Supercomputing Secrets: Unraveling the 'Cray'-zy Designs

Before the NVIDIA's and the Intel's there was Cray. Who would have thought that the entire supercomputing revolution was the vision of a single man.

Part 1: Legacy

The Tyranny of Numbers

In the tumultuous landscape of early computing, Seymour Cray stood as a pioneering force, bravely confronting the 'Tyranny of Numbers.' A decade after the transistor's inception at Bell Labs, a new challenge befell Silicon Valley. The transistor, akin to a miniature vacuum tube, promised a revolution in information processing. However, assembling custom modules manually, each serving a specific function, gave rise to reliability issues due to potential faults in components or solder joints. Interconnecting numerous modules compounded complexity, creating what was termed the 'Tyranny of Numbers.' This hurdle, hindering the seamless operation of early computers, was eventually overcome by innovations like integrated circuits and microprocessors.

From Gliders to Supercomputers

Seymour Cray's journey commenced in Chippewa Falls, Wisconsin, where he was born. In 1950, he joined Engineering Research Associates (ERA) in St. Paul, Minnesota. ERA, born from a former U.S. Navy laboratory and initially focused on codebreaking machines, transitioned into general-purpose computing. Cray's expertise burgeoned, notably showcased in his work on the ERA 1103, the first commercially successful scientific computer. This computer, utilizing magnetic core memory with a capacity of 4096 words, marked a significant leap forward. The gliders group, where Cray worked, was later acquired by the Remington Rand Corporation.

Pioneering Supercomputing

"Thank heaven for startups; without them we'd never have any advances." ~ Seymour Cray

Post-acquisition, a faction of ERA employees, led by William Norris, founded Control Data Corporation (CDC) in Minneapolis, Minnesota. Cray joined as the chief designer, propelling CDC to create some of the world's fastest machines at the time, including the renowned CDC 1604. Cray, driven by an obsession to push the boundaries of computer speed, convinced management to establish a separate lab in his hometown, Chippewa Falls. Here, Cray and a team of 34 engineers developed the CDC 6600, a supercomputer that outpaced its competitors tenfold, marking a pivotal moment in 1964 and earning the title of the first supercomputer. Subsequent projects, such as the CDC 7600 and 8600, though groundbreaking, almost led CDC to bankruptcy, culminating in Cray's amicable parting with Norris.

Innovations and Challenges

As transistors rapidly scaled in speed during the 1960s, a slowdown prompted Cray to realize that CPU speed was only part of the system's overall speed. He shifted focus to efficient peripheral designs, famously stating, "Anyone can build a fast CPU. The trick is to build a fast system." This shift led to several revolutionary technologies, including the introduction of the first computer cooling systems.

Challenging the Norms

"If you were plowing a field, which would you rather use? Two strong oxen or 1024 chickens?" ~ Seymour Cray

Amidst this backdrop, Cray established Cray Research Inc. at the Chippewa Falls laboratory site. The pursuit continued to build the world's fastest supercomputer, resulting in the development of the Cray-1. Employing a revolutionary Vector Processing Approach, the Cray-1 could perform 32 calculations simultaneously. The National Center for Atmospheric Research's investment in the Cray-1, despite its hefty $8.8 million price tag, solidified the organization's financial standing.

GaAs and the Technological Frontier

Cray's innovative spirit persisted as he contemplated Gallium Arsenide (GaAs) for its faster properties during the Cray 2 project. Five years later, the Cray 3 incorporated GaAs for logic and Si for memory. Cray's bullish stance on GaAs, despite its challenges such as process and grain structure unreliability, exemplified his commitment to pushing technological boundaries. The Cray 3 project, faced with day-to-day challenges, prompted a relocation to Colorado Springs in 1988, partly influenced by the recession that left many defense personnel unemployed. This move attracted considerable publicity and interest in the 1990s. The period saw a surge in the recognition of the critical role supercomputing played in advancing other technologies, given its sophistication requirements.

A Visionary's End

The narrative takes a poignant turn in 1996 when, on September 22, around 3:00 PM, Seymour Cray met a tragic end in a car accident in Colorado. A collision on I-25 claimed his life, leaving behind a legacy synonymous with high-performance computing. Cray, by his thirties, had solidified his status as a genius in designing fast scientific computers, forever etching his name in the annals of computing history.

Legacy Lives On

As we reflect on Seymour Cray's journey – from the wooden glider factory to cutting-edge laboratories – his legacy endures. The man who challenged the 'Tyranny of Numbers,' envisioned a world of faster, more efficient computers, and whose name became synonymous with high-performance computing, Seymour Cray remains an enduring icon in the ever-evolving landscape of technology.

Part 2: Man behind the Legacy

Crafting Computers with Precision

In the intricate tapestry of Seymour Cray's career, his working style and technical philosophy stood as pillars defining his journey as a computing pioneer. From his early days at ERA to the establishment of Cray Research Inc. and beyond, Cray's approach bore the indelible marks of innovation, simplicity, and an unwavering commitment to speed.

Odd Hours and Deep Concentration

Cray's work ethic was characterized by an intense dedication to his craft. During his time at ERA, he would often work odd hours, seeking solitude to avoid distractions from coworkers. This habit persisted throughout his career, becoming a permanent part of his life. The odd hours allowed him the mental space needed for deep concentration, essential for the intricate task of designing cutting-edge computers.

Customer-Driven Innovation

His design philosophy was grounded in an astute understanding of customer needs. Cray had a methodical approach to innovation – he would take inputs from users of previous models, meticulously study their complaints and desires, and synthesize this information into the blueprints for his next design. This process, essentially a solo project for Cray, required support technicians, but the fundamental concepts were uniquely his own. He firmly believed that this was the most efficient and effective way to bring groundbreaking ideas to fruition.

Simplicity as a Key Principle

The first computer entirely designed by Cray himself was the ERA 1103, and this marked the inception of a design lineage that stretched to his Cray-4. Cray's key principle was simplicity. He avoided incorporating anything unnecessary, a methodology reminiscent of Reduced Instruction Set Computing (RISC) thinking, which gained prominence later. Time was of the essence for Cray, and he prioritized completing projects swiftly. Speed remained a fundamental concept in his quest to build faster computers.

Innovative Solutions Amidst Constraints

His tenure at CDC further solidified these principles. The CDC 1604, his initial project at the company, was a testament to Cray's ingenuity. In an unconventional move, he opted to use reject transistors from a local retail store, proving that substandard components could be designed around to achieve the goal of a working computer.

Founding Cray Research Inc.

Cray's dedication to scientific computing remained unwavering amidst changes in ownership at ERA and CDC. Despite shifts in focus prompted by acquisitions, Cray joined forces with Bill Norris to found Control Data Corporation, ensuring that the goal of building fast scientific computers persisted. Even as CDC later phased out large-scale scientific computing, Cray saw an opportunity to continue this pursuit. In 1972, he founded Cray Research Inc., using integrated circuits and vector register technology to usher in a new era of supercomputing.

Technological Visionary

"One of my guiding principles is don't do anything that other people are doing. Always do something a little different if you can. The concept is that if you do it a little differently there is a greater potential for reward than if you the same thing that other people are doing. I think that this kind of goal for one's work, having obviously the maximum risk, would have the maximum reward no matter what the field may be." ~ Seymour Cray

Cray's technical outlook became a hallmark of his career. His shift to Gallium Arsenide (GaAs) in the Cray 2 and Cray 3 projects showcased a forward-thinking approach. While GaAs offered faster properties, Cray was acutely aware of its challenges, including unreliability in process and grain structure. Yet, he pressed on, relocating the Cray 3 project to Colorado Springs in 1988, driven not just by technical challenges but also by the economic landscape.

Attention to Detail and Artistic Expression

Cray's meticulous approach extended beyond the design phase. As CDC grew into a larger company, distractions mounted, prompting him to retreat to Chippewa Falls and establish a research and development facility. His attention to detail was visible in the design of the Cray-1, not just in its powerful capabilities but also in its visually striking conical shape, showcasing an artistic expression rarely associated with computers.

Advancing Supercomputing in the 1990s

As Cray Research Inc. gained prominence, Cray's focus on creating the fastest scientific computers continued. The Cray 2, with its innovative design, and subsequent projects like the Cray 3, which incorporated GaAs for logic and Si for memory, pushed the boundaries of what was technologically possible. His emphasis on efficient designs and innovative technologies set the stage for supercomputing advancements in the 1990s.

Challenges and Unwavering Vision

Beyond his successes, Cray faced setbacks, as seen in the failed ventures of Cray Computer Corporation. Yet, undeterred, he continued to chase his vision. The tragic end in 1996, resulting from a car accident, marked the untimely conclusion of Cray's remarkable journey. His legacy, however, endures as the man who revolutionized high-performance computing.

Reflecting on Seymour Cray's Legacy

In reflecting on Seymour Cray's working style and technical outlook, his commitment to simplicity, speed, and customer-driven innovation shines through. His solo-driven design process, odd work hours, and ability to navigate technical challenges with unconventional solutions underscored a unique approach that left an indelible mark on the world of computing. Seymour Cray's legacy remains not just in the machines he designed but in the paradigm-shifting ideas and methodologies he introduced to the realm of high-performance computing.