Chapter 1

The History of ARM

WHAT’S IN THIS CHAPTER?

• The beginnings of Acorn
• How Acorn became ARM
• ARM naming conventions
• ARM processor architecture

In the late 1970s, the computer industry was going through a period of huge change and substantial growth. Up until that time, computers were the size of a room and took dozens of people to operate. The ENIAC weighed 30 tons, took up a surface of 1,800 square feet (167 square meters), and required 150 kilowatts of energy. Huge technological advances were being made, and people started talking about the possibility of creating a computer that didn’t weigh more than 2 tons. Then, the transistor revolution happened. Suddenly, all the power held in a room full of electronics (some would say electrics) could be put into a single microchip, just a few centimeters squared. Businesses could afford to have multiple systems; schools bought computers for research and education; and even families could finally enjoy personal computing. The 1970s saw the birth of some of the major players in computers: Apple Computer, Atari, Commodore, and Acorn, to name but a few. It would take only a few years to see some of the mythical names in personal computing: The Amiga, the Atari ST, and the Commodore 64, for example — gift-giving hasn’t been the same since.

THE ORIGIN OF ARM

In late 1978, Hermann Hauser and Chris Curry founded Acorn Computers, Ltd. in Cambridge, UK. Initially working as a consultancy group, Hauser and Curry obtained a contract from Ace Coin Equipment to develop a microprocessor-based fruit machine. After an initial research and development phase, Hauser and Curry chose a MOS Technology 6502 processor for the job. The 6502 was produced in 1974 and was one of the most reliable processors available at that time, which would soon revolutionize the computer industry.

The 6502 was an 8-bit microprocessor that was by far the cheapest microprocessor at the time, while still remaining comparable to competing designs. The 6502 was easy to program, and its overall speed was good. The 6502 was also well known for its low interrupt latency, making it good at handling interrupt-driven events. The 6502 had a simplistic design; it had “only” 3,510 transistors. (Intel’s 8085 had 6,500, the Zilog Z80 had 8,500, and the Motorola 6800 had 4,100.) It also had comparatively few registers, requiring frequent RAM access.

The 6502 went on to power some of the most famous names in computing history: the Apple II series, the Atari 2600, and the Commodore VIC-20, among others.

Acorn Computers won multiple contracts and continued to use the 6502 for its projects, capitalizing on its knowledge of this microprocessor.

With its experience with the 6502, Acorn went on to develop the Acorn System 1. It was designed by Sophie Wilson and was based on the 6502. It was a small system (for the time) and was launched into the semi-professional market. With a $96.43 (£65) price tag, it was accessible not only to professionals in laboratories, but also to enthusiasts. It consisted of two simple boards: the top one holding a hex keypad and a small seven-segment LED display standing on a ribbon cable, the bottom one carrying the processor and support circuitry. The simple ROM monitor enabled memory to be edited and software to be stored on cassette tapes. Most were sold as self-assembly kits. With each success, Acorn launched a new System, up to the System 5 in 1983, still featuring a 6502, but rack-mounted, including a disk controller, video card, and RAM adapters.

Acorn then released the Acorn Atom — a personal computer that still used the 6502. By now, Acorn had an excellent knowledge of the 6502, allowing it to push the processor to its limits and sometimes even beyond. Acorn’s experience with the 6502 was legendary. However, the 6502 was becoming an old processor, and technological advancements meant that faster processors were available.

From here, Acorn shifted slightly; internal discussions debated on which market to pursue. Curry wanted to target the consumer market, while many other factions did not believe in the consumer market and wanted to target a more professional target, namely laboratories. Several research projects were launched, including the possibility of a 16-bit computer, but Hauser suggested a compromise, an improved 6502-based machine with more expansion capabilities: the Proton.

At the time, the British Broadcasting Corporation, the BBC, took an interest in the microcomputer market and began the BBC Computer Literacy Project. Television episodes on microelectronics and the predicted impact of computers in industry were designed to be made available to students, who had no time to design their own computer system. The BBC wanted a computer to go with their television series and started to look for candidate systems.

The BBC had a long list of subjects that it wanted to demonstrate in its series; graphics capabilities, sound and music, teletext, external I/O, and networking were all requirements. Several companies competed for the contract, and the Proton project was an ideal candidate. The only problem was the Proton didn’t actually exist. It was only in the design stage; it wasn’t prototyped. Acorn had little time, only 4 days, and spent those 4 days working night and day, prototyping the design, and getting the Proton ready to show to the BBC Finally, only hours before the meeting with the BBC, the Proton was ready. All the hard work done during that week paid off; not only was the Proton the only machine to meet the BBC’s specifications, it also exceeded them. Acorn was awarded the contact, and the project’s name was changed. The BBC Micro was born.

The BBC Micro sold so well that Acorn’s profits rose from a little more than $4,800 (£3000) to more than $13.6 million (£8.5 million) in July 1983. Acorn anticipated the total sales of the BBC Micro to be approximately 12,000 units, but after 12 years, more than 1.5 million units were sold. By 1984, Acorn claimed sales of 85 percent of the computers in British schools, with deliveries of 40,000 units a month. The BBC Micro was extremely well designed for its use; a robust keyboard that could withstand anything children could throw at it (literally), a carefully designed interface, and the right machine at exactly the right time.

Why Acorn Decided to Create a New Processor

Acorn was now faced with a major problem; almost all its projects had been done on a 6502, so it knew the hardware well. When seeking a new processor to replace the aging 6502, it found that other processors just weren’t up to the job. Graphics systems were emerging, and it was clear that the 6502 couldn’t keep up in the graphics field. The Motorola 68000 was a 16/32-bit microprocessor that was used in many family and business computers, but slow interrupt response times meant that it couldn’t keep up with a communication protocol that the 6502 had no problem running.. Numerous processors were studied and excluded. One by one, the processors on the market were studied, analyzed, and rejected. Finally, the list ran out, and Acorn was left with no choice; if it wanted to do things its way, it had to make its own processor.

Creating processors wasn’t necessarily something new; it was the golden age of multipurpose CPUs, and several companies were designing CPUs, using little more than transparent film and pens, but what Acorn was about to do went well beyond simply designing a new CPU with 4000 transistors; ARM set out to create a 32-bit processor.

The project was started in October 1983, with Sophie Wilson designing the instruction set and Steve Furber doing the hardware design using BBC Micros to model and develop the chip, and on April 26, 1985, the first Acorn RISC Machine processor was born, the ARM1. It also worked perfectly the first time, which was rather exceptional for a processor that was basically designed by hand.

The primary application of the ARM1 was to be a coprocessor on BBC Micros and to help create the ARM2. Subsequent chips were sold as specialized coprocessors, or as development boards linked onto BBC Masters. It was in 1987 when the first ARM-based computer was sold, the Acorn Archimedes.

Why Acorn Became ARM

With the education market starting to fall off, Acorn’s priority was to open new markets and to promote its processor design. VLSI, Acorn’s partner, had been tasked with finding new applications for ARM processors, and Hauser had a separate company, Active Book, that was developing a mobile system based on an ARM2 CPU. Apple Computer’s Advanced Technology Group (ATG) contacted Acorn and started to study ARM processors. Apple’s ATG objective was to create an Apple II-like computer, but the project was abandoned for fear of creating too much confusion with Macintosh systems. However, the final report stated that ARM processors were extremely interesting — both their initial design and their power usage and processing power ratio.

Later, Apple Computers again studied the ARM processor. Apple had set strict requirements for its Newton project; requiring a processor that had specific power consumption, performance, and cost, but also a processor that could be completely stopped at any given moment by freezing the system clock. Acorn’s design came closest to Apple’s requirements but didn’t quite fill them. A number of changes were required, but Acorn lacked the resources necessary to make the changes. Apple helped Acorn develop the missing requirements, and after a short collaboration, it was decided that the best move would be to create a new,...