By Richard Sharpe
In the last week of April, 37 years ago in 1985, Acorn Computers received the first samples of a new microprocessor it had designed for its next range of micros. When they tested the samples that week they found a surprising result: the microprocessor worked on hardly any power. It certainly did not need mains power: it could work with a battery. It became the processor design of choice for the soon to explode mobile phone market and other mobile devices.
Hermann Hauser set the ball rolling at Acorn. Acorn could not find a microprocessor suitable to power its next generation of pc after reviewing the options mostly from US makers. Hauser decided that Acorn should go it alone using a new type of processor: a reduced instruction set computer. They called it Acorn then Advanced RISC Machines.
“I gave two advantages to our design team that neither Intel nor National Semiconductor or Motorola ever managed to give to their design teams, and the first advantage was, I gave them no people, so it’s the only microprocessor in the world that was designed by just two people, Steve Furber and Sophie Wilson; and the second advantage I gave them was, no money, because we didn’t have any. So rather than 50 people burning through many millions of pounds, we had to keep things incredibly simple and cheap.
“It was when Stephen tried to measure the power consumption of the first ARM that he realised that the power pins were actually not connected. So, we thought, well that’s rather good, you know, it’s one that, we’ve got a processor here that doesn’t need any power at all. And as you say, the secret for why the, the thing actually worked quite well was. the leakage currents through all the, all the other pins was enough to power up the chip. So we, we knew that we were onto a winner on the power front.”
Steve Furber describes the division of labour between him and Sophie Wilson: “Sophie did the instruction set architecture design; I did the micro-architecture. There were other factors. I mean, Acorn had decided strategically, I think this was advice from Andy Hopper at the Computer Lab, to invest in silicon design tools, and silicon designers. So we had identified what we thought was the best silicon design software from VLSI Technology in California. We employed some experienced chip designers. But we didn’t really have anything for them to do. There were some chips, there was the second generation of the electron chip, and, and one or two other things, but, basically, we had this capability, this capacity that was under-utilised. And so when, when I sketched a micro-architecture, they were there waiting, and they took the micro-architecture specs and, and did the silicon implementation. And, and within eighteen months we had the first working ARM chip… Apple came. They wanted the joint venture. Pushing on open door. And, by November that year ARM Limited was set up in an old barn in Swaffham Bulbeck, and, the processor was now in independent hands, Robin Saxby was brought in. He introduced this business model that made it all work.”
Sir Robin Saxby was brought into the joint venture to provide some proven executive experience: “They rang me up and said, Acorn were thinking of spinning out the twelve-man design team into a new company. Apple were thinking of investing. VLSI were thinking of investing. But Apple wouldn’t put the money on the table unless they had found a chief executive. And I was a name, so I went and had a chat. And my chat was, how are we going to make this company work? And one of the jokes, I said, ‘We’ll make chips over my dead body.’ And one of the other jokes is, ‘I’m still alive, and ARM doesn’t make chips.’ …When I analysed it with the engineers, we realised the big benefit was the lower power consumption. Another big benefit was the low cost. And the first company brochure I did, I said, we’re going to be the best MIPS per watt, that’s millions of instructions per second per watt, and the best MIPS per dollar, millions of instructions per second per dollar.”
ARM has struck to the mantra of not making silicon. It licences out its design, provides tools for implementation and collects royalties. No expensive fab plants, no competition with possible customers: pure intellectual property.
Simon Segars, then head of ARM when interviewed for the Archive in 2019, said: “Having a chip with a processor in it that can, you know, run software and utilise all the other features of the chip, turned out by that point to have been a universally good idea, and because by that point so many chip companies already had an ARM processor in their kind of portfolio of IP, and had engineers who had experience of using it, it was just the natural technology choice. And the pricing of our product was such that, it was, it was economically an easy decision to make as well. So we found ourselves being designed into things that we had had no visibility of. Because we had done a license for mobiles, say, and suddenly you find you’re in the anti-lock brake system of a car, having not done anything to particularly make that happen. So, so we found just, all these design wins were happening, and of course we were building a, a field-based team to help on some of that, and, build our ecosystem to, you know, provide some kind of specialist software if there was some market we were going after. But, but the fact that we had done all of this licensing off the back of mobile, suddenly meant we were getting exposed to all these other markets. And, and so just, the, the momentum was building. And with any processor, the more software that’s written for it, the stickier it becomes. And so, you know, suddenly, the barrier to entry for somebody else coming into this space was, was very high.”
And so it remains.