Seymour Cray Paved The Way For Supercomputers
September 04, 2007
-------------------------------------------------------------------
 
Seymour Cray wanted to talk to his sister.
 
But it was already after bedtime, and the young Crays were supposed to stay in their beds.
 
Young Seymour had a strategy: If he could figure out how to rig a communication system between his room and her room, he could talk to his sister without breaking his parents' rule.
 
As recounted by former colleague Charles Breckenridge, who later became president of SRC Computers Inc., Cray (1925-96) set up a telegraph system between the two rooms so the siblings could chat via Morse code after bedtime.
 
When his father complained of the clicking sounds, the younger Cray found another way: He used blinking lights instead of the telegraph.
 
Cray's reach spanned just a few feet then, yet his vision continued to grow. It eventually became so great, it sparked a computer revolution.
 
Cray's name is synonymous with the supercomputers he developed, a range of high-end machines that perform billions of calculations per second. They're useful for everything from predicting weather to modeling nuclear reactions.
 
At the height of the Cold War, they gave the U.S. a vital edge during its arms race against the Soviets. And as the technology filtered into the marketplace, it changed science, commerce and society.
 
The son of a civil engineer, Cray showed an early interest in science and electronics. In addition to the home Morse code system, Cray eagerly took apart radios, fans and blenders to see how they worked, then put them back together. He read about science voraciously, plowing through textbooks and scientists' biographies.
 
Realizing he needed a solid educational base, Cray studied electrical engineering and applied mathematics in college.
 
After coming home from World War II, Cray followed his passion, landing a job at Electronic Research Associates, an early computer maker. He played a key role in developing the 1101, one of the first computers geared toward businesses.
 
A year into the job, Cray was already gaining a reputation for his design genius.
 
Convinced he'd be better able to make his ideas a reality on his own, Cray left the firm in 1957 to help start a new company called Control Data Systems. His CDC 6600 was the first machine to get the designation of "supercomputer."
 
That wasn't enough for Cray. Seeking to kick up the computer's power, he soon followed up with the even faster CDC 7600.
 
He wanted to improve on that right away. But frustrated with slower progress on the 8600 and his partners' timid approach to it, he left to found Cray Research in 1972.
 
Cray worked diligently on his computer, the CRAY-1, which he unveiled in 1976. It was hailed for its vector processing, a technique that speeds up big calculations. Los Alamos National Laboratory, which bought it for $8.8 million, was its first customer. In his typical drive to improve, he introduced the faster CRAY-2 in 1985.
 
The CRAY line, with its trademark futuristic horseshoe design to cut the length of inside wiring, cemented Cray's reputation in the field.
 
As Joel Birnbaum, a former Hewlett-Packard (NYSE:HPQ) HPQ chief technical officer, once said, it's hard to overstate Cray's contribution.
 
"Seymour combined modesty, dedication and brilliance with vision and an entrepreneurial spirit in a way that places him high in the pantheon of great investors in any field," Birnbaum said. "He ranks right up there with Edison and Bell in creating an industry."
 
Cray's intense focus was a big part of his achievements. "Cray would clear his desk of everything except a pad and a pencil," wrote Charles Murray in "The Supermen." "Then he would sit for hours, scribbling Boolean logic equations and thinking." Sometimes this meant drifting off into his private world and avoiding the interruptions that came with his management duties.
 
Cray didn't like bureaucracy and he eschewed micromanaging, which he considered meddling. Several times during his career he moved his engineers to remote offices to avoid corporate red tape.
 
While his peers often spent millions looking for ways to blaze new paths, Cray sought to solve his engineering problems with a more frugal approach. He insisted on using older, better-tested technologies. For one of his earliest machines, he found bargain-basement transistors -- rejects from an electronics firm -- and designed around their flaws.
 
Despite his reputation as a tech pioneer, he wasn't always the first to embrace new technologies. His natural thriftiness was a lifesaver in the early days of Control Data Corp., Engineering Research Associates and Cray Research.
 
But the approach went deeper than that. It let Cray learn from the mistakes of his rivals and gave technologies time to work the kinks out.
 
This approach remained a trademark throughout his career. He didn't use transistors for a decade after their invention in 1947. The same was true of silicon transistors and integrated circuits. "The pioneer never wins," Cray said, according to Murray. "It's always easier to be the one who goes second."
 
Cray was still working on ideas for his next computer when he was killed in a car crash in 1996.
 
Cray Inc. CRAY bears his name.
 
This story originally ran Dec. 30, 2003, on Leaders & Success.