Interesting People mailing list archives
IP: A MUST READ --Seymour Cray == an "appreciation," aimed
From: Dave Farber <farber () central cis upenn edu>
Date: Mon, 07 Oct 1996 13:11:02 -0400
7 October 1996 The Washington Post, Style section APPRECIATION Cray Saw Beauty in Computers By Elizabeth Corcoran Washington Post Staff Writer According to the old saying, if a man made a better mousetrap, the world would beat a path to his door. Even if he lived in the woods. For decades, Seymour Cray lived up to that adage. From a lab in rural Chippewa Falls, Wis., he designed the Lamborghinis of the computer world and gave them his surname. Building computers was his livelihoodbut it was also his passion. Cray designed the machines to suit his whimsy, each time challenging his imagination to do something that hadn't been done before--to build the fastest machine in the world. Throughout the years, he designed with the joyfulness of a young man in love a blaze with new ideas, unencumbered by earlier mistakes. For that, many in the engineering and scientific worlds thought of him as a hero. He once said he was surprised that anyone else was willing to buy his creations. But they were. Like a Stradivarius violin, a Cray supercomputer became a mark of excellence, of prowess. It also became a symbol of the sophistication of American engineering at a time when its primacy was gravely challenged. Cray, an intensely private man who often said he didn't deal well with people, died Saturday in Colorado Springs of head and neck injuries he received in an auto accident last month. He was 71 years old. For millions of people, Cray was a name best known from spy novels. In such tales, the hero would wangle a few precious computation minutes on a prized Cray supercomputer, jealously guarded behind windowless walls in the heart of the CIA or the National Security Agency. The Cray invariably offered up an intriguing insight into the mystery at hand. But Crays were more than hulking computer muscle. Much as Frank Lloyd Wright became known for the simple grace of his wood and stone buildings, Seymour Cray was revered for the elegant, clean designs of his machines, both inside and out. "I've always been interested in aesthetics," Cray told Smithsonian historian David Allison, who interviewed the engineer last year for an oral history of computing. "So many computer products are rectangular boxes and don't seem to have any aesthetic appeal." When Cray debuted the Cray-1 in 1976, it was cylindrical and featured a cozy, warm seat. It was, after all, located right above a power converter. For a technologist, Cray had a rare genius for design. He invented tricks that made others gasp with admiration. At the heart of the early blueprints, for instance, was the idea that every command the machine carried out must travel through the central processorlike cars filing one by one through a tunnel. Cray invented graceful ways to speed up the traffic by transforming those cars into double-decker buses and magically creating additional routes. One early machine, the Model 6600, put existing electronic leviathans to shame. When it debuted, then-IBM Chairman Thomas J. Watson Jr. issued a biting memo to his staff, demanding to know how Cray's team of "34 people--including the janitor" had bested the industry's mightiest corporation. If Cray had invented only clever new computer designs, he would have been admired and called a genius. But it was not just what he did, but how he did it. Engineering is a conscientious discipline of compromise. Among the most difficult are those that balance past and future technologies. The design of a computer or piece of software includes genes that must be passed down to the next generation for the system to be compatible-- for this year's model to work with last year's. But compatibility has its costs. As engineers work on their designs, they better understand the flaws in their earlier work. To make up for those problems, they must turn to ever more elaborate tricks and sophisticated puzzle piecesall of which makes the final thing more ponderous and complex. That burden becomes like the emotional scars of a world-weary lover, each new relationship shaped by earlier ones. Not Cray. He started every design afresh, literally on a blank sheet of paper. "The blank sheet of paper is not a blank mind," he said. "I wanted to take advantage of all the things that I remembered and all the inputs I had gotten from people, ... By the blank piece of paper, I mean that I liked to start over with the technical details, review all the things that the world offers at this point in time, rather than to reuse things that were just used." Every design was a fresh romance, cut loose from the flaws of the past. Every design was risky. There was little artifice to Cray's approach. He would tell people that he wasn't sure if the ideas would work this time. But it was exhilarating--like falling in love for the first time, again and again. Others watched with admiration. And because Cray did no boastful preening, because he was just the guy in the lab doing what he had always wanted to do, his peers' delight was not traced with bitterness or envy. Cray wasn't trying to best them--he was trying to out-do himself. As time went on, it became harder for Cray to build his new machines. By the early 1990s, an idea that Cray had helped pioneer parallelism, or carrying out multiple operations simultaneouslyhad been exploited by others. While Cray's designs continued to use a relatively small number of enormously powerful microprocessors, massively parallel computers, which harnessed hundreds of punier processors, became the vogue. Cray's last major project, the Cray-4, was designed to use exotic materials that no one else in the computer industry had dared to touch. It almost defeated him. After lengthy delays, his team finished one working machine and closed its books last year. But Cray was undaunted. He was enthusiastically plunging ahead, looking to the human brainfor inspiration. Sighed Gordon Bell, a pioneering computer designer at Digital Equipment Corp.: "I was counting on him to make another breakthrough."
Current thread:
- IP: A MUST READ --Seymour Cray == an "appreciation," aimed Dave Farber (Oct 07)