Friday, July 19, 2013

From a small Acorn to 37 billion chips: ARM's ascent to tech superpower

From a small Acorn to 37 billion chips: ARM's ascent to tech superpower

The BBC Micro. A British computing legend

Every day, you use one - or probably more - of the 37 billion ARM chips produced so far. They're inside your phone, inside your tablet, inside your TV, and inside numerous other devices.

But they began life as a second development processor for the rather beige mid-1980s BBC Micro at the Cambridge, UK-based Acorn Computers.

The remarkable Sophie Wilson was the designer behind the ARM (Acorn RISC Machine) instruction set, which originally began in October 1983. And so we thought TechRadar's Brit Week was a great time to speak to the recent European Inventor Award nominee about her time at Acorn, the beginnings of ARM, and the huge project she's now working on.

"When we designed the BBC machine in the 1980-1981 period, we were essentially designing our own ideal machines," explains Wilson. "We thought it was a good machine. The BBC had asked for 12,000 and they thought they were being pessimistic."

Selling the BBC Micro

The BBC had commissioned the MOS Technology 6502-based computer to go alongside the BBC Computer Literacy Project for education, but the shipments were far bigger than anybody had envisaged.

BBC Computer Literacy project
The logo of the BBC Computer Literacy project

"We'd thought about 25-50,000 units, but it was a million and a quarter in the end. They were everywhere and they were doing everything. People bent and twisted them in ways we'd never imagined.

"We'd be forever seeing something that we had no idea could be done with a machine that we'd developed - a case in point being David Braben's Elite. We could not believe that he got that into the machine."

As for the ARM microarchitecture itself, we wondered whether Wilson realised how much potential it had at the time.

"When we set the project up we had a slogan internally to remind us what we thought we were doing, and that was 'MIPS for the masses', i.e. lots of processing power for everybody. We were aiming at the mass market." (MIPS means Microprocessor without Interlocked Pipeline Stage, a RISC – or Reduced Instruction Set computer microarchitecture.)

Wilson is at pains to stress that ARM's rise to success has taken place gradually over a 30 year period. "We had our first working chips back in April 1985, and we put them into Acorn machines and they were very good, and we got our market and people liked it a lot," she says, rather matter-of-factly.

The original ARM chip [Image credit: Broadcom]

Moving ARM forward

Acorn started being approached by other people to use the ARM design. "We set up a division in Acorn for third parties and eventually we spun out ARM in 1990 because there seemed to be a market. By then we'd been approached by Apple, for example, and throughout 90s they kept making little bits of process. Nokia came on board in 96 or 97, then TI."

"At every stage there was just another customer, a little bit extra. And it just kept adding up, and when we did the sums in 2008 we'd shipped 10 billion ARMs. Now we can be remarkably blasé about shipping 36 or 37 billion of them. It's a gradual success over 30 years. "

"It's extremely well grounded - that's why there's so much depth to the architecture. It's been there for so long, and it's only been about in the last five years that the public has even vaguely started knowing about it.

"It's also to do with the decisions taken by our management over many years to have all that depth. So it's not merely the high profile apps processors that everybody talks about competing with Intel and taking sockets in mobile phones and tablets, it's all the Cortex-R, Cortex-M series and before them in particular the ARM7 TDMI that have just got absolutely everywhere.

"And that's the secret. It's an enormous ecosystem. ARM succeeds through being in partnership with everybody, essentially. Even Intel has an ARM license. Even Intel still sell single chips for phones with ARMs in them."

ARM
More thabn 37 billion ARM processors have now shipped

More Acorn machines and RISC OS

One of the more surprising aspects of Wilson's chat about the early days at Acorn was how certain the team was that they would succeed. "We were supremely confident," she says, without a hint of irony or doubt.

"The team of people that created ARM, particular Steve Furber and myself, had been working together for long enough to have a good rapport and working relationship, and we'd never failed at doing anything. Everything that we tackled we'd succeeded.

"Designing a microprocessor as Steve has remarked is just another complicated piece of digital logic and he was good at designing digital logic! And designing the instruction set, I'd actually designed fantasy instruction sets before so even that was another logical step forward. It all felt extremely possible.

"Furthermore we had a conviction that we knew what people were doing wrong. We had chips in our hands from Intel, Motorola and National Semiconductor, and we could see why they weren't performing well. We set out to remedy that in making ARM, and we were quite right."

After the BBC Micro, Acorn launched the fully ARM-based Archimedes in 1987. But RISC OS - Acorn's advanced and rather Windows 95-like operating system - wasn't ready. "There was a year which we had to go with an operating system that was essentially a clone of the BBC operating system, and that was painful because it wasn't good enough," says Wilson regretfully.

"But yes, in 1988 RISC OS came out and that was dramatic because it was then a fully-featured system that could do things that few machines could at the time.

"It was a machine with a high-resolution machine with anti-aliased graphics with WIMP (Windows, Icons, Mouse, Pointers). The Macintosh had been around for a time but it didn't have anti-aliased graphics, so the on-screen experience was very poor. RISC OS gave you WYSIWYG like nobody have ever seen."

RISC OS lives on, and you can download it for the Raspberry Pi - check it out as part of our feature Raspberry Pi operating systems: 5 reviewed and rated.

But while Acorn's microprocessor was strong, its advanced hardware had a surprisingly short lifespan due to the success of the IBM PC, and even by the time it was releasing the early 1990s RISC OS-based machines such as the Acorn A4000, A5000 and RISC PC, it was clear that time was running out for the company as a British computer manufacturer.

Acorn A5000
1991's A5000 was ahead of its time, but Acorn was already losing to Windows PCs

"I think by the time Acorn was capable of tapping business with RISC-based technology, the IBM PC already had a strong foothold", says Wilson. "You can't really blame one thing, but there was VisiCalc on the Apple II and Lotus 123 on the IBM PC - you had to have one of those two programs to run a business. If we'd written an equivalent program for BBC machines, it would never be the same program."

Wilson's recent work

Wilson now works at semiconductor giant Broadcom, working on a processor line she also created - FirePath, a DSL chip that has also had a major impact. "If you have a DSL line going into your house, the kit at the other end that sends you the data is run by a FirePath processor. Hundreds of millions of them have been shipped," says Wilson.

Wilson today [Image credit: Broadcom]

She takes up the story: "In 1990 I started playing with new ideas for a processor inside Acorn. And obviously lots of other things were going on. I wrote the RISC OS multimedia subsystem Acorn Replay, so what with that and launching Acorn's Online Media division and designing the SA1500/1501 digital media processor, there wasn't a lot of time for my little experiments.

"Anyway sometime in late 1996, John Redford [now head of UK engineering for Broadcom] found out what I was doing." The pair founded a new company - Element 14.

If Element 14 sounds familiar to you, it isn't the company of the same name behind the Raspberry Pi.

Wilson's Element 14 was spun out of Acorn in 1999 and clearly Broadcom knew the potential - Element 14 was sold just under two years after founding for a huge £366 million (US$356 million, AU$607 million).

Wilson on tablets - and Windows 8

Finishing up our chat, we ask Wilson what devices she uses on a day-to-day basis - but the result was some surprisingly forthright opinions on Microsoft's operating system woes.

"I use whatever does the job. I have an iPad, an Android powered Sony Xperia phone with Ice Cream Sandwich. I also have machines running Windows XP and Windows 7." Has she tried Windows 8 yet? "I have tried Windows 8. I have machines running Windows 7…"

"For machines without a touchscreen [Windows 8 is] a disaster. But they [Microsoft] have a history of violating usability guidelines. There's a whole subsection of the computer community for the usability of computer interfaces and they know precisely what makes things good, and Microsoft just ignore them.

"The [Microsoft Office] ribbon in particular is crazy from a usability viewpoint. One nice thing that's happened with the ribbon is that Microsoft have gradually been reliant on the right-click pop-up menus that we had in RISC OS, and that is straight out of the usability manuals.

"You don't have to do a great deal of research to develop that stuff, you just have to read a usability manual. It says you [need to travel] the shortest possible distance to [do something].

Acorn RISC OS machines had a three button mouse with a middle "menu" button, so we asked Wilson if usability was the key driver behind this. "Yes, so we said right, we'll dedicate a button to it. Press the button and you get a context-sensitive context menu. The more control and non-modality of your interfaces, that meant more input buttons.

"Xerox had used a three button mouse before us, so we developed RISC OS around a three button mouse. It gives you more actions. There was a massive amount of acceleration from that, and the fact the system was so fast. One button [as Apple used] introduces a lot of modality into your interface."

Sophie Wilson
Sophie Wilson giving a presentation in 2009 [Image credit Jussi Mononen from Flickr]

As one of the key figures behind the chips inside them, we ask Wilson whether she feels tablets can replace PCs as devices for content creation as well as content consumption. Again, she puts forward some strong views and, interestingly, highlights Microsoft's key problem with trying to break into the tablet market: "It's the person who's creating, not the device."

"You may wish that a tablet was better at some things but there are many excellent [apps] for them. They're so cheap compared with a computer, so light, so easy to use.

"I don't think that popularity is going to go away no matter what the Windows team does - their prime problem is to produce hardware and software that persuades business to move away from Windows XP and Windows 7. Windows 8 has a long way to go."

Now check out 30 years on, the Spectrum's DNA is everywhere in tech


Source : http://www.techradar.com/news/computing/from-a-small-acorn-to-37-billion-chips-arm-s-ascent-to-tech-superpower-1167034

0 comments:

Post a Comment