Quicklinks

• What makes a Workstation a Workstation?

The Workstation Era

Workstations suddenly emerged in the early 80's. They dominated the technical computing scene for about twenty years, until they disappeared almost as fast as they arrived.

But what were those workstations, what was the reason behind the phenomenon paying as much for a desktop computing system as for an upper class sports car?

I think it was a quite simple but finally persuasive philosophy behind the workstations: Build a computing system under inclusion of the latest technology, and without any trade-offs. Consequently eliminate any bottlenecks, and deliver cutting-edge performance. Something comparable could be found in the area high fidelity consumer electronics. Some guys were willing to focus on the quality much more than on the price, so there was a market for high end audiophile systems.

Well, in fact the two examples are not completely comparable, because the workstations were used to solve real problems and most of all to earn real money. However, it is certainly true that when acquiring a workstation, the performance and availability had a much higher priority than the price.

At the time the first workstations emerged, one have to consider that there was a computing landscape which was dominated by mainframes and minicomputers at the one side and - still quite new - the home computers at the other, and lots of empty space in between. Systems with graphics capabilities were quite rare, and the PCs as we use to know them didn't yet exist. So computers were mainly a matter of computer professionals or of a handful of home computing enthusiasts. Imagine that most mainframe and minicomputer systems were controlled and used over terminals, connected to the host system with a communication line where 9600 baud were high-speed. Processing time was a high value and was assigned according to sophisticated time sharing schemes to each user.

On the other side, home computers started as basic microprocessor kits with very limited features beyond just running self-programmed routines and storing them on audio cassette tapes (which changed significantly with well-designed systems like the Apple II or the CBM PET). The IMSAI Altair 8800 and the IMSAI 8080 were the most famous of those early home computer systems, based on the S-100 bus.

Rapidly a wide ecosystem emerged around the 8-bit processors (mainly the Intel 8080 and the Zilog Z80), and the S-100 bus, with hundreds of companies and products, mostly targeted towards hobbyists with lots of knowledge on electronics (many items came as kits). Over just a few years lots of groundbraking innovations were created, which built the fundament of personal computers and mobile equipment we are using today.

But 8-bit microcomputers were still limited, and there was a need especially from engineers and scientists for unshared high-performance computing without the overhead and size of minicomputer systems, but with the ability to support graphical applications. And this is the point where the workstations came in. And it was the time, where the HP 9845 appeared years before all the workstations from SUN, Apollo or Silicon Graphics.

What makes a Workstation a Workstation?

Basically, the first workstations were minicomputers scaled down to desktop computers. This downscaling required a high level of integration. Typically, a minicomputer used multiple boards for the CPU, which had to be sized down to one single board, sometimes already with microprocessor technology. So the internal architecture of the early workstations was very similar to that of the larger minicomputer ancestors. This was true for the DEC MicroVAX (which was a desktop version of Digital Equipments popular VAX minicomputers) and also for the HP 9845 (which was basically a desktop version of the HP 2116 minicomputer, but some years before the MicroVAX).

In contrast to the home computers, which were limited to 8-bit microprocessors at a CPU clock rate of around 1 MHz and 64 kByte address space, the early workstations had a 16-bit or 32-bit processor and a much higher memory limit and CPU clock rate. But this was not the only difference between workstations and home computers. Even the early workstations were highly integrated, complete systems. They had networking and exceptional graphics capabilities, and a broad range of peripherals and professional software. And a service plan, which was optimized for high availability. Once a workstation had a failure, the complete faulty assembly was exchanged just on-site.

In 1969, Xerox created the Palo Alto Research Center (PARC) to develop products for the "office of the future". As we know, those revolutionary research efforts originated cutting-edge concepts like window based graphical user interfaces, mouse control and WYSIWYG, which all became essential for today's computers. Another development result was the Alto, a personal computer system incorporating the latest in technology and capable of demonstrating PARC's new concepts. The Alto was developed in 1972 and is commonly considered as the first system implementing the workstation concept, however as a prototype it had never been produced in higher numbers or offered commercially. It consisted of a fridge-size storage/processor cabinet with separate high-resolution black&white monitor, keyboard and mouse. Actually not a desktop system - due to the size of the storage/processor cabinet. As with most workstations, the architecture was a miniaturized version of a minicomputer, the 16-bit Data General Nova 1220. The Alto already featured an Ethernet network interface for connectivity to other workstations.

Certainly the HP 9845 was one of the first commercially produced workstations, and in my opinion it actually was the first real one. But the high time of the workstations startet years after with the systems from Apollo and Sun Microsystems. It was the potential of a single microprocessor, the Motorola 68000, together with a new kind of control, the graphical user interface, and the UNIX operating system, which opened the race on the workstations. At the time I personally was fascinated by the 68000, mainly because it had a fairly orthogonal design and actually was a 32-bit microprocessor with a 4GB address space. And of course I was impressed by graphical user interfaces, because they were extremely intuitive to use. The market for workstations was that for scientific and engineering computing. I used to work with a couple of Sun-1 during my study time, and, although the optical mouse mechanism was far from perfection, the choice between our VAX 11/750 and the Sun was no question. Mostly, because I didn't have to share the workstation with others. And this is what really makes up a workstation: unshared computing power, latest technology, uncompromised design and a graphical user interface.

The workstation development proceeded rapidly. Shortly after the Sun-1, the Sun 2 and Sun 3 workstations came on the market. Apollo, one of the real workstation pioneers who started their workstation development in 1981 with the DN100, was acquired by HP in the late 80's, who merged their own Series 9000 workstation line with that of the Apollo systems. So I had the opportunity to have a fully equipped HP Apollo 9000 model 755 under my desktop, not a quiet system, but for a price of roundabout $60,000. Also DEC wasn't lazy and continued their workstation line from the MicroVAX towards their DECStation systems based first on the MIPS processors and later on their own Alpha architecture, which was always ahead of the megahertz race, and first entered the 64-bit-world in the early 90's.

With Silicon Graphics another player entered the workstation market in the mid 80's, specialized for three-dimensional graphics in realtime. I still own an SGI Indigo, which marked the transition from the classical CISC microprocessor to the RISC CPU's. Still famous for its 3D flight simulator. Some workstation manufacturers like Intergraph moved towards the Intel architecture, others like Sun, DEC, IBM and HP started their own RISC processor line. During my first employment I worked in a project where all those workstation manufacturers like HP, Sun, IBM, DEC and SGI were involved and we had lots of fun benchmarking one against the other (later the PC and the Mac were added, but it was another league).

My first personal encounter with the 9845 was in the early 80's, must have been at a fair or something like that. I was tremendously magnetized by the futuristic design. I really was convinced, that this was what a computer should look like. And it was the space shuttle demo which made me a computer graphics believer. However I lost track of the 9845 and focused more on other systems (home computers, of course, and later other types of workstations). When I had to implement a graphical kernel system in the mid 80's, I remembered the space shuttle demo and used it as a template for my own GKS demo. Much later I turned back and tried to find out what the system was in the early 80's which had been such fascinating to me. And fascination still goes on.

But back to the workstations. In the late 90's, workstations became less important. The Intel and AMD processors gained more and more power, so did the VGA successors, and the need for special systems faded. In 2002, when even Sun Microsystems decided to use their SPARC architecture essentially for their servers, the workstations as we used to know them were definitively gone.