SRC Computers

Seymour Cray
Born	Seymour Roger Cray (1925-09-28) September 28, 1925 Chippewa Falls, Wisconsin, U.S.
Died	October 5, 1996(1996-10-05) (aged 71)
Occupation(s)	Electrical engineer, supercomputer designer
Years active	1950–1996
Known for	Founder of Cray Research, pioneer of supercomputing

Seymour Roger Cray (September 28, 1925 – October 5, 1996) was a pioneering American electrical engineer and supercomputer designer who developed a succession of the fastest computers globally and established Cray Research. Often hailed as "the father of supercomputing," Cray is recognized for founding the supercomputer industry. Joel S. Birnbaum, former chief technology officer at Hewlett-Packard, remarked that numerous high-performance computing features now commonplace were once considered cutting-edge innovations envisioned by Cray. Larry Smarr, who directed the National Center for Supercomputing Applications at the time, described him as "the Thomas Edison of the supercomputing field."

Early life

Cray was born in 1925 in Chippewa Falls, Wisconsin to Seymour R. Cray, a civil engineer. From a young age, he showed a fascination with science and engineering. At just ten years old, he constructed a machine that translated punched paper tape into Morse code signals using an Erector Set. His family’s basement served as his experimental workshop.

After finishing at Chippewa Falls High School in 1943, Cray was drafted during World War II as a radio operator, serving in Europe and later contributing to codebreaking efforts against Japanese naval communications in the Pacific. Upon returning to the United States, he earned a Bachelor of Science degree in electrical engineering (1949) followed by a Master of Science in applied mathematics (1951), both from the University of Minnesota.

Engineering Research Associates

In 1950, Cray began working at Engineering Research Associates (ERA) located in Saint Paul, Minnesota. ERA originated from a former United States Navy laboratory focused on cryptographic machines. The company expanded into digital computing technology and notably produced the ERA 1103, recognized as the first commercially successful scientific computer. Cray remained with ERA through its acquisitions by Remington Rand and later Sperry Corporation, where ERA became the scientific computing division within their UNIVAC business unit.

Control Data Corporation

Unhappy with ERA’s direction under Sperry Rand management, Cray partnered with William Norris to establish Control Data Corporation in 1957. By 1960, Cray had completed designing the CDC 1604, an enhanced and cost-effective iteration of the ERA 1103 that delivered strong performance relative to its price. While CDC prioritized business-oriented systems such as the CDC 3000 series, Cray concentrated on developing high-performance machines, culminating in the revolutionary CDC 6600.

The CDC 6600 earned distinction as the first commercially available supercomputer, surpassing competitors significantly despite its premium price. Cray focused on maximizing overall system speed by balancing processor capabilities with input/output (I/O) throughput to avoid data congestion. He introduced a novel approach by eliminating I/O interrupts in favor of polled requests managed by ten peripheral processors—specialized mini-computers responsible for all memory transfers—addressing critical design obstacles that impeded rivals like IBM’s IBM 7030 Stretch. The follow-up model, the CDC 7600, boosted speed by a factor of five.

In a 1963 Business Week feature on the CDC 6600, Cray explained a performance-to-cost concept frequently mistaken for Grosch's law:

CDC's Chippewa Falls Facility

Throughout this timeframe, Cray became increasingly exasperated with the interference from CDC's management. He insisted on a tranquil workspace with minimal oversight; however, as CDC grew, he was often disrupted by middle managers who, in his opinion, contributed little and mainly utilized him as a promotional figurehead.

To preserve his concentration, Cray opted to move to a location distant enough from St. Paul to deter casual drop-ins and phone calls, yet still accessible for board meetings. Norris endorsed this relocation, setting up a new laboratory on Cray’s own land in Chippewa Falls. Additional factors influencing the decision included concerns about nuclear conflict and safety issues in the Twin Cities area. Cray’s nearby residence featured an extensive fallout shelter.

The Chippewa Falls lab was launched midway through the 6600 project without hindering its progress. Following shipment of the 6600, the CDC 7600 was developed at this site, achieving a tenfold increase in speed. Nevertheless, the subsequent CDC 8600 initiative faced significant challenges, bringing Cray’s successful tenure at CDC to an end by 1972.

Although both the 6600 and 7600 were commercially successful, their development nearly pushed CDC to financial collapse. The problematic 8600 project led Cray to suggest starting anew, but Norris favored concentrating on the internally advancing CDC STAR-100. Unwilling to continue under these conditions, Cray departed from CDC.

Cray Research

The separation was conducted on friendly terms; within a year, Cray established Cray Research on his Chippewa property with Norris contributing $250,000 in investment. Research and development stayed in Chippewa Falls while corporate headquarters were located in Minneapolis; unlike CDC, manufacturing operations were also local.

At first uncertain about its strategic direction and funding potential, Cray Research quickly gained strong investor trust due to Seymour Cray’s reputation. This confidence allowed full financing for new computer development.

After several years of effort, Cray Research introduced its inaugural product in 1976: the Cray-1. Staying true to Cray’s design principles, it maximized overall system performance and outpaced nearly all rivals including the STAR-100. Only the specialized ILLIAC IV approached its speed but lacked comparable flexibility.

Unit number 001 was loaned to Los Alamos National Laboratory in 1976; that same summer saw the first complete system sold to the National Center for Atmospheric Research for $8.8 million. While initial sales projections based on CDC-era machines anticipated about a dozen units sold, over 80 were eventually purchased. The company achieved significant financial success and earned Seymour Cray the moniker "The Wizard of Chippewa Falls."

Later achievements were variable. Although Cray developed the Cray-2, other teams released the dual-processor Cray X-MP and subsequently a four-processor variant—both commercially successful. The Cray-2 debuted after six years but delivered only slightly higher speeds than the X-MP due to its sophisticated memory system and complex pipeline design, leading to modest sales figures.

When launching the Cray-3 project, Cray faced renewed distractions from day-to-day management responsibilities and resigned as CEO in 1980 to serve as an independent contractor. In 1988, he moved the Cray-3 development effort to Colorado Springs.

By 1989, problems with the Cray-3 resurfaced amid concurrent development of an enhanced X-MP model utilizing Cray-2 memory technology. Management decided to prioritize this lower-risk option, releasing it as the Cray Y-MP.

Cray Computer Corporation

The Colorado Springs research facility was separated from Cray's main operations to establish Cray Computer Corporation, which assumed responsibility for the Cray-3 initiative.

The 500 MHz Cray-3 represented Cray's second significant setback. In pursuit of a tenfold boost in performance, Cray opted to utilize gallium arsenide semiconductors, diverging from his customary dependence on established technologies. This decision necessitated the creation of every element from scratch, including the semiconductor chips themselves.

Despite numerous obstacles, the team successfully delivered the inaugural Cray-3 unit to NCAR on 24 May 1993. The system functioned largely as a prototype, primarily serving debugging purposes. By that time, multiple massively parallel computing platforms had entered the marketplace, offering superior price-to-performance ratios compared to the Cray-3. In response, Cray began developing the Cray-4, aiming for 1 GHz speeds and enhanced power irrespective of cost considerations.

Nevertheless, no further Cray-3 units were sold by 1995, and the reduced demand for costly supercomputers like the Cray-4 in the post-Cold War era exacerbated financial difficulties. Consequently, Cray Computer Corporation filed for Chapter 11 bankruptcy protection on 24 March 1995.

SRC Computers

Initially skeptical about massively parallel architectures, Cray maintained that a single high-speed processor outperformed numerous slower ones. However, advancements in compiler technology and shifting market needs by the mid-1990s challenged this perspective. To address these changes, Cray founded SRC Computers, focusing on creating a massively parallel system that tackled communication and memory bottlenecks typical of such designs. Following Cray's untimely death in a car accident, SRC Computers continued its work with an emphasis on reconfigurable computing.

Technical approaches

Two fundamental design tenets guided Cray’s work: efficient thermal management and precise signal timing. His machines incorporated sophisticated cooling mechanisms, including integrated coolant channels within both chassis and circuit boards; some designs even immersed components entirely in coolant fluids. Safety interlocks ensured operation only when cooling was active, thus preventing hardware damage.

To mitigate clock skew, Cray carefully equalized signal path lengths on circuit boards, occasionally routing traces in complex patterns to maintain consistent electrical lengths. He applied Maxwell's equations to address radio frequency phenomena affecting signal velocity and timing accuracy.

Despite these complexities, Cray favored straightforward design tools, often beginning layouts on lightly ruled quadrille graph paper pads.

List of computers

Cray played a role in designing the following machines:

• ERA 1103 (1953)
• AN/USQ-17 (1958, unproduced)
• CDC 1604 (1960)
• CDC 3000 series (1963)
• CDC 6600 (1964)
• CDC 7600 (1967)
• CDC 8600 (cancelled 1974)
• Cray-1 (1976)
• Cray-2 (1985)
• Cray-3 (single unit built, 1993)
• Cray-4 (cancelled)
• Cray-3/SSS (cancelled)

Personal life

In 1947, Cray wed Verene Voll, a childhood acquaintance who was a nutritionist and daughter of a Methodist minister—paralleling Cray’s own mother’s background. The couple had three children before divorcing around 1978. Subsequently, he married Geri M. Harrand. Among his descendants is LGBTQ rights advocate Andrew Cray, his grandson.

Known for shunning publicity, Cray enjoyed activities such as skiing, windsurfing, tennis, and constructing an underground tunnel beneath his residence—a hobby he credited with inspiring engineering problem-solving. Stories shared by then-CEO John A. Rollwagen illustrate Cray’s unconventional management style, characterized by brief progress updates and ambitious objectives like creating "the biggest computer in the world."

Cray passed away on October 5, 1996, at age 71, two weeks after being involved in a rollover accident caused by a reckless driver while merging onto Interstate 25 near Colorado’s United States Air Force Academy.

Posthumous Recognition

Since 1997, the IEEE Computer Society has presented the Seymour Cray Computer Engineering Award to acknowledge groundbreaking achievements in the realm of high-performance computing systems, embodying the innovative legacy of Seymour Cray.

See also

• Charles Babbage Institute
• Cray-3/SSS
• John Vincent Atanasoff

External links

• Seymour Cray Oral History
• What's All This About Gallium Arsenide? — keynote address delivered by Seymour Cray on November 15, 1988, at Supercomputing '88
• An Imaginary Tour of a Biological Computer: Presentation by Seymour Cray at the Shannon Center for Advanced Studies, University of Virginia, May 30, 1996