While Ken Olsen and his Digital Equipment Corporation (DEC) are widely acknowledged as originators of the minicomputer, it appears it was, in fact, another IT luminary, one who is better known for a far more powerful class of machine.
Seymour Cray is the celebrated creator of the supercomputer (more in next month's issue of Brainstorm), but he apparently also built the first commercial minicomputer when he worked for Control Data Corporation (CDC).
Author Paul Ceruzzi elucidates in A History of Modern Computing (MIT Press, 2003): “Around 1960, CDC introduced its model 1604, a large computer intended for scientific customers. Shortly thereafter, the company introduced the 160, designed by Cray (‘almost as an afterthought', according to a CDC employee) to handle input and output for the 1604… When CDC offered a standalone version, the 160A, for sale at a price of $60 000, it found a ready market.”
Meanwhile, DEC had been working on its first machine and the Programmable Data Processor (PDP-1) was announced in 1959. However, the first commercial sale only took place three years later, according to Ceruzzi.
While Cray's design was created by necessity, reports indicate that DEC found its inspiration in an earlier body of work: the Massachusetts Institute of Technology's (MIT's) Whirlwind project. Funding a Revolution: Government Support for Computing Research (National Academy Press, 1999) refers: “Whirlwind originated in 1944 as part of the [US] Navy's Airplane Stability and Control Analyser (ASCA) project. At that time, the navy made extensive use of flight simulators to test new aircraft designs and train pilots. However, each new aircraft design required a separate computer specially created for its particular design.”Apparently, ASCA was intended to negate the need for individual computers for flight simulators by creating a single general-purpose simulator that could emulate any design programmed into it. Project leader Jay Forrester very quickly recognised that analogue computers (typically used on aircraft simulators) would not be fast enough to operate in real time. Learning of work in electronic digital computing as part of ENIAC at the University of Pennsylvania, Forrester started investigating the potential for real-time digital computers for Whirlwind.
By early 1946, Forrester decided to pursue the digital route, expanding the goal of the Whirlwind program from building a general aircraft simulator to designing a real-time, general-purpose digital computer that could serve many functions other than flight simulation. This proved to be rather fortuitous given the looming Cold War.
A symbiotic relationship
The Whirlwind computer project was refocused to underpin a new air defence system known as Semi-Automatic Ground Environment (SAGE).
“SAGE… made several important contributions to computing in areas as diverse as computer graphics, time-sharing, digital communications and ferrite-core memories. Together, the two projects shared a symbiotic relationship that strengthened the early computer industry. By late 1951, a prototype ferrite-core memory system was demonstrated, and by 1953, the Whirlwind's entire memory was replaced with core memory boasting a nine-microsecond access time, effectively ending the research phase of the program,” continues the report, full text of which is available online at www.nap.edu.
But Whirlwind was remarkable for another reason more pertinent to the story of minicomputers: one of DEC's founders also worked on the project. Ceruzzi explains: “As a student at MIT, [Ken] Olsen had worked on fitting the Whirlwind with core memory in place of its fragile and unreliable storage tubes, and in the mid-1950s he worked for MIT's Lincoln Laboratory... In 1955, he took charge of a computer for the Lincoln Lab called the TX0, a very early transistorised machine. Under his supervision, the TX0 first operated at Lincoln Lab in 1956.”
A year later, Olsen, his brother Stan and Harlan Anderson started DEC and, as mentioned earlier, the company unveiled the PDP-1 in 1959. According to Edgar Schein, author of DEC is Dead, Long Live DEC, the machine was designed by Ben Gurley and was revolutionary for its small size (analogous to a refrigerator) and relatively low initial price of $120 000. Despite this, however, the new minicomputer was hardly a commercial success because, as Ceruzzi notes, DEC only sold about 50 machines. However, he asserts that it deserves a place in history for its architectural innovations – innovations that he claims were as profound as those embodied in John von Neumann's report on EDVAC (Brainstorm, April 2007).
By design, PDP-1 incorporated many of the architectural and circuit innovations Olsen had seen in the TX0. And where typical computers of the day were often simply transistorised iterations of vacuum tube models preceding them – replacing tubes with transistors on a one-to-one basis – DEC's PDP-1 owed nothing whatsoever to tube design. It was intended to take full advantage of what transistors had to offer from the start.
In the end, DEC designed 16 PDP models (www.computernostalgia.net hosts a timeline spanning 34 years), but not all of them were built. And those that actually left the drawing board met with varying degrees of success but, typically, important lessons were learned along the way. A classic example was the PDP-5. Designed by Gordon Bell and Edson DeCastro, it represented DEC's first attempt at a 12-bit machine and emerged in 1963. According to Schein, the lessons learned led directly to a much-improved successor, the PDP-8, two years later.It was, as Ceruzzi, points out, the minicomputer that first revealed the potential size of the target market. He notes: “The modest appearance of the PDP-8 concealed the magnitude of the forces it set into motion. Mainframe computing would persist, although its days of domination were numbered... The mini showed that with the right packaging, price and, above all, a more direct way for users to gain access to computers, whole new markets would open up.”
Schein's book echoes this sentiment through a 2002 memo penned by a DEC alumnus: “DEC popularised personal computers, although it wasn't called that and it wasn't for home use. The concept was for a researcher or engineer to be able to justify the purchase of a computer dedicated to his own tasks/project.”
Ceruzzi adds: “The simplicity of the PDP-8's architecture, coupled with DEC's policy of making information about it freely available, made it an easy computer to understand. This combination of factors gave rise to the so-called original equipment manufacturer (OEM), a separate company that bought minicomputers, added specialised hardware for input and output, wrote specialised software for the resulting systems and sold them (at a high mark-up) under its own label.”
Indeed, OEMs sprung up left, right and centre throughout the company's history, increasing sales and broadening the appeal of its designs. But despite having two distinct markets, it still wasn't plain sailing for DEC because, as is often the case, some of the company's biggest strengths were also weaknesses. Schein reports that the company deliberately courted the best and the brightest, and encouraged them to come up with proposals that, if approved, would be their responsibility to implement. Since it is obvious that not all such proposals would be approved, it was inevitable that resentment would rear its ugly head.
He writes: “The main consequence of that kind of culture was that people developed self-confidence and became more willing to trust their own judgement. Combining that with a track record of success and growth led inevitably to powerful subunits that developed their own strategic agendas and subcultures. The first indication of that kind of process was the departure in 1968 of DeCastro and a group of engineers to form Data General when their view of what the next product should be was not approved by DEC senior management.”
Nova rears its head
Given that the group's product plans where already well advanced by the time they left DEC, it took the new company mere months to produce its first product: the Nova. ComputerHistory.org provides an indication of how much price had plummeted with all the new players in the market. The Nova shipped with 32 kilobytes of memory and sold for only $8 000.
Back at DEC and the minicomputer revolution started by the PDP-8, the company found itself at a disadvantage in 1969. Not only had DeCastro and his cronies deserted and become competitors, but others were coming out of the woodwork too. As the University of Sheffield's Stuart Bennett recalled in a paper presented to the 15th Triennial World Congress in Barcelona in 2002: “Scientific Data Systems, with its SDS-910 and SDS-920 computers, General Electric with the GEPAC4000 series machines and the Computer Control Corporation, with its 16 bit DDP-116, were all early competitors.”
Ceruzzi reports that between 1968 and 1972, around 100 new companies or divisions of established companies offered minicomputers on the commercial market, an average of one every three weeks for that five-year period. Inevitably, consolidation followed: Honeywell acquired both Computer Control Corporation (1966) and General Electric's computer business (1970). Around the same time, Xerox bought Scientific Data Systems, renamed it Xerox Data Systems and then closed it down in 1975.
Long-forgotten names from the period include: Computer Terminal Corporation aka Datapoint (taken over by corporate raider Asher Edelman in 1985), Modular Computer Systems aka Modcomp (diversified into the integration business and eventually acquired by CSPI in 1996), Prime Computer (closed its hardware division in 1992 to focus on computer-aided design software) and Varian Data Machines (incepted in 1966; acquired by Sperry in 1977). A few well-known names entered and then subsequently exited the business too. These included Lockheed Electronics, Motorola, Raytheon, Texas Instruments and Westinghouse. HP had also dipped its toe in the water in 1966 by launching the first model in its 2100 minicomputer series (later renamed to HP 1000).
All this activity was having an effect and, although DEC had kept up with industry developments by releasing a version of the machine based on integrated circuits (Brainstorm, February 2007), by 1969, the PDP-8/I's 12-bit word length and limited instruction repertoire had become dated and no longer compared well with the competition. That year also saw the arrival of the 300-pound gorilla, IBM, introducing its System/3 in July of that year. HP also introduced its first time-shared operating system on a minicomputer.
DEC needed to do something, fast. Fortunately, the company's engineers hadn't been idle and in January 1970, the company was able to announce the PDP-11. Better yet, it was able to ship the first models of what would become its most successful minicomputer ever in mere months.
Fate also intervened in the form of the recession that hit the computer industry that year. Companies still needed computers; they just couldn't afford the sums demanded by the likes of IBM. As a result, the minicomputer business boomed, the biggest beneficiaries being DEC with its PDP-11 and Data General with its Nova, according to Ceruzzi.
Mini boom time
“Nova's success came from its elegant design and innovative packaging; the PDP-11's from its innovative architecture, which opened up minicomputers to a host of applications that had previously been the domain of mainframe computers,” he writes.
Answers.com elaborates on the PDP-11: “Possibly the single most successful minicomputer design in history, a favourite of hackers for many years and the first major Unix machine. The first PDP-11s shipped in 1970; and the last in 1990. Along the way, the 11 gave birth to the VAX, strongly influenced the design of microprocessors such as the Motorola 6800 and Intel 386, and left a permanent imprint on the C language (which has an odd preference for octal embedded in its syntax because of the way PDP-11 machine instructions were formatted).”
This dominance is illustrated by Ceruzzi in his text when he notes that DEC sold over 170 000 PDP-11s in the 1970s alone. The 1980s proved even more successful for the company, if Computergram International's Gary Flood is to be believed. He wrote that the company had shipped more than 600 000 of the systems by 1989. Similarly, Answers.com asserts that Data General shipped its 100 000th computer during the course of 1981.
The advantage enjoyed by the two Massachusetts-based companies was maintained despite a constant barrage of new products from competitors. IBM, for instance, made numerous attempts with its System/32 (1975), Series/1 (1975), Systems/34 (1977), System/38 (1978), IBM 8100 (1978) and System/36 (1983). It was all to no avail because DEC and, to a lesser extent, Data General dominated the segment into the late-1980s by constantly releasing more powerful machines. In the end, however, both companies were victims of their own success.
Writes Ceruzzi: “The mini generated the seeds of its own destruction by preparing the way for personal computers that came from an entirely different source.”
But Paul Kampas, who worked at DEC from 1976 to 1994, expresses it differently. In a chapter he contributed to Schein's book, he refers to the multiple disruptive technologies that were emerging in the mid- to late-1970s, which he says started to usher in the new computing model of client-server architecture and low-cost PCs and workstations.
He suggests that as the market grew and technology matured, a new breed of competitor started to emerge: category killers. Kampas defines these competitors as vendors who specialise in one or a few very closely related product categories.
“The emergence of category killers created a huge competitive challenge to DEC. Two of the biggest competitive issues were cost and openness,” he writes.
However, Paul Carroll puts forward a different reason in Big Blues: The Unmaking of IBM (1993, Weidenfeld & Nicolson). After detailing many reasons IBM was in dire straights in the late 1980s, not least of which was its consistent lack of success in the booming minicomputer market, he writes: “The only thing that saved IBM in 1988 was a project... to combine two of IBM's incompatible minicomputer lines – the System/36 and System/38 – while preserving the best features of both.”
The resulting machine was dubbed the AS/400 and, Carroll reports, it quickly grew into a $14 billion a year business and “also sent arch minicomputer competitor DEC and its snotty ads crashing to earth. DEC had taunted IBM for years by exploiting its penchant for pre-announcing products with adverts under the slogan ‘Digital has it now!'.”
In the final analysis, the truth lies in your own perspective. If you believe the minicomputer died with DEC, you'll probably accept Ceruzzi and Kampas' explanations. If, on the other hand, you maintain that it lives on in the form of today's midrange server category, then you'll probably favour Carroll's perspective. One man's minicomputer is another man's midrange server.
One thing, however, is indisputable: both Data General and DEC are no more. The erstwhile minicomputer giants both became takeover targets in the 1990s. As you will probably recall, the latter was acquired by Compaq, which itself was later absorbed by HP. Data General, on the other hand, managed to hold out a little longer and diverged into storage with remarkable success in the form of its Clariion range of network-attached storage (NAS) devices. But that very success proved its undoing when storage giant EMC went on the acquisition trail in 1999. So ends an era.