How the BBC Micro helped conquer the world

The designers of the venerable BBC Micro computer recalled this month how it led to the development of one of the world’s two dominant processor architectures.

Sophie Wilson and Steve Furber were Cambridge graduates working for a local start-up called Acorn in the early 1980s when the BBC launched its Computer Literacy Project, one of the most successful mass-education exercises ever.

Acorn was one of seven companies asked to submit designs for a machine that would provide a standard platform for a series of TV teach-ins on computing. Wilson and Furber put together a prototype in just five days and it won the contract.

The Beeb, as it became known, came at an extraordinary time, when the Great British public was beginning to realise that almost anyone could afford a computer and learn to use it. Even so, the BBC was astonished by the response to its programmes. One in six of the population, men and women of all ages, viewed at least one of them; and sales of the BBC Micro, predicted to be around 12,000, eventually reached 1.5 million.

There was, according to John Radcliffe, executive producer of the literacy project, a lot of anxiety among viewers about whether they would be able to cope. “And the older people feared they would be outclassed by the younger ones,” he told a reunion at London’s Science Museum of BBC and Acorn people involved.

It used a six-year-old processor, the MOS 6502, and the first model had just 16KB of Ram. But it had lots of stuff proto-geeks could get into: a well-structured Basic language and ports capable of networking, controlling add-ons, downloading software from the TV via a Teletext adapter and even linking in a co-processor.

To keep the price down it used a TV as a monitor, connecting via the aerial socket using a design Furber adapted from one he found in Wireless World magazine. The greatest immediate impact on Acorn was psychological, said Furber. “The engineers became very confident that the things that they did would work.” They soon began to look round for a processor to power a successor to the Micro.

“We looked at the 16-bit processors that were around at the time, the Motorola 68000 and the National Semiconductor 32016, and we didn’t like what we found.

“These were very complex processors based on mini-computer architectures and they took a very long time to do some things. In particular they had a very poor interrupt latency, so that every time you wanted them to do something different it took them a long time to stop what they were doing and pay attention to what you wanted them to do.”

Acorn had taken on some chip designers and did not know quite what to do with them. A decision followed a trip Wilson and Furber took to the Western Design Center at Phoenix, Arizona, where the successor to the MOS 6502 was being drawn up.

“We expected to find big shiny American buildings full of big computers. What we found were a bunch of people working in a bungalow using Apple 11s and employing high-school kids over the summer to do circuit design. We came away saying that if they could design a processor, then so could we.”

Furber drew up a reference model, a kind of design template, for a new processor in 808 lines of BBC Basic code; and Wilson, now chief architect at Broadcom, worked on the instruction set. The project was kept secret in case nothing came of it. “Eighteen months later we found ourselves with a working, rather effective ARM [then standing for Acorn RISC Machine] chip. It was the 26th of April 1985,” Furber said.

“When we decided to make it public I had the strange experience of ringing up journalists and saying ‘We’ve made a new processor.’ And them saying: ‘We don’t believe you.’ And hanging up.” The first ARM was used as a co-processor for the BBC Micro. The next version, the ARM2, powered the fabled Archimedes desktop computer.

But Acorn, unlike Apple in the US, never had a home market big enough to allow it to withstand the dominance of Wintel machines, despite having technology that was in many ways superior, and the company was bought by Olivetti in 1985. However, the 32-bit ARM architecture had two things going for it.

Its reduced instruction set meant it had fewer hard-wired functions, a lower transistor count, and a smaller footprint than Intel chips. And it was designed to run cool to avoid the expense of fans in the price-sensitive educational market targeted by Acorn.

“That was serendipitous,” said Furber, ICL Professor of Computer Engineering at Manchester University. “We had to keep the power consumption below 1W. The [chip] design tools were not very good at the time and when we got the chip in it turned out to be drawing only a tenth of that.”

The result was that Apple used ARM chips in its ground-breaking 1993 Newton handheld. The machine was a flop, but it opened doors for Advanced Risc Machines, spun off from Acorn in 1990 to develop the ARM processor.

Two other trends buoyed up the company: the emergence of mobile phones, and the increasing use of systems-on-a-chip (SoCs) ­ packing all the modules for one application around a central processing core on a single piece of silicon. ARM’s business is now built around providing core designs for other companies to use in SoCs.

The number of devices using ARM cores exceeded 10 billion in January ­ more than one for each person on earth and far outnumbering x86 processors.

“It would not have happened without the BBC Micro,” said Furber. “Without that success we would not have had the confidence to design a microprocessor.”

There will be an exhibition dedicated to the BBC Micro at the Science Museum in 2009.

Like this story? Spread the news by clicking below:

 del.icio.us     Digg this     reddit!