📸 Seymour Cray posing alongside the Cray-1. (Source: Minnesota Star Tribune)
📸 A CRAY-1 AT THE MAGNETIC FUSION ENERGY COMPUTER CENTER (Source: Department of Energy)
Advances in computers are always a matter of miniaturization. Data that once took a car-sized computer to process can now be done on a system a fraction of the size in a fraction of the time. New machines naturally eclipse their predecessors, but there are still certain milestones that mark the way from the earliest supercomputers to the machines of today. When it was finished in 1975, the Cray-1 was the fastest supercomputer in the world, its sixty miles of coiled wire processing data ten times more efficiently than its closest competitor.
Fitting for such a revolutionary machine, Cray-1’s namesake was a legend in the world of early digital computers. A code-breaking veteran of World War Two, Seymour Cray worked for several prestigious computer firms, eventually founding Control Data Corporation in 1957, located in Bloomington, Minnesota. Dissatisfied with management interference, Cray split again to form Cray Research. Eschewing the methods of the past, this new company would be based on four main principles: simplicity, size, discipline, and cooling.
📸 A Cray-1 with its 60 miles of wires exposed (Source: Computer History Museum)
Part of the split from Control Data Corporation had been necessitated by Cray’s demand for absolute privacy and silence while working on his designs. Cray’s eccentricities were seemingly boundless. His fear of a nuclear war had led him to build his CDC lab in his hometown, Chippewa Falls, Wisconsin. Later in his career, he worked out of a cabin. In his spare time, Cray dug a massive tunnel under his house, complete with a periscope, and joked that elves in the tunnel would help him solve whatever computer problem he was faced with. As Cray himself once said: "I'm up in the Wisconsin woods, and there are elves in the woods. So when they see me leave, they come into my office and solve all the problems I'm having. Then I go back up and work some more."
These eccentricities helped to build a mystique around Cray, something of great use to him. The man was able to inspire confidence and devotion, with many computer scientists following him from CDC to his new company. From there they set their eyes on a simple enough task: making the fastest computer in the world and finding new ways to do it. Earlier attempts to create a viable supercomputer involved the use of incredibly complex integrated circuits. The Cray-1 used just three different types of integrated circuits, vastly simplifying the architecture.
"My guiding principle was simplicity. I think there is an expression for that. Don't put anything in that isn't necessary. Whereas many other places at that point in time and for several years after that were adding all the bells and whistles that could be imagined. Later on much more recently there came the term "RISC" which says "back to the basics", make it as simple as you can. I thought I was a RISC person all the time even though I didn't know the name." -Seymour Cray, 1995 interview
📸 CRAY-1 CABINET DESIGN FROM SEYMOUR CRAY'S 1978 PATENT.
Cray’s use of integrated circuits for the new machine was curious. They were hardly a new invention, being cheaply available as early as 1966, but this was a core idea of Cray’s design philosophy. He always made a point to lag behind innovations, preferring others to try and fail for him. In spite of using older technology when he could, demands on the system were still extreme. For cooling, freon circulated through stainless steel tubing bonded between vertical wedges of aluminum fitted between the stacks of circuit boards. The machines were so hot that years later they were used to heat a Cray facility during a cold Minnesota winter.
The end result was a supercomputer faster than any system built before it. The Cray-1 was so advanced that a bidding war ensued between the Department of Energy and Lawrence Livermore National Laboratory for the first machine off the line, called Serial One. This made the Cray-1 the first commercially successful supercomputer and launched the legend that became Cray Research. The system revitalized a flagging supercomputer industry, finding a home in places like the NSA and other government agencies where it was used as a tool during the Cold War.
📸 The finished Cray-1.
The iconic look of the Cray-1 is more than just 1970's aesthetics at play. Everything was thought through to provide advantages in performance. The columnar design of the cabinet allowed Cray to minimize the amount of wiring between processing stacks, while the cushions ringing the unit covered the enormous power supplies at the base of each tower. Its successor the Cray-2 added its famous “waterfall” cooling system.
Years later in 1988, Cray set his sights on the Cray-3, the next evolution of his computing system. This machine would make use of circuits made from gallium arsenide, which would in theory allow the machine to run more efficiently and thus more cooly, but it had had limited success in supercomputing. Fearing a costly failure, Cray Research’s managers made a proposal: Cray could continue his work, but as a separate entity, Cray Computer Corporation. For the second time, Cray broke from a company he had founded to chase his dream.
Although his gallium arsenide gambit paid off, the end of the Cold War dried up demand for supercomputers, and the CCC shuttered its doors. Cray was in the process of getting a new company off the ground when he died after a car accident in 1996, age 71, depriving the world of many new innovations he doubtless would have made. Although his story ends on a tragic note, the machines Cray built revolutionized an industry, the common ancestor of all supercomputers today.
First Supercomputer Cray-1
First Supercomputer Cray-1
Further Reading
Cray, Seymour. Live presentation at Los Alamos National Lab (LANL), 1976. https://www.youtube.com/watch?v=vtOA1vuoDgQ
Murray, Charles J., and Arthur L. Norberg. The Supermen: The Story of Seymour Cray and the Technical Wizards Behind the Supercomputer. Wiley, 1997.
Igarashi, Yoshihide, et al. Computing: A Historical and Technical Perspective. CRC Press, 2014.