. But every now and then, he surprises people. Last week a beaming Cray, looking more like a retired dairy farmer than one of the age's commanding intellects, sat at the ComputerWorld Smithsonian Awards in Washington, then walked to the podium to accept an honor for innovation. The crowd applauded hard; Cray, 68, smiled anxiously. There was no hint of the trouble his company has encountered as it sticks with an old-style design that it insists is the best. Many investors, in fact, had expected it to close its doors this month. But last week it got a $17 million loan that should take it through late in the year. Cray's name is synonymous with the supercomputer, a term that means the fastest class of machine existing at a given time. Today's supercomputers conduct billions of calculations per second. They aid science, industry and the military through tasks as diverse as predicting the course of a hurricane, simulating a nuclear bomb blast or helping design an auto engine. Born in Wisconsin, Seymour Cray co-founded Control Data Corp. in Minnesota in 1957, then left the company to create Cray Research Inc. In 1989 he moved on again, this time to set up Cray Computer, where he is now based in Colorado Springs. He is the hub of the wheel in the shop there, working closely with several teams. Cray's love from the start has been the product, not the management, though to keep customers and investors happy, he has had to trot out occasionally to shake hands. "I really love blank paper," he said in 1988 in one of his rare surfacings. "Thank heaven for start-up companies or we'd never make any progress." People who know Cray describe him as intensely private and uneasy with all the fuss his work has engendered. He sees himself as a craftsman. "I just make a tool and other people do things with it," he said in a brief interview after the ceremony, where he won an award sponsored by MCI Communications Corp. Cray generally turns aside a reporter's efforts to get him to talk about the impact computers have wrought. But he did say that "the real question is how can we as humans, who have a very slow rate of development, compete with something that's advancing exponentially." People are managing the flood of information well, Cray said. "I'm pretty pleased, so far. God's taking care of us." Receiving the award on the 50th anniversary of D-day, Cray offered only a few sentences to his audience, many of whom had never seen the legendary man. He noted that a St. Paul, Minnesota factory where he once worked had produced gliders for the invasion. That said, he thanked the group and stepped off to thunderous applause. People were disappointed, but most of them understood. In the computer industry, "Seymour stories" abound: He once hand-dug an underground tunnel in his yard in Minnesota; whenever he hit an impasse in system design, he would retire there with shovel in hand and work up a sweat. Ideas came. For several summers running, he designed a sailboat, built it by hand, then held a party to burn it at the end of the season. "Start anew" is his constant message to his associates. His latest effort to do that is the Cray-3, which can handle about 4 billion calculations per second. Many analysts, however, predict it will be the last of a breed. Its inventor, they say, founded an industry, then failed to catch on to new technology. His machines are based on the traditional "architecture," or design, of a small number of processors that each operate extremely fast. Much of the rest of the industry, including his Cray Research, is experimenting with stringing together huge numbers of relatively slow processors for so-called "massively parallel processing." This, they say, can bring economies of scale to an industry where machines to date have been largely built by hand and sold for tens of millions of dollars. But Cray sticks with the old approach, and the market is not biting. To date it has built only one prototype Cray-3. It has no orders for more machines. Investors are bailing out. In 1991 the company's shares sold for a high of more than $19. Last week, they closed at just over $1. "We'd all like to see Seymour successful," said Gary Smaby, president of Smaby Group, a Minneapolis firm that does market research on the supercomputer industry. "He's an icon for the industry." "I'm sure that every time he reads the obituaries for the company, he says, 'I'm going to show them.' He's not out till he's out."