The Father of Supercomputers: The Story of Seymour Cray and the Cray Corporation.

[Man]

Imagine the early 1970s. Humanity has landed on the moon, the Cold War is in full swing, and the need for computing power is growing every year. It is at this point in the computer industry that a man named Seymour Cray appears - a brilliant engineer and true innovator who believed that only very fast and very powerful computers would help humanity accelerate technological development and make many scientific discoveries.

Who is Seymour Cray?​

Designing and even building a supercomputer is only half the battle. It is much more difficult to make it commercially successful. The man who managed to do this, Seymour Cray, was born on September 28, 1925, in the town of Chippewa Falls, Wisconsin. His father, a civil engineer, instilled in the boy an interest in technology, although young Seymour was more drawn to electronics than architecture. Having organized a real laboratory in the basement of his parents' house, Seymour constantly made various crafts from radio components. Thus, at the age of ten, he built an electrical device that converted holes punched in punched tape into Morse code signals. Perhaps this design in particular, and his passion for radio electronics in general, played an important role in his immediate fate: after graduating from school in 1943, Cray was called up to the front as a radio operator.

Seymour Cray served in Europe and the Pacific, where he helped break Japanese naval codes, among other things. Upon returning from the war, he attended the University of Minnesota, where he earned a bachelor's degree in electrical engineering and a master's degree in applied mathematics.

His career began at Engineering Research Associates (ERA), one of the first firms to build computers. There, Cray helped develop the ERA 1103, a computing machine that is considered the first truly commercially successful computer in the United States. But over time, Seymour became disillusioned with working for someone else. Why work for a big corporation when you can start your own? Together with his friend William Norris, he created Control Data Corporation (CDC), whose first product was the CDC 1604 computer, an improved and improved version of the ERA 1103.

However, some disagreements began to arise within the CDC management. Norris decided to focus on relatively inexpensive and simple computers for banks and commercial companies, while Cray dreamed of large and powerful supercomputers that could perform complex scientific calculations in a fraction of a second. Cray wanted to "build the fastest computer in the world," so he handed over the work on simple commercial computers to other engineers and began designing the CDC 6600, a flagship mainframe with a performance of up to three megaflops.

Even in the sketchy design stage, however, Cray had come to the conclusion that performance was more than just processor speed, as he quickly ran into the technical limitations of the I/O subsystem as he pushed the machine's performance. "Anyone can build a fast processor", he wrote, "the trick is to build a fast system".

Nevertheless, the CDC 6600 became a legend: sometimes called the world's first supercomputer, it outperformed all other computers then available, including the IBM 7030. It was considered the most powerful computer in the world from 1964 to 1969, when it lost that status to its successor, the CDC 7600.

But Cray was a man who always wanted more, and CDC was having financial problems that were hampering his ambitious projects. In 1972, already an established computing architect, he left CDC and founded his own company, Cray Research, Inc. The move reflected his conviction that the future of supercomputing required radically new approaches.

Cray Research Corporation​

Cray Research became Seymour Cray’s personal laboratory and a testing ground for his boldest ideas. Cray literally lived for his business, immersing himself in drawings and diagrams from morning until night. Legend has it that there was no telephone in his office, and he preferred to communicate with colleagues using notes left by the door. Isolation was part of his working method – he believed that the best solutions come when there are no distractions.

The company’s first brainchild, the Cray-1, which appeared in 1975, was not only the most powerful computer of its time, but also the embodiment of innovative thinking and futuristic design. This was not a standard angular cabinet with a bunch of wires and printed circuit boards. The computer had a cylindrical shape and was covered in bright upholstery, and at the bottom of the case there were peculiar “benches”. This design was not only a tribute to fashion: the cylindrical shape reduced the length of wires inside the machine, which also increased its productivity. And the "benches" around the base of the supercomputer played the role of a cooling system - freon tubes ran inside these seats. It can be said that back in the early 70s, Cray sought to build computers with an elegant, but at the same time functional design: a principle that Apple arrived at more than a decade later.

One of the key innovations of the Cray-1 was vector processing. This meant that instead of processing data one element at a time, as other computers of the time did, the Cray-1 could process entire arrays of data in a unit of time. The supercomputer was created using an incredibly fast and compact data transfer system between the processor and memory, and the memory itself worked several times faster than that of its competitors. The Cray-1 architecture used four different types of chips. The computer's performance was 160 megaflops, it had a 64-bit architecture, 24-bit addressing, and a maximum aggregate memory bandwidth of 638 Mbps. The first version of the supercomputer weighed 4,800 kg (including the freon cooling system) and consumed about 115 kW of power.

The first Cray-1 was sold within a month for $8.8 million, and the company built about 80 copies of this computer in various modifications. There was even a serious battle between Lawrence Livermore National Laboratory and Los Alamos National Laboratory for the first supercomputer.

In 1982, the company introduced the Cray-2, a machine that marked the birth of the second generation of supercomputers. It was a multiprocessor computer that also used a unique cooling system based on an electrically inert liquid pumped directly through the blocks with circuit boards. Despite the fact that great hopes were pinned on this supercomputer, the Cray-2 turned out to be only slightly faster than the quad-core successor of the first-generation supercomputer, the Cray X-MP.

Seymour Cray always demanded a completely calm working environment with a minimum of management tasks, but as the company grew, he was constantly distracted by managers and department heads. As a result, he could not stand such working conditions and left Cray Research in the early eighties. The company continued to develop and produce supercomputers - Cray Y-MP, Cray C90, Cray T90, but without its founder.

In the early 90s, Cray Research launched the production of computers with massive parallelism based on DEC Alpha 21064 processors - Cray T3D and Cray T3E. By February 1996, Cray actually remained the only developer of supercomputers in the United States - it was at this point that Silicon Graphics acquired it for 740 million dollars. But this stage in Cray's history did not last long: in 2000, Tera Computer Company bought the Cray Research business from SGI. And 19 years later, this organization became part of the Hewlett-Packard holding company.

Cray after Cray​

After leaving Cray Research, Seymour Cray founded a new company, Cray Computer Corporation (CCC), which began developing the Cray-3 supercomputer. Cray set a goal of achieving a tenfold increase in performance compared to his previous supercomputer, and decided that the machine should be built using gallium arsenide semiconductors. But while Cray's previous supercomputer models were partially built using components from other manufacturers, Seymour decided to develop everything himself, including even the chips.

At a time when the industry was moving toward increasing performance through the development of parallelism, Cray was categorically against this technology. He said: "If you had to plow a field, which would you choose: two strong bulls or 1024 chickens?" Cray believed that computing power should be increased through innovations in architecture, and parallelism was a road to nowhere.

By 1989, when the Cray-3 project was launched, perestroika was in full swing in the USSR, and the Cold War between the Soviet Union and the United States was essentially over. The military no longer needed super-expensive, super-fast, and super-powerful supercomputers, so the Cray-3 prototype built by 1993 remained a single copy. Cray conceived the Cray-4, which was supposed to be a compromise option in terms of price/performance in the high-power computing systems market, but his company began to run out of money, and in 1995 it went bankrupt.

Despite his engineering successes, Cray always remained a person who valued the process of technical creativity more than fame. He preferred to avoid publicity, did not give interviews, and was known for his modesty. Cray loved skiing, windsurfing, played tennis, and devoted his free time to a strange hobby: he dug a tunnel under his house. When asked about the purpose of such a strange structure, Cray smiled mysteriously and answered that this was the secret of his success: “While I’m working in the tunnel and digging it, elves often come to me, suggest solutions and help me deal with all the problems that arise.”

In October 1996, Seymour Cray was driving his Jeep Cherokee onto Highway 25 in Colorado when another car crashed into him. Cray died in the hospital from his injuries. Despite the fact that Seymour Cray’s life was cut short tragically, he is remembered as a man who proved that true innovation requires not only intelligence, but also the courage to go down a path that no one has gone before. Remembering his life story, it is difficult not to be imbued with respect for how one person was able to influence the development of an entire industry.

Source