The following article is from the Computer museum History Center's "CORE" vol. 1.4
(Most images are not included)
by ALEXIS DANIELS
Gene Amdahl's WISC is currently on display at the The Computer Museum History Center's Visible Storage Exhibit Area.
Recently, over several hours of videotaped interviews conducted by William Aspray, Executive Director of Computing Research Association, Gene Amdahl reflected on his professional experiences and documented the cours of his amazing technical life. The following material condenses some of the story that was gathered.
Gene Myron Amdahl was born in Flandreau, South Dakota in 1922. Even though his father had only three years of schooling, the elder Amdahl knew the importance of education. When Gene declared his intention to go to South Dakota State to study engineering, his father encouraged him to get a liberal arts education instead, emphasizing that learning how to make a living was not as important as learning how to get the most out of life. Nevertheless, Amdahl went on to South Dakota State and accomplished both goals. Although he began as an average student, his performance changed dramatically when he took a physics course during the summer of his freshman year. He became consumed by a passion that not only altered the course of his life, but which later had a profound impact on the entire computer industry.
Although his undergraduate work was interrupted during World War II by a twoyear stint in the U.S. Navy, Amdahl returned to South Dakota State and received his bachelor's degree in engineering physics in 1948. He then began his graduate work at the University of Wisconsin with a thesis on "The Contributions to the Magnetic Moments of Heavy Nuclei Due to Spin Anti-Symmetry and Velocity-Dependent Forces."
Meanwhile, he began designing computers on his own time. When the Electrical Engineering department heard about this "other" work, Amdahl was encouraged to build a computer that could be used to train graduate students in the emerging field of digital computing. The resulting computer, known as the Wisconsin Integrally Synchronized Computer (WISC), was designed in the summer of 1950, and submitted as Amdahl's doctoral thesis in June 1951. His ideas were so innovative that the Physics Department felt unqualified to evaluate it and sent it to others for review and acceptance. His thesis passed the test, and Amdahl received his doctorate in theoretical physics in 1952.
After graduate school, Amdahl wanted to start a company building computers but he lacked sufficient financing. He interviewed with International Business Machines (IBM) and was hired, in part, because IBM was impressed with the quality of the writing in his doctoral thesis. Rather than the dry, technical style of most theses, Amdahl's writing had a missionary's zeal that engaged his readers. He accepted a position with IBM in 1952 and was the most highlypaid person in the history of IBM to be hired directly out of school.
In the fifties, the environment at IBM was one of innovation and excitement when new technologies emerging from the war effort were beginning to be applied in industry. Amdahl initially worked on machine designs for character recognition and simulation studies to determine if a machine could be made to behave like a human brain. He was the chief architect for the IBM 704 computer, IBM's first commercial machine with floating-point hardware and the first widely-used machine to use indexing and a high-level programming language (FORTRAN). While the marketing department at IBM predicted a market of only six machines, Amdahl himself predicted a market for 32 machines, and the price of the 704 was based on that projection. Since 140 machines were sold, the 704 proved to be highly profitable to IBM and secured Amdahl's place within IBM as a bold, innovative thinker and manager.
In 1955, Amdahl, John Backus and others at IBM began work on the 7030 project, also known as "STRETCH." The goal of the STRETCH project was to build a supercomputer for the Los Alamos National Laboratory with 100 times the performance of anything else available at that time and to "stretch" IBM internally in terms of design, manufacturing, and device technologies. Frustrated with management's directions, Amdahl left IBM in 1956. He worked for other computer companies on a variety of projects that included designing airborne computers for fighter planes to maximize the plane's capabilities in a dogfight, as well as creating a data entry system for FAA flight planning. Back at IBM, the first of nine STRETCH computers was delivered in 1959 and, although each was sold at a loss, the intellectual debt IBM's later System 7000 and System/360 family of computers owed to STRETCH was to be enormous.
Despite his earlier disenchantment with IBM, Amdahl agreed to return to the company in 1960. He was named Manager of Architecture for the IBM System/360 family of mainframe computers. The System/360, announced in April of 1964, was a series of instruction-set compatible machines covering a 400:1 performance range. It became the greatest success story in the history of computing and IBM's most profitable product line ever-in fact, the basic System/360 architecture is still embedded in many current IBM products today.
By 1969, Amdahl had been named an IBM Fellow, that company's highest honor, and was made director of IBM's Advanced Computing Systems (ACS) Laboratory in Menlo Park, California. After a time, Amdahl again became disenchanted with IBM's bureaucracy and the internal barriers he felt were hampering the company's growth and ACS product development. Even though many company executives believed his ideas had merit, they refused to change direction, and so, once again, Amdahl left IBM.
When Amdahl resigned from IBM for the second time, he decided to pursue the dream he had held since completing graduate school: to start a company that would build computers. In order to circumvent future legal problems, he fully disclosed his plans to senior management at IBM who cautioned him that there was no money to be made in large computers.
In 1970, Amdahl Corporation was formed in Sunnyvale, California, with the mission to build more innovative mainframe computers (called PCMs--Plug Compatible Mainframes) and to compete head-to-head with IBM. Most industry analysts thought Amdahl was foolish to take on IBM and he experienced problems raising the capital he needed. Despite the difficulties, Amdahl was able to simplify design, improve technology, and build discounted computers that could be substituted for the more costly IBM models. The company's first computer, the Amdahl 470 V/6, shipped in 1975 and sold briskly, being a direct, drop-in replacement for IBM's System 360/165 but one-quarter the size and four times as fast (the price was the same at $3.5 million).
Although IBM had not originally considered Amdahl Corporation as a potential competitor, the company soon learned that it had underestimated its former employee's determination. At its peak, Amdahl Corporation captured 22°/0 of the large systems market and had a pre-tax profit of 30%. Amdahl Corporation became the biggest threat to IBM's domination of the mainframe market and forced IBM to re-align its marketing strategies to take PCM manufacturers into account.
Ever in search of new challenges, Amdahl left Amdahl Corporation in 1980 and went on to establish three other companies: Trilogy Systems (now part of Elxsi Corporation), Andor Systems, and Commercial Data Servers (CDS). In 1991, The Times of London named him one of the "1,000 Makers of the 20th Century," and Computerworld called him one of the 25 people that "changed the world."
Gene Amdahl not only followed his father's advice to learn to make a living and to get the most out of life, but he also left a lasting mark on the computer industry with his well-known law on the theory of computer architecture itself.' His innovative and pioneering spirit showed the world that it was possible to compete with IBM on its own terms. Yet perhaps most notable and memorable are his sustained records of accomplishment and energy over a lifetime. ::
Amdahl's Law states: `It x of a program is inherently sequential. the Maximum attainable speedup is 1/x. Experience has shown this law to be fundamental to computer designs which incorporate multi-threaded kernels and parallelism.