First Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory

By combining a new design that permitted vector processing with an earlier innovation of an instruction pipeline, Seymour Cray and his team of engineers at Cray Research built a new supercomputer with unprecedented processing power.

Summary of Event

The computer firm Engineering Research Associates (ERA) of St. Paul, Minnesota, was founded in 1946 with the idea of continuing the work of a team of naval cryptographers who had worked on deciphering the Nazi Enigma code during World War II. The firm established a pattern of encouraging creativity and enthusiasm in their engineers. In 1957, William Norris, the company’s founder, and others left to form Control Data Corporation Control Data Corporation (CDC), a small company also based in St. Paul. The next year, Seymour Cray, a talented young engineer from ERA, joined CDC. With a solid understanding of logic and computational mathematics as well as electrical engineering, Cray was able to design a fast, efficient, and relatively inexpensive computer using transistors instead of the bulky, unreliable vacuum tubes which were standard at the time. During the early 1960’s, this computer, the CDC 1604, became popular in the scientific, military, and corporate communities. In spite of the firm’s relatively tiny size and modest funding, CDC began to displace larger companies in the technical computing field. Computers;Cray-1[Cray one]
Cray-1 supercomputer[Cray one supercomputer]
[kw]First Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory (Mar. 4, 1976)
[kw]Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory, First (Mar. 4, 1976)
[kw]Supercomputer Is Shipped to the Los Alamos National Laboratory, First Cray-1 (Mar. 4, 1976)
[kw]Los Alamos National Laboratory, First Cray-1 Supercomputer Is Shipped to the (Mar. 4, 1976)
Computers;Cray-1[Cray one]
Cray-1 supercomputer[Cray one supercomputer]
[g]North America;Mar. 4, 1976: First Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory[02330]
[g]United States;Mar. 4, 1976: First Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory[02330]
[c]Computers and computer science;Mar. 4, 1976: First Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory[02330]
[c]Inventions;Mar. 4, 1976: First Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory[02330]
[c]Science and technology;Mar. 4, 1976: First Cray-1 Supercomputer Is Shipped to the Los Alamos National Laboratory[02330]
Cray, Seymour
Norris, William
Davis, Les
Chen, Steve
Hillis, W. Daniel

CDC’s lead was cemented in 1964 with Cray’s next computer, the CDC 6600. Much more powerful than its predecessor, the CDC 6600 achieved a processing power of three million FLOPS (floating-point operations per second). The term “supercomputer” came into use as a way of distinguishing these ultrafast and very powerful machines from those used in business. Next, the CDC 7600 was released. Five times faster than the 6600, the computer utilized an instruction pipeline that allowed incoming data to be fed in while previous instructions were being executed. However, there were some problems with these new computers. It was difficult to cool the machines, which were made up of circuitry and wires that were densely packed together for maximum speed. Also, the 7600 code was not compatible with the code of the earlier CDC machines, and they broke down frequently.

In 1972, Cray, with other key engineers such as Les Davis, who was effective at organizing team projects, left CDC and formed Cray Research, Inc., based in Cray’s hometown of Chippewa Falls, Wisconsin, where he had already moved the CDC development facilities. Because of his fame, Cray was able to attract funding and began work on his next design, which combined the innovations of his CDC computers with yet another new concept, vector processing. In contrast to scalar processing, in which numbers are processed in sequence, vector processing increases speed by working with long arrays (lists) of numbers simultaneously. This radical new design was combined with further increases in speed, so that scalar processing was also dramatically improved. Meanwhile, CDC continued to develop and refine new models, essentially finishing some projects that Cray had started. Cray found himself competing with his own former colleagues for lucrative sales.

The demands made on supercomputers during the Cold War Cold War continued to expand. The leading customers for supercomputers were the Los Alamos National Laboratory Los Alamos National Laboratory in New Mexico, the Lawrence Livermore National Laboratory Lawrence Livermore National Laboratory in California, and the National Center for Atmospheric Research National Center for Atmospheric Research (NCAR) in Colorado. Among other projects, the Los Alamos and Livermore facilities needed to simulate the complex and rapid changes that occurred during nuclear reactions, and NCAR needed to model statistics for studying weather and climate and for predicting severe storms. Scientists at the client institutions participated in the development of supercomputers by writing operating systems and programs to run on them and by consulting with CDC, International Business Machines (IBM), and other computer manufacturers during development and testing.

Cray’s new vector-processing computer was called the Cray-1. It used one million 64-bit-wide units of rapid memory for math processing. Its clock speed was measured at 80 megahertz. After the Cray-1 was announced in 1975, the Los Alamos and Livermore labs, which often competed with each other for research funding, struggled to be the first to own one of the new machines. The rivalry between the two labs actually delayed the purchasing process. To resolve the situation, on March 4, 1976, Cray Research shipped its first Cray-1 to the Los Alamos National Laboratory for a free trial. The following year, NCAR became the first customer to order one of the machines. Each sale was worth almost $9 million, and Cray Research quickly became profitable and hired more employees.

The next innovation in supercomputers was multiprocessing, the use of two or more central processing units (CPUs) in one computer system. Conceived by Davis, the Cray X-MP combined two Cray-1’s to achieve three times the processing power of a single unit. Davis selected Dr. Steve Chen, who had recently completed his dissertation on parallel processing, to be the lead designer. The X-MP was succeeded in 1985 by the Cray-2, which used four processors and was the earliest computer machine to go beyond one billion FLOPS. Although the Cray-2 was not as profitable as the Cray-1, many scientists felt that parallel processing was the wave of the future.

A Cray-1 Supercomputer at the Computer History Museum in Mountain View, California.

(Ed Toton)

Fortunately, the individual processors needed in parallel processing became less expensive and more powerful over time, as the logic circuits that had once been laboriously handcrafted by Cray and his team became miniaturized, printed onto circuit boards, and mass-produced for the growing personal computer industry. W. Daniel Hillis, a doctoral student at the Massachusetts Institute of Technology, was interested in “massive parallelism,” in which thousands of processors are linked to form one computer. His Thinking Machines Corporation, Thinking Machines Corporation founded in 1983, produced several variations on his Connection Machine, a massively parallel supercomputer based on biological models, with the thousands of processors designed to function as individual neurons in a single brain. Although Thinking Machines led the supercomputer industry in the early 1990’s, the end of the Cold War resulted in decreased government funding for supercomputers, and the company filed for bankruptcy in 1994. After this period, the development of supercomputer hardware was largely taken over by large companies such as IBM and Sun Microsystems, which purchased the hardware of Thinking Machines Corporation.

Related to the idea of parallel computing, but much less expensive, is the concept of “grid” or “distributed” computing, in which individual workstations are networked to divide tasks. Although grid computing is usually slower than multiple processors in a single machine, it has advantages, such as eliminating the need to cool many closely packed processors.


The development of supercomputers depended to a great extent on the creativity and skills of scientists and engineers who went to great lengths to stay focused on their goals, enabling them to challenge much larger and more established organizations. When the practice of building processing hardware from scratch became too expensive and specialized, the same spirit of creativity and independence entered into the development of new possibilities. Scientists and engineers used the achievements of the supercomputer pioneers as building blocks to create the infrastructure of the Internet, to explore the frontiers of artificial intelligence, and more.

Because of almost constant advances in processing power and speed, the term “supercomputer” has been defined only loosely by the most powerful machine at any given time. Just a few decades after the Cray-1 was introduced, some of its features were matched or even exceeded by high performance machines made available to millions of consumers by Apple and other popular computer manufacturers. Nevertheless, the capabilities of the supercomputers, which were able to simulate nuclear explosions and therefore avoid the terrible destruction of actual detonations, acted as a deterrent to the proliferation of weapons during the Cold War. The achievements of Seymour Cray and the other engineers who developed these machines led to advances in other areas of science as well. Ultimately, the flexibility and creativity they showed in being able to switch quickly from vacuum tubes to transistors to printed circuits to multiple processors will provide a useful model for future changes in fundamental technology. Computers;Cray-1[Cray one]
Cray-1 supercomputer[Cray one supercomputer]

Further Reading

  • Fritz, Sandy, et al., eds. Understanding Supercomputing: From the Editors of “Scientific American.” New York: Warner Books, 2002. Futuristic exploration of trends in hardware, with mention of Seymour Cray’s use of gallium arsenide for the Cray-3. Illustrated, index.
  • Moschovitis, Christos J. P., ed. History of the Internet: A Chronology, 1843 to the Present. Santa Barbara, Calif.: ABC-CLIO, 1999. Includes description of the development of the Internet as a means of information sharing by supercomputer users in the scientific community. Photos, glossary, bibliography, index.
  • Murray, Charles J. The Supermen: The Story of Seymour Cray and the Technical Wizards Behind the Supercomputer. New York: John Wiley & Sons, 1997. Definitive, not overly technical narrative describes the main inventors, the evolution of their creative work environment, the constantly shifting business arrangements, the corporate sponsors, and the high-powered customers. Photos, notes, index.

Intel Introduces the First “Computer on a Chip”

Jobs and Wozniak Found Apple Computer

IBM Introduces Its Personal Computer

Rise of the Internet and the World Wide Web

Sun Microsystems Introduces Java