Monday, 14 June 2021

Texas Instruments TI-99/4A (1981)

Introduced June 1981

When Texas Instruments (often known as just “TI”) entered the home computer, it wasn’t a typical player. Most machines were made by startups, or companies that had specialised in calculators and electronic games. Instead, TI was a massive and long-established electronics manufacturer which could trace its origins back to the 1930s, and by 1981 it was the largest semiconductor company in the world.

Rivals such as Motorola were happy to supply all the important bits and bobs to go into these new microcomputers, but that was as far as it went. However, TI chose to leverage its considerable expertise in silicon to try to carve out a slice of the market for itself.

In 1979 TI launched the TI-99/4, based on the 16-bit Texas Instruments TMS9000 CPU. The TI-99/4 was expensive, had a horrible keyboard and was limited in expansion capabilities. Two years later, TI fixed many of these issues with the improved TI-99/4A with a massively improved keyboard, clever expansion system and – crucially – a price tag that was half that of the original. The TI-99/4A looked promising to consumers, and sales started to take off.

It was a good-looking machine, with plenty of brushed aluminium and black which was in line with the aesthetics of the time. It wasn’t cheap, but the 4A’s price ticket of $525 was at least competitive unlike the 4. The graphics and sound were amongst the best in its class, so initially at least it seemed like a compelling proposition.

Texas Instruments TI-99/4A
Texas Instruments TI-99/4A

At its heart was the TMS9000 CPU, a sophisticated beast that was essentially a 1970s Texas TI-990 minicomputer on a single chip. It should have allowed the 99/4A to be one of the most powerful microcomputers on the market, but instead it was a major source of problems. Because building a full 16-bit system would be prohibitively expensive, almost all the internal architecture is just 8-bit which negated the possible performance impact. More difficult still was the fact that the 16-bit CPU’s instruction set was twice as memory hungry as a contemporary 8-bit CPU.

To get around this, TI essentially created an 8-bit virtual machine using an intermediate language called GPL. This made coding more efficient and was a technically advanced technique, but the processing limitations of the hardware meant that all of this sophistication created a computer that was significantly slower than its 8-bit rivals, despite running with a 16-bit core.

No computer of the era was perfect though, so the TI-99/4A wasn’t disadvantaged as much as you might think. But there were other problems – and the main one was software. TI were reluctant to share information about the platform with independent developers, instead TI wanted to produce the bulk of the software and peripherals for the 99/4A themselves – and thus profit from them. In truth, the TI-99/4A was probably better than most offerings but it was much weaker than the likes of the venerable Apple II or the upstart Commodore VIC-20.

But there was trouble brewing, and it was the Commodore VIC-20 which would deliver it in a giant tankard with a single raised finger painted on the side. Commodore’s boss – the legendary Jack Tramielloathed TI for nearly bankrupting his business during the pocket calculator wars of the 1970s. The VIC-20 ran on the 8-bit 6502 CPU (built by Commodore subsidiary MOS Technology) which was cheap, fast and well understood by programmers. The VIC-20 wasn’t as sophisticated as the TI-99/4A, but it was about half the price… at first.

Tramiel dropped the price of the VIC-20, TI followed suit. A price war emerged with both Commodore and TI dropping the prices until both units were shipping at less than $100. TI was haemorrhaging cash at this price point, but sales were good and it thought it could make the money back on software and peripherals. It couldn’t. TI started to lose hundreds of millions of dollars in this price war, driving the whole corporation into a sea of red ink. Even cost-cutting in production couldn’t turn it around – late 99/4As swapping to a cheaper beige case rather than the snazzy aluminium-and-steel one.

Late model fully-expanded TI-99/4A
Late model fully-expanded TI-99/4A

TI couldn’t sustain these losses, and in late 1983 it announced that the TI-99/4A would be discontinued. Production ended in the spring on 1984 and TI cancelled the interesting TI-99/2 and TI-99/8 systems that it was working on. Instead TI switched its efforts to 8088-based PCs running DOS, machines that were better than the IBM PC but weren’t IBM PC-compatible. In retrospect this was not a winning market strategy either. On and off TI stuck with the PC business, coming up with the TravelMate line of laptops which were quite successful, but TI sold their PC business to Acer in 1997.

Ultimately TI went back to concentrating on making the components that make the world go around apart from one consumer product – calculators. Yes, the product that so ired Jack Tramiel is still a profitable line for TI and outlasted the Company that dared to challenge it.

Image credits:
Max Mustermann via Wikimedia Commons – CC BY-SA 2.0
Leigh Anthony Dehaney via Flickr – CC BY-NC 2.0

Wednesday, 26 May 2021

Zuse Z3 Computer (1941)

Completed May 1941

Who made the world’s first programmable digital computer? The Americans? The British? The Japanese? Well, in what is perhaps a forgotten part of history it was quite possibly the Germans with the Zuse Z3 which was completed in May 1941.

The electronics of the time were not sophisticated, the Z3 relied mostly on relays and the whole machine ran at a little over 5 hertz (no, not megahertz.. just 5 cycles per second). It weighed a ton and drew 4000 watts of power, but it was actually remarkably capable.

Zuse Z3 detail
Zuse Z3 detail

Floating point numbers were supported, the Z3 could not only add and subtract, but divide, multiply and calculated the square root. Many of the computer’s operations were actually implemented in microcode rather than being hard wides. A keyboard and row of lights formed the basis of the operator console, and the Z3 could store data on punched celluloid tape.

Sometimes considered a design flaw, the Z3 was incapable of performing a conditional jump – i.e. the program couldn’t take a different path depending on different circumstances, an essential feature of a multipurpose computer. Still, the Z3 could perform complex calculations more quickly and accurately than a human, which is pretty much all early computers were used for.

It might not have escaped your attention that a lot was going on in Germany in 1941. Designer Konrad Zuse struggled to get resources for his project, but the much simpler predecessors of the Z3 (the Z1 and Z2) persuaded the Nazi government to support it in a limited way. Despite commissioning the Z3, the authorities were not convinced of its value and it was not used to its full capabilities. A bombing raid in 1943 destroyed the computer, and by this time Zuse had gone on the design the Z4 – arguably the world’s first commercially available computer - which was released a few years after the end of the war. Zuse continued to develop computers into the 1960s.

Zuse Z3
Zuse Z3

It is perhaps fortunate that the Nazis didn’t see the potential of the computer – in Britain the Colossus computer was breaking high-level German codes produced by the Lorenz cipher. This allowed the Allies to read communications from German high-command, including some from Adolf Hitler himself.

The Z3 ended up being largely forgotten, although a reconstruction was made after the war which now resides in the Deutsches Museum. In different circumstances – probably not very good circumstances considering – the Zuse Z3 might have been the progenitor of modern computing. But it wasn’t, instead the Zuse company was taken over by Brown Boveri in 1964 and then was sold on to Siemens in 1966, eventually disappearing in 1971. In 1999 the computer division of Siemens merged into a joint venture with Fujitsu, eventually being wholly taken over by the Japanese firm in 2009. Perhaps somewhere in there a little bit of Zuse DNA lives on.

Image credits:
DKsen via Wikimedia Commons – CC0
Floheinstein via Flickr - CC BY-SA 2.0

Saturday, 22 May 2021

IBM 7030 Stretch (1961)

Introduced May 1961

Sometimes products are released that look like they are sure-fire successes at the time, but end up in the long run as being insignificant. Sometimes products are launched that look like failures, but end up changing the world in some way. The IBM 7030 Stretch is a little of one and a little of the other.

The 7030 was IBM’s first fully transistorised computer, and at launch it was the fastest computer in the world. Projected to be priced at an eye-watering $13.5 million dollars in 1961 money (about ten times that today), this was a serious computer for serious organisations – coming in at 32 metric tons and consuming 100kW of power.

Transistor technology had been developing at a rapid rate by the start of the 1960s and IBM proposed using diffusion transistors for the new design. This was a risk move for the typically risk-averse IBM, but competition with companies such as UNIVAC was heating up. The initial goals for the 7030 was impressive – a 64-bit system capable of a processing capacity of 10 MIPs. When the technical complexities of the project began to dawn, this was dropped to 4 MIPS. When the 7030 was launched, it actually shipped with 1.2 MIPS. 

IBM 7030 Stretch
IBM 7030 Stretch

The system performance was a disappointment – even though it turned out that the 7030 was the fastest computer in the world. IBM cancelled new orders and halved the price for those who had already ordered it. In IBM’s eyes, the 7030 was a failure. Just 9 units were sold – including one secret version known as “Harvest”. There were significant internal recriminations at IBM, with plenty of finger-pointing going round and people anxious to assign blame.

But the 7030 was more of a technological success than was realised, and the innovations in hardware and software found their way into other IBM products, especially the successful IBM System/360 series which found their way into corporations everywhere. And although many of the technologies in the 7030 were soon obsolete, they all provided an important stepping-stone in the development of 1960s computing.

IBM 7030 Stretch

Two key figures in the 7030 Stretch were Gene Amdahl, a legendary designer of powerful early mainframes and Frederick Brooks who went on to write the seminal software engineering tome “The Mythical Man Month”. This book attempted to learn from the mistakes in the project management of Stretch and other projects, significantly the idea that adding more people to a late software project will only make it later. Despite being in print for more than 40 years, corporations continue to make the same mistakes that IBM did in the early 1960s.

Image credits: Don DeBold via Flickr - CC BY 2.0
[1] [2]

Thursday, 29 April 2021

Fiat 127 vs Morris Marina (1971)

Introduced April 1971

Two cars, two different design philosophies – the Fiat 127 and Morris Marina were both introduced in April 1971. One ended up being celebrated, the other derided. But which is which?

Fiat 127

Ah yes – the Fiat 127, the cute and inexpensive Italian hatchback of the 1970s. But it wasn’t a hatchback… well, not at launch, in 1971 you had a two-door saloon that had a boot at the back. Remember than even in the 1970s, the idea of the hatchback was a radical one… even though today it is an obviously versatile way to build a small car. The little 900cc engine gave a respectable 46 horsepower for such a car weighing about 700kg. The design was innovative enough, with crumple zones and excellent road holding – helped by the car’s front-wheel drive - with decent interior space as well.

Fiat 127
Fiat 127

A year later the hatchback version arrived – this is the version that really sold well – a major facelift in 1977 gave a more modern look and better engines. A further revision in 1982 sneaked in just before the launch of the Fiat Uno in 1983, and licenced versions built overseas lasted even longer.

This radical car was designed by Pio Manzù, who was tragically killed in a car accident before the 127 came into production. Manzù was an exceptionally talented young designer of lamps, clocks and furniture before turning his hand to automotive design. Just 30 years old when he died, it is likely that Manzù would have become one of the great car designers given the chance.

The Fiat 127 was massively influential – arguable the first modern hatchback design (well, eventually) – it set a pattern for small cars that it still in use today. Despite selling in huge numbers, only about 100 are still on British roads.

Morris Marina

Where the Fiat 127 predicted the future, the Morris Marina was instead a quick fix to British Leyland’s problems in the late 1960s with competing with the Ford Cortina. Available as a traditional four-door saloon or a rather rakish coupé, the Marina used tried-and-tested components to come up with something that wasn’t all that exciting, but for a while was certainly successful.

Morris Marina
Morris Marina

A reputation for unreliability and variable build quality, the Marina fell out of favour by the late 1970s and quickly became something of a joke, but this was probably unfair. It had been designed in a hurry and with a minimal budget, and yet it did everything that an early 1970s fleet buyer would want. It was certainly competitive with the Cortina.

The Marina’s Cortina-like capabilities were perhaps no coincidence. Designer Roy Haynes who created the Marina was also largely responsible for the Mark II Cortina. Haynes went on to other things before the Marina was launched however, and again here was another designer who had a chance to be one of the all-time greats but things didn’t quite pan out.

The Marina continued on until 1980 when it was replaced by the Ital – essentially a heavy facelift of the Marina – which continued until 1984. The Ital was the end of the line for Morris though, in the end the Marina was a dead end. Fewer than 400 Marinas are still on British roads.

Well, almost – the Ital briefly emerged again as the Huandu CAC6430 in China in the late 1990s. But it was the utterly magnificent door handles that had a life of their own, turning up in all sorts of exotic designs such as Lotuses and Ginettas.

Image credits:
Robert Capper via Flickr – CC BY-NC 2.0
Qropatwa via Flickr - CC BY-NC-ND 2.0

Monday, 26 April 2021

Hayes Smartmodem (1981)

Introduced April 1981

The sound of modems belongs to a certain era before ubiquitous broadband. The characteristic screeching noise – a bit like a fax – joined in the rat-a-tat of dot matrix printers and the clunk clunk of mechanical disk drives. Modems had been around since the 1950s, a way of connecting computers together over the omnipresent telephone network. But initially they were very expensive and aimed at really big computer systems. As time went on, they became more affordable.. but not very usable.

By the late 1970s modems for computers had developed in two different ways – the acoustic coupler was a device where the screeching noise was fed into the telephone handset (often via a serial cable) or alternatively where it was permitted, the modem would be a card inside the computer with a telephone connection built in. But the acoustic coupler was bulky and slow, but the alternatively tended to need a modem designed specifically for every model of computer on the market – and these were only available for large systems and not the booming microcomputer market. There had to be a better way.

What was needed was a modem that could work with just about anything. That was the Hayes Smartmodem.

Hayes Smartmodem
Hayes Smartmodem

The “Smart” in Smartmodem wasn’t just marketing speak. It took advantage of the fact that almost every microcomputer system on the market had a serial port fitted, or one available as an optional extra. By using a serial port, all you needed to connect the Smartmodem to your micro was the appropriate cable, which if you couldn’t buy you might be able to make yourself.

But the implementation of serial ports varied from computer to computer, the Smartmodem needed to know (for example) when to hang up the connection. More sophisticated computers used pins to indicate when the connection was up or down, but cheaper ones didn’t. The serial port could also be running at a variety of speeds, so matching the port with the modem could be tricky on some devices.

Hayes designed a series of commands to control the modem, all beginning with “AT” for “attention”. But because all the commands started with the same two latters, the modem would attempt to use this to automatically match the baud rate of the computer with the modem. This made setting up the Smartmodem much simpler, but there was a more tricky problem… how could you tell the modem to hang up if you had a basic serial port that didn’t have the right wires to signal that to the modem?

To achieve this, the Smartmodem would look for a sequence “+++” followed by a once-second pause. This would break out of the communications sequence and make the modem ready to receive a command. If that command was ATH then the modem would hang up. It was an elegant solution, the pause minimised the possibility of it happening by accident. When competitors tried to copy the functionality of the Smartmodem, this was often implemented badly (in part because Hayes had patented it and demanded a fee). This meant that on some non-Hayes modems could hang up completely at random if they were sent the +++ sequence accidentally.

In these pre-internet days, modems would be use to connect computer directly to another computer, or perhaps to a BBS (bulletin board system) or if you were one of a small number of privileged people, they could act as a gateway to what was to become the internet.

The original Hayes Smartmodem was a 300 baud unit, but as technology improved the Hayes modems became faster. However, competition was also becoming fierce and although Hayes products were expensive, they were also very reliable and they carved themselves a healthy share of the market.

All good things come to an end though, and in the 1990s Hayes bet the barn on ISDN products, a market that never materialised. Competitors such as USRobotics were taking the analogue modem market. By the mid-1990s Hayes were in serious trouble. Bankruptcy and mergers led to them being subsumed into rivals Zoom Telephonics, where the brand eventually vanished. However, the Hayes website is still available although it was last updated in 2013.

Surprisingly, analogue modems are not quite a dead technology but their heyday has certainly passed. However, Hayes certainly made it easier to connecting disparate computer systems together and in part that drove the uptake of the internet in the 1990s.

Image credit:
Michael Pereckas via Flickr - CC BY-SA 2.0

Sunday, 18 April 2021

Xerox Star 8010 (1981)

Introduced April 1981

By the early 1980s, hardware and software designers had great dreams about what they wanted products to be. Portable perhaps, affordable or business-oriented… but all of these were constrained by technology and price. But what if you dreamed big and without compromise, and built the best computer system you possibly could? This is what Xerox did.

The catchily-named Xerox 8010 Information System – more commonly known as the Xerox Star – introduced potential customers to the graphical user interface, mouse, Ethernet, servers and email. A great deal of modern computing technology was first available in the Star, but it certainly came at a price.

Xerox Star 8010
Xerox Star 8010

It had been a very long journey. Doug Engelbart’s Mother of All Demos in 1968 had introduced many of these modern concepts, but running on primitive hardware. Many of Engelbart’s team migrated away to the giant Xerox corporation which had pioneered photocopiers and laser printers. The fortunes of Xerox were very much based in paper, but the concept of the paperless office loomed large and Xerox wanted to still be in business when paper was consigned to museums.

The Xerox Alto was their first attempt, launched in 1973 it incorporated a GUI (graphical user interface) and a mouse, but it was never sold commercially. Instead the Alto was deployed around the Xerox PARC as well as some universities and research organisations. It took another eight years for Xerox to realise a commercial product – the 8010 – but even though it had taken over a decade since Engelbart had shown the concepts, the Star was still way ahead of everyone else.

Strictly speaking, “Star” referred to the software rather than the hardware. And this wasn’t simply a computer you could buy and take home. Doing anything required a network, some servers and perhaps $100,000 in 1981 money for a small installation (about $250,000 today).

Japanese market Fuji Xerox 8012-J
Japanese market Fuji Xerox 8012-J

The price like the name was astronomical. But what that substantial wedge of cash bought you was a computer system with a high-resolution 17” monitor, a carefully thought out software interface that could work collaboratively with others, based on the high-end AMD Am2900 CPU. And the software was like nothing else.

Everything was WYSIWYG (“what you see is what you get”) – you could edit two pages of a document side by side, including charts and tables from other applications and when they printed out they matched what was on the screen. You’d expect that today, but in 1981 it was revolutionary. The clever object-oriented operating system delivered features to the desktop that wouldn’t be common until a decade later.

There’s a problem with trying to sell customers a product that they don’t know they want at a price they can’t afford... all the efforts of Xerox to create an advanced computer system did not translate into many sales. Xerox tried to reposition the 8010 into a desktop publishing platform called the 6085 (aka Daybreak) which included a laser printer, and although this was a capable system it was still expensive and sales were slow. Later attempts to port the software to OS/2 and other platforms also failed. Xerox weren’t done with WYSIWYG though, a spin-off created the iconic Ventura Publisher, but that was only a passing success.


Xerox Star UI
Xerox Star UI

Despite being a sales failure, the Star was a technological success. In particular elements from the Star user interface found their way into the Macintosh, Windows and a host of other platforms. The networked environment too was increasingly emulated by competitors. As is sometimes the case with big, sprawling companies the Xerox Corporation itself did not seem to understand or be able to protect its own intellectual property. As with many pioneers, it was other adapting their idea that made it a success. Today elements of the Star user interface are pretty much everywhere, but this pioneering system is long dead.

Image credits:
Rhys Jones via Flickr - CC BY-NC 2.0
Fuji Xerox - Courtesy of FUJIFILM Business Innovation Corp
Amber Case / Digibarn via Flickr - CC BY-NC 2.0

Sunday, 11 April 2021

Victor 9000 / ACT Sirius 1 (1981)

Introduced April 1981

By 1981 the business microcomputer market was developing very rapidly. First generation 8-bit systems were giving way to more powerful 16-bit systems, and so too a new generation of computer companies were challenging the early pioneers.

One of these companies was Sirius Systems Technology, founded by (among others) the legendary Chuck Peddle who had designed the Commodore PET and MOS Technology 6502. Peddle and his team then set about designing a next-generation computer system based around Intel’s 8/16-bit CPU, the 8088, called the Victor 9000.

Now you’ve probably heard about the IBM PC, also launched in 1981 and eventually finding its descendants on just about every work desk everywhere. The Victor 9000 was better and hit the market first, but would it be enough to succeed? The answer is complicated.

Victor 9000
Victor 9000

It was based around a 5MHz 8088 CPU with between 128Kb to 896Kb of RAM, a high-resolution 800 x 400 pixel display, clever variable-speed floppies with up to 1.2Mb of storage, a bunch of interface ports and a very pleasing industrial design. On top of this the Victor 9000 could run CPM/86 (the 8086/8088 version of CP/M) and could also run Microsoft’s new (although slightly recycled) MS-DOS operating system. A useful wordprocessor, spreadsheet and financial management software could be bought to run on it.

Overall, this was a good and extremely competitive system… and perhaps it could have been a world leader if it wasn’t for the launch of the IBM PC in the US in August 1981. The PC was more expensive and less capable, but the magic three letters “IBM” ensure that larger corporations went out and bought it. Sales of the Victor 9000 were disappointing in the United States… but IBM waited another 18 months to launch the PC in Europe where the market was wide open.

Sleeker Victor 9000 with half-height drives
Sleeker Victor 9000 with half-height drives

In Europe, the Birmingham-based Applied Computer Techniques (ACT) acquired a licence to sell the Victor 9000 as the ACT Sirius 1. With little competition, the Sirius 1 became a major success in the UK and Germany in particular, even though it wasn’t really PC compatible in any meaningful way. Of course when IBM did start shipping into Europe, sales of the Sirius I were hit badly.

ACT Sirus 1 advertisement
ACT Sirus 1 advertisement

For the US-based Sirius Systems, their history was short one that followed a traditional path – only three years after the launch of the Victor 9000 they were bankrupt. It was a different story for ACT who launched several generations of advanced but not-quite-PC-compatible computers under the “Apricot” brand afterwards including the world’s first production system based on a 486 CPU. A takeover by Mitsubishi in 1990 was effectively the end of the independent Apricot brand - indeed Mitsubishi shuttered operations in 1999 – but it outlasted Sirius Systems, and along the way ACT kept innovating and was probably far more influential than its American partner.

Image credits:
Samuel via Flickr - CC BY-NC 2.0
Bradford Timeline via Flickr - CC BY-NC 2.0
The Henry Ford Museum - CC BY-NC-ND 3.0