The remarkable story of ARM
The story of how ARM took over the world without anyone realising, is an amazing tale. The fact that most people don’t even know who or what ARM is, is surprising, maybe a bit depressing.
It’s a story of a tiny group of geeky engineers in a small company in Cambridge in the UK doing the impossible, taking on the largest silicon-valley computer chip designers, and over a period of 35 years, slowly, by stealth, becoming the world’s dominant microprocessor.
I knew the designers of the original ARM processor. It was created by a ridiculously small group of ridiculously talented engineers working for Acorn computers in the 1980. The same team that created the BBC Micro, (that made me very rich, but that’s another story) were trying to develop the next generation personal computer. They wanted a new processor — the core of any computer — and couldn’t find anything that was suitable, cheap enough, fast enough.
So they decided they’d design one themselves. Even back then processors were typically complex products usually designed by huge teams in huge corporations such as Intel. So the idea that a handful of engineers with no experience of designing processors could pull this off was, at the time, a bit daft.
The processor for the BBC Micro, the 6502, was about as simple as it gets. The BBC Micro was an 8-bit computer which meant it could only read and write single bytes (characters) at time. By the mid-80’s the mainstream IBM PC was a 16-bit computer using the Intel 8086.
Simple, clean and fast.
The ARM processor design skipped the whole 16-bit generation and was a super-clean 32-bit design. Simple, elegant and very, very fast.
By comparison, the Intel processors were (and still are) monsters of evolved complexity, with hacks to help address increased amounts of memory (paged memory handling was a nightmare for programmers). Writing assembly language for Intel was a serious pain, writing assembly language for ARM was a delight, and easy.
A consequence, a side-effect of this simple design, that was to become fortuitous was that it was very low-power. I’m not talking ‘computing power’ which was high, but current consumption power — Watts. That wasn’t a particularly important consideration in the day, but turned out, a decade or so later, to be the critical difference that propelled ARM to world dominance. Why? Mobile phones.
There’s one other critical point that made ARM successful. Acorn split the processor design side of the business into a separate company and found a brilliant CEO to run the company (Sir Robin Saxby). At the time Acorn computers was being run by salesmen and accountants, who clearly didn’t really understand what they had on their hands. They misunderstood how technology companies should be run, and IMO, lacked vision. Acorn failed. ARM succeeded, and I put this down to the fact it was headed by an engineer who happened to also be a great marketeer.
At this time I had created the largest software house for Acorn computers, Computer Concepts, later renamed Xara. I was also the publisher of the very first book on programming the ARM processor, and I personally wrote a ton of ARM assembly language programs that made me a lot of money.
I just discovered that the author of the ARM programming book, Peter Cockerell, has put the book online — so here it is the very first book on ARM programming:
Arm Assembly Language Programming.
Part of the reason ARM was split into a separate company was Apple. They were developing their Newton PDA (remember them — Personal Digital Assistant — a forerunner of the modern smart phone in some ways). This was a hand-held, battery powered device that required a high performance, low power processor, and the Acorn’s ARM processor was it. But Acorn were a competitor of Apple and preferred not to buy processors from a competitor. So together Acorn and Apple created this new company, to be called ARM.
Originally ARM stood for Acorn Risc Machine, but later when it was split into a new company, they changed the meaning to be Advanced Risc Machine.
For the small team of designers, it was a case of being in the right place at the right time. There was university research being done at the time on the concept of RISC processor design, simple, but fast. And the ability to more easily create custom chip designs was just becoming a thing. Finally, the unusual imagination of the Acorn founders (particularly Herman Hauser) to back the idea of creating their own processor.
It’s usually taken that RISC stands for Reduced Instruction Set Computer. But here’s a video of Sophie Wilson, (@9min) the co-designer of the ARM when at Acorn, where she states it stands for Reduced Instruction Set Complexity. She should know being the designer of the world’s most popular RISC chip.
The Apple Newton was a failure, and it was finally cancelled by Steve Jobs after his return to Apple. But the experience of using ARM meant this was the processor they wanted to use for the upcoming iPads and iPhones.
Little known fact — the iPad development predated the iPhone despite the fact the iPad didn’t appear on the market until three years after the phone.
A key differentiator between ARM and some other chip companies such as Intel, is that ARM actually manufacture nothing. ARM are what are called Fabless chip design company. They license their designs to other companies, such as Apple and Samsung, who then incorporate the ARM design into their own chips. Apple doesn’t in fact manufacture chips either, despite the fact the recent announcement was all about ‘Apple silicon’ Apple use chip a Taiwanese chip manufacturing company TSMC.
This was a critical part of their success — licensing the processor design to anyone who wanted to make their own versions of the chip.
The happy accident that ARM processors had the best MIPS/Watt (highest performance for the lowest power) is the key thing that propelled their processor to success in the exploding world of smart phones. Of course now ARM are focused almost entirely on that MIPS/Watt figure because that’s what customers are demanding — longer battery lives for their laptops, phones and watches.
So where is ARM now?
Around 140 billion ARM processors have been manufactured, and it’s estimated this is growing between 10 and 20 billion per year. Considering the world’s population is 7 billion this is an astounding figure.
Just this week (June 2020) Apple confirmed they are moving their entire line of computers away from Intel to ARM processors. They stated that, going forward, the ARM architecture was the only way they can create the high performance computers they want. Presumably they mean ‘at a competitive cost’.
Coincidentally, at about the same time as Apple made their announcement, a new ‘world’s most powerful super-computer’ was announced. Guess what? It’s running on ARM processors. See this.
It’s been an incredible, decades-long, journey that is not finished yet. ARM has taken over the world, shipping more processors than any other type. Covering entire spectrum of devices from the smallest, lowest powered IoT devices, to just about all mobile phones and tablets, to mainstream laptops and desktop computers, and finally to the world’s most powerful computer.
It shows no sign of slowing down. If anything the ARM takeover is accelerating.
I wonder what Steve Furber and Sophie Wilson, the principle designers of the ARM architecture, think of their legacy? They must be astounded. If you had told me when I started programming ARM code in the 1980s, or when I published that first ARM programming book, that in 35 years, this processor would dominate the world, I would have been incredulous.
It slightly bugs me that so few people know this British success story. I’ve seen a number of commentators refer to ARM as a Japanese company (it’s true the ARM’s parent company is now Japanese). But it was and still is a British company, probably the most successful British technology on a world-stage. I know Brits are shy about ‘blowing their own trumpet’, and I know the press seem only interested in spreading bad news (pet peeve), but still, more people should know the story.
