Cray, Seymour Roger

(b. 28 September 1925 in Chippewa Falls, Wisconsin; d. 5 October 1996 in Colorado Springs, Colorado), major figure and pioneer in the design and engineering of supercomputers, whose technological genius was responsible for a series of innovations in the building of very fast computers.

Cray was one of two children born to Seymour Roger Cray, a civil engineer who worked for an electric power company and as a city manager in northwestern Wisconsin, and Lillian Scholer, the daughter of a Methodist minister and a homemaker. While growing up Cray exhibited a talent and passion for science and mathematics and tinkered extensively with electrical apparatus around his home. After graduating from Chippewa Falls High School in 1943, Cray entered the U.S. Army, spending time in Europe and south Asia during World War II as a radio communications specialist.

After returning from military service Cray entered the University of Minnesota to study engineering, like his father before him. In 1947 he married Verene Voll, the daughter of a Methodist minister. The couple had two daughters and one son and divorced in 1975. Cray completed a B.S. degree in electrical engineering in 1950 and an M.S. degree in applied mathematics in 1951. His seemingly innate talent for both the practical and the applied, nurtured by combined study in both engineering and mathematics, would later serve him well in the field of computer science.

Cray’s first job after completing college was in Minneapolis with Engineering Research Associates (ERA), a company formed after World War II to perpetuate the U.S. Navy’s cryptography operation. The cold war push to build ever more complex weapons and defense systems—which began after the war and continued without diminution until the early 1990s—was a persistent stimulus to the designing and building of fast computers. The first head of ERA was William Norris, also an electrical engineer, who had worked during the war with the navy’s cryptography laboratory.

Soon after Cray began work with ERA in 1951 the company was sold to Remington Rand, and within four years Remington Rand merged with the Sperry Corporation to form Sperry-Rand. Cray’s first experience in building marketable computers was with Remington Rand’s UNIVAC computers. The culture of this large corporation was not conducive to the innovation that both Norris and Cray thought was necessary to engineer computing machines, so Norris left Sperry-Rand in 1957 to found Control Data Corporation (CDC) in Minneapolis/St. Paul. Cray followed within a few months. Cray, with his rapidly developing command of all aspects of computer design and engineering, quickly became the mastermind of CDC’s computers, and the CDC 1604 reached the market in 1960 as the fastest computer in the world. One major innovation in the 1604 was the use of transistors, which had been introduced about ten years earlier by Fairchild Electronics, to replace the vacuum tubes that had been the standard in computers up to that time.

In 1962, because of his desire for independence and freedom from interference, Cray reached an agreement with CDC that allowed him to move his project to Chippewa Falls, about eighty miles from company headquarters. This was to remain his base of operations for more than twentyfive years. The CDC 6600, now generally recognized as the first supercomputer, was completed and marketed during 1963. The CDC 6600 introduced several innovations, including the use of Freon as a coolant and the use of silicon-based transistors. This computer, the fastest in the world at the time, could carry out 3 million interactions per second. However, to give a perspective on the ephemeral nature of the description “supercomputer,” the desktop computer of the late 1990s could perform more than 1 billion interactions per second.

With the success of its CDC 1604, CDC 6600, and CDC 7600, Control Data Corporation quickly became the third largest computer manufacturer in the world (IBM was first, Sperry-Rand second). Supercomputers were required in the advancing fields of high technology and science, and they were especially in demand by U.S. military laboratories. In a relatively short time Cray had led this start-up company to beat far larger and more established companies in the race to make the most advanced computing machines of the time. The CDC 7600, which was marketed in 1969, became a widely used supercomputer during the early 1970s. In 1972 Cray was awarded the Harry Goode Memorial Award for outstanding achievement in the field of information processing.

Cray left CDC the same year, discontented with actions of the company that he perceived to be intervention, to form his own company, Cray Research Corporation (CRC), also in Chippewa Falls. The new company’s first computer, the CRAY-1 introduced in 1976, brought CRC success and recognition as the leader in producing supercomputers. One of its notable innovations was the use of vector processing, which replaced the earlier technique of scalar processing and allowed faster simultaneous calculation rather than series operations. Cray Research was the leading producer of fast computers through the late 1970s and early 1980s. Once again Cray’s engineering genius had taken another company to the top. In 1980 Cray married his second wife, Geri M. Harrand, and became stepfather to her three children.

The CRAY X-MP, introduced in 1982 and engineered by a team led by Steve Chen, had multiple processors and was a faster version of the CRAY-1. The commercial success of the CRAY X-MP boosted Cray Research during a period when the company was struggling to complete its next project, the CRAY—2. However, Cray believed the CRAY–2 had become too encumbered with the “upgrade” image as an improvement of the CRAY—1. He tended to create everything from a blank page, and this led him to look beyond the CRAY–2 to the CRAY–3, allowing others to complete the CRAY–2. Straying from his normal pattern of using only tested materials and components, Cray attempted to use gallium arsenide, rather than silicon, for the circuits in his next computer. Gallium arsenide was known to permit an increased rate of transfer of electrons at switches and a concomitant reduction in heat generation, compared with silicon. But Cray never succeeded with gallium arsenide and returned to silicon.

In 1989, feeling that fresh surroundings and a new start were needed to sustain his drive to complete the next supercomputers, Cray, along with his wife, moved a branch of Cray Research from Chippewa Falls to Colorado Springs, Colorado. The following year he reached an agreement with Cray Research to partially finance a new company, Cray Computer Corporation, whose immediate goal would be to complete the CRAY-3 project. Although the CRAY—3 supercomputer was completed in the new company, it was not marketable and none were sold. Cray Computer filed for protection under bankruptcy law in 1995.

The supercomputer industry was changing, in no small part because the demand for such machines had lessened with the end of the cold war, an event signaled in late 1989 by the fall of the Berlin Wall. In the early 1990s Cray Computer and Cray Research were the only companies producing supercomputers. The wide availability and low cost of microprocessors also brought revolutionary changes in the use of personal computers (PCs), especially the capacity of PCs for integrated uses through connections with other, large-capacity computers. The industry was thus drawn in new directions.

Even after the failure of Cray Computer Corporation in 1995 Cray still believed there was a market for supercomputers. In 1996 at the age of seventy he began to raise capital to found a new company. But this new venture ended when Cray was seriously injured in an automobile accident in Colorado Springs on 22 September 1996. He died two weeks later. His remains were cremated and scattered in the Colorado mountains at an undisclosed location.

Seymour Cray is recognized as the single most successful designer and builder of supercomputers during the first thirty years of the development of the computer industry. A number of traits contributed to his success, among which was his facility with both the applied and the theoretical. Cray insisted upon personally knowing and working on every detail of a project, and his single-minded dedication and concentration to the tasks at hand led him to resist any interruption. He was frequently in conflict with management in attempting to protect his time and that of his staff. In 1989 Cray declined to accept the National Medal of Technology because it would have required his taking a day or two from his work to go to Washington, D.C., for a ceremony with President George Bush.

Another characteristic conducive to Cray’s creativity was his personal belief that innovation must begin with a clean slate and not be encumbered by existing ways of doing things. This penchant for starting anew on every venture, and for restarting some projects by clearing out everything and starting from ground zero, has been called “The Cray-Way” by Charles J. Murray. Cray’s desire to pursue original engineering on every project was in interesting contrast to his aversion to using newly discovered materials or invented components (the gallium arsenide effort being an exception). For instance, he delayed the use of transistors, silicon-based transistors, and integrated circuits until they had been tested for years in other products. Cray is the undisputed major figure in the history of supercomputer architecture and engineering. His technological genius remains the hallmark of the supercomputer era.

Cray’s professional life, as well as a history of the development of supercomputers, is given by Charles J. Murray in The Supermen: The Story of Seymour Cray and the Technical Wizards Behind the Supercomputer (1997). R. Slater, Portraits in Silicon (1989), presents additional biographical information. An account of the role of supercomputing and Cray’s contributions in the advancement of science and high technology is provided by William J. Kaufmann III and Larry L. Smarr in Supercomputing and the Transformation of Science (1993). See also Philip Elmer-Dewitt, “Computer Chip off the Old Block: Genius Seymour Cray and the Company He Founded Split Up,” Time (29 May 1989); Russell Mitchell, “The Genius,” Business Week (30 Apr. 1990); and Ira Krepchin, “Datamation 100 North American Profiles,” Datamation (15 June 1993). Obituaries are in The New York Times (6 Oct. 1996) and Washington Post (6 and 7 Oct. 1996).

W. Hubert Keen