IP: a marvelous obit re: Seymore Cray by John Markoff

[Dave Farber <farber () central cis upenn edu>]

   The New York Times, October 7, 1996, p. A47.




   Seymour Cray, Computer Industry Pioneer and Father of the
   Supercomputer, Dies at 71


   By John Markoff




   Seymour R. Cray, a computer industry pioneer and the father
   of the supercomputer, died yesterday at a hospital near his
   home in Colorado Springs. He was 71 and had been in the
   hospital since an automobile wreck two weeks ago.


   Officials at Penrose Community Hospital said the cause was
   the severe head injuries Mr. Cray had received in the
   accident.


   Mr. Cray, who as a young electrical engineer at the Control
   Data Corporation in the late 1950's led the design of the
   world's first transistor-based computer, went on to develop
   a string of supercompupting machines that were known for
   their elegance and simplicity, but most of all for their
   blazing speed.


   Used first by military weapons designers and in
   intelligence agencies, Mr. Cray's first supercomputers,
   notably the Control Data 6600, 7600 and the Cray 1,
   permitted researchers to simulate nuclear weapons
   explosions and crack enemy codes. They were soon turned to
   tasks like weather prediction and oil exploration.


   Known as an idiosyncratic and quirky computer wizard, Mr.
   Cray had a remarkable ability to focus on a single
   challenge: the need to extract more and more speed from
   each new machine he designed.


   "He had a profound effect on the computer industry," said
   John Rollwagen, a business executive who worked with Mr.
   Cray for many years at Cray Research Inc. "He was always on
   the leading edge."


   Danny Hillis, a supercomputer designer who early in his
   career adopted Mr. Cray as his role model, said: "You
   rarely see someone who knows their calling so precisely. He
   knew every transistor and every wire in his computers."


   Mr. Cray's legacy of scientific computer designs stretches
   as far back as the Univac 1103, which he began working on
   in the mid-1950's.


   During the 1970's and 1980's, Mr. Cray was instrumental in
   creating a constant stream of design advances and
   innovative technologies that were later adopted by the rest
   of the computer industry.


   In the late 1970's, Cray Research, the company he founded
   in 1972 after leaving Control Data, became the world's
   leading maker of supercomputers, which were increasingly
   viewed as a measure of national technological prowess and
   economic competitiveness. That led many governments to
   invest in supercomputing technologies as a way of creating
   or protecting a national advantage, whether military or
   commercial.


   Mr. Cray left Cray Research in 1989 to form the Cray
   Computer Corporation, which ended in bankruptcy in 1995. In
   August, Mr. Cray announced that he had formed a new
   company, SRC Computers, where he was planning to begin the
   design of the Cray 5 supercomputer.


   Perhaps Mr. Cray's most significant contribution was an
   approach to solving the daunting scientific and engineering
   problems known as "vector processing," which involved
   chaining together long series of calculations in
   specialized hardware to expedite solutions.


   He was also widely seen as a genius at the art of the dense
   packaging of the components that make up a computer, a
   design approach that slashes the time it takes electrical
   signals to travel between circuits.


   Mr. Cray was perhaps the most remarkable of an elite group
   of computer designers who were able to build elegant
   systems by assembling simple components in clever ways.


   The Control Data 6600 was the first computer to be cooled
   with Freon, the same fluid used in air-conditioners. To
   bring its components closer together, the Cray 1 was folded
   in on itself like a giant "C." The Cray 2 supercomputer
   used an even more exotic fluid, Flourinert, in which its
   circuit cards were immersed to draw away excess heat.


   The son of a civil engineer, Mr. Cray was born on Sept. 28,
   1925, in Chippewa Falls, Wis., where he attended school,
   displaying an early fascination with radio, electric motors
   and electrical circuits.


   After graduating from high school in 1943, Mr. Cray went
   into the Army, joining an infantry communications platoon.
   Arriving in Europe after D-Day, he participated in the
   Battle of the Bulge and then "tramped" across Europe to
   meet the Russians. He then went to the Philippines, where
   he was involved in supporting the Filipino guerrilla army.


   Mr. Cray experimented with electronics in college but did
   not learn about digital computers until after graduation
   from the University of Minnesota in 1951, having received
   a bachelor's in electrical engineering and a master's in
   mathematics.


   "I was fortunate in having an instructor at the University
   of Minnesota who was looking after me in the sense that
   when I said, 'What's next?' he said, 'If I were you, I'd
   just go down the street here to Engineering Research
   Associates, and I'd think you'd like what they're doing
   there," he said in a Smithsonian interview. The Minneapolis
   company was doing contract work for the Navy in
   cryptography.


   Mr. Cray said that while he was at Engineering Research
   Associates, he met John Von Neumann, the mathematician who
   is the father of the modern computer.


   Located in a converted wooden glider factory, Engineering
   Research Associates was in the forefront of developing
   digital computers. During his years at E.R.A., Mr. Cray
   began to develop his solitary working style and his
   philosophy of simplicity in computer design.


   Indeed, while credit is given to I.B.M. for inventing
   Reduced Instruction Set Computing, or RISC, during the
   1970's -- a now popular design approach that calls for
   simplifying computer hardware to gain speed -- computer
   historians note that Mr. Cray's computers were always
   designed along RISC lines.


   It was also at E.R.A. that Mr. Cray's legendary impatience
   with corporate bureaucracies and management began to take
   shape. When a series of mergers brought the company first
   under the control of Remington Rand, the typewriter
   company, and later Sperry Rand, Mr. Cray decided to leave
   to join William C. Norris, E.R.A.'s founder, who had
   started a new company, Control Data.


   There Mr. Cray was freed to pursue his vision of building
   large scientific computers. Shunning committees, he felt
   that the best computers were the ones where a single
   architect offered a unified vision. After the machine had
   been delivered, it was then appropriate, Mr. Cray felt, to
   listen to feedback from customers and, if necessary, start
   over from "a clean sheet of paper."


   At Control Data, Mr. Cray led the design of the CDC 1604,
   the first commercial computer to replace large vacuum tubes
   with individual transistors that he had bought at a local
   electronics store. In the scientific computing market, the
   1604 competed against machines from the more powerful
   I.B.M.


   The new company grew rapidly as large corporations began to
   adopt mainframe computing. However, Mr. Cray's disdain for
   management had followed him to Control Data which soon
   became too big for his comfort. He persuaded Mr. Norris to
   permit him to relocate his laboratory 100 miles from
   Minneapolis, back to his hometown of Chippewa Falls.


   Indeed, Mr. Cray had so little patience for traditional
   corporate activities, Mr. Norris once recalled that when he
   asked Mr. Cray to write a five-year plan for C.D.C., his
   response was: "Five-year goal: Build the biggest computer
   in the world. One-year goal: Achieve one-fifth of the
   above."


   The move to Chippewa Falls turned out to be a remarkably
   important one for Control Data. Mr. Cray's small team
   designed the CDC 6600. Introduced in August 1963, it had a
   speed of three million instructions per second,
   dramatically faster than the market leader at the time,
   I.B.M.'s 7094 computer.


   Frustrated, I.B.M.'s chairman Thomas J. Watson Jr., wrote
   a memo to his staff noting that the 6600 team totaled only
   34 people, "including the janitor," and asked how I.B.M.
   had let such a small team offer the world's most powerful
   computer.


   Five years later, the CDC 7600 furthered Control Data's
   lead in the scientific computing market. However, in 1972
   Mr. Norris delayed Mr. Cray's next project, the 8600, in
   favor of a competing design.


   Mr. Cray left Control Data with a small team and founded
   Cray Research. Four years later, the Cray 1 unseated
   Control Data as the world's fastest computer.


   Surviving with venture capital investments of about $8
   million during the development period, Mr. Cray took the
   company public in 1976, raising $10 million only a month
   before the company sold its first Cray I computer to the
   Los Alamos National Laboratory for $8.8 million.


   In 1975, while he was developing the Cray 1, Mr. Cray was
   divorced from his first wife, Verene. A year later, he met
   Geri M. Harrand, whom he later married. He began to do more
   things outside his work regimen. He learned to ski and
   began windsurfing and traveling widely.


   As Cray Research grew rapidly, Mr. Cray continued to focus
   on the technical challenges of building a supercomputer,
   increasingly turning over the responsibility of running the
   company to Mr. Rollwagen, who became chief executive in
   1980 and chairman in 1981.


   During his career, Mr. Cray also had both technical and
   business disappointments. For example, the Cray 2 was
   initially supposed to use an ultrafast semiconductor
   material gallium arsenide, but gallium arsenide proved a
   vexing material for chips, and Mr. Cray ultimately returned
   to silicon chips and finished the machine in 1985.


   Although it provided a tenfold performance increase above
   the Cray 1, it was late to market. To keep the company
   alive while Mr. Cray attempted to complete the Cray 2, Mr.
   Rollwagen had turned to a young computer designer, Steve S.
   Chen, to design a faster version of the Cray 1. That
   machine, the Cray X-MP, succeeded better than anyone had
   expected and saved the company.


   That began a rivalry between Mr. Cray and Mr. Chen that was
   to last until Mr. Rollwagen was forced to choose between
   the two designers and went with Mr. Chen. Mr. Cray left
   Cray Research in 1989 to found Cray Computer Corporation,
   basing it in Colorado Springs.


   After finishing the design of the Cray 2, Mr. Cray had
   begun working on the Cray 3, still focusing his attention
   on the possibilities of gallium arsenide processing chips.
   The project required special robots to assemble the
   processing units, and costs soared. By 1989, the Cray 3
   development project had already cost $120 million with no
   working computer on the horizon. Mr. Rollwagen was forced
   to spin off the Cray 3 project, keeping a 10 percent
   investment in the new company.


   Three years later, Cray Computer had consumed more than
   $300 million, and Mr. Cray had fallen victim to the rising
   power of cheap microprocessor chips. More than a dozen,
   start-up companies had emerged to cobble together massively
   parallel supercomputers, using the new chips and building
   systems at a far lower cost than Mr. Cray's ambitious
   gallium arsenide-based systems.


   Mr. Cray began work on the follow-on to the Cray 3, the
   Cray 4, a machine that would have 64 processors.


   However, he was never able to complete the Cray 4. In a
   bitter final chapter signifying the decline of the
   supercomputer industry, Cray Computer sought Chapter 11
   bankruptcy protection in March 1995 after failing to raise
   an additional $20 million.


   Mr. Cray's venture was undermined by the twin
   transformation of the computing world in the 1990's. First,
   the arrival of cheap and powerful microprocessor chips had
   dramatically undercut the multimilliondollar "big iron"
   systems that were Mr. Cray's hallmark. Moreover, the end of
   the cold war meant declining Government budgets for
   purchasing machines that were once the mainstay of the
   nation's weapons labs and at the heart of the Strategic
   Defense Initiative, called "Star Wars."


   In a letter to his employees, he pointed to larger changes
   in the world that had prevented his company from selling
   even one system in its six years of existence.


   "Our problem is basically one of timing," he wrote. "The
   business world and our government are in a cost-cutting
   mode."


   In addition to his wife, Mr. Cray is survived by two
   daughters, Susan Borman of Eau Claire, Wis., and Carolyn
   Arnold of Minneapolis; a, son, Steven, of Chippewa Falls,
   a sister, Carol Kersten of Rochester Minn., and five
   grandchildren.


   [Photo] Seymour R. Cray, with the Cray 2 supercomputer he
   developed.


   [End]