In August 2019 Arm Education Media will be publishing our newest title, System-on-Chip Design with Arm Cortex-M Processors, written by Joseph Yiu, Distinguished Engineer, Embedded Technology, at Arm.
This is a new line of products from Arm Education Media developed specifically to suit the needs of researchers and professional engineers.
Arm Education was set up specifically to address the education and skills gap in computing and STEM with a lifelong learning approach. This approach is crucial since many of the education and skills gaps happen at the interfaces between different stages of education and learning, for example, from school to university, and from university to the workplace.
The textbooks currently published by Arm Education Media focus on undergraduate student learning, combining theory and practical applications that use state-of-the-art technologies from the Arm ecosystem.
Our first reference book aims to expand upon this learning journey, targeting researchers, engineers that are transitioning form university to the workforce, or existing practitioners looking to engage with Arm technology for the first time.
In anticipation of the title’s release, we’d like to publish an extract of the foreword, written by Arm’s Director of Business Innovation Strategy, Mike Eftimakis.
Mike provides the reader with an excellent summary of why understanding SoC Design is so important in today’s connected world. Along the way, readers will also discover how Arm came into existence, and why Arm is at the heart the exciting revolution that’s taking place in computing today.
Right now, you are probably surrounded by Arm processors without even knowing they are there. More than 100 billion chips containing an Arm processor have been produced up to now – this is 13 for every human on the planet.
The most surprising thing is that Arm does not produce chips. It just designs the technology and enables its partners to manufacture differentiated devices that integrate them.
Many more of those chips, also called SoCs (system-on-chip), are expected to be produced in the coming years. We even start talking about trillions of devices for the Internet of Things (IoT). Of the total number of SoCs currently out in the market, the great majority use the smallest processors in the Arm product range: the Cortex-M series. Small, very energy efficient and powerful enough for many applications, they are at the heart of many of today’s electronic devices.
This book is here to explain how SoCs based on the Arm Cortex-M processor portfolio cores are designed, detail the different elements that compose such a system, explain the different design issues, describe the integration into systems, and discuss how these SoCs are programmed.
In the 1980s, Acorn, a British company, became very successful with the BBC Micro-computer, which was used in many schools throughout the country. For its future generation computers, the company wanted an updated processor and started a quest for such a component. Unfortunately, none of the available microprocessors were suitable for its needs. Most of them were either too complex or not available and required a large number of external components. The Acorn team then learned about the Reduced Instruction Set Computer (RISC) concept and found it could lead to powerful, yet low-cost, solutions.
At the time, RISC processors were confined to high-end computers, where cost was less of an issue, since no existing RISC processors were exactly suitable. That led the team to embark on the journey to develop their own piece of silicon.
This secret project was named “Acorn RISC Machine” (ARM, in short). The first processor, ARM1, was launched in 1985. It was produced by VLSI Technology in a 3μm technology (almost 500 times larger than the most advanced designs now) and could run at 6 MHz. One of the side-benefits of this simple processor architecture was its lower power consumption (compared to contemporaneous CPUs), which allowed the component to use a lower-cost plastic package without melting it.
At the heart of the processor design was the Arm instruction set, which progressively evolved to optimize the performance and efficiency of new generations of processors. This is a key element of what is called the ‘architecture.’
The Arm processors powered several models of Acorn computers, but a major change happened when VLSI Technology, which was manufacturing the components in its factories, signed an agreement with Acorn to re-sell the chips to other companies. This was the first ‘Arm license.’
In 1990, after discussions with Apple Computer, who needed a new processor for the Newton project, Acorn decided to spin-off its processor division and form a joint venture with Apple and VLSI Technology. The team then changed the meaning of Arm to ‘Advanced RISC Machines’, which became Arm Ltd later on.
This evolution came at the same time as a great change in the new company’s business model. On the one hand, Arm had unique assets: great expertise in processor design and an original architecture. However, producing chips required caring about fabrication, yield, quality, logistics, sales channels, complex application-specific marketing, or any other tasks that a silicon manufacturer should do to be successful. This was not optimal.
On the other hand, silicon manufacturers had a hard time staying competitive, because they had to excel at these activities while simultaneously investing in design and innovation around processors, at an increasingly fast pace. This was not great either.
The revolutionary idea for the newly-formed company was to become a specialist in R&D and focus on the processor design only. Instead of selling components, Arm would license ‘Intellectual Property’ (IP in short) to semiconductor manufacturers, who would then use this IP to design their chips, in combination with other elements that would be more application-specific.
The IP model selected from the start by Arm required a very tight relationship with the other companies using the IP. As the company did not manufacture products, its success was entirely
dependent on the success of chip manufacturers embedding the Arm IP into their chips. Conversely, to make sure that they always get the best performance and efficiency for their products, silicon manufacturers had to make sure that the success of their products also benefited Arm, so that part of the increasing revenues would be invested in improved and competitive IP. Together, Arm and partners solidified the symbiosis using a royalty-based model: Arm revenues were largely dependent on the success of the chips containing its IP. This resulted in a strong partnership between the company and its customers, and a great sign of this very special relationship is that customers were called ‘partners’ (This is still the case more than 25 years after the foundation of the company).
Another great benefit from these partnerships was that each semiconductor ‘partner’ could focus on a different set of applications, on different market segments, and integrate its own expertise and ‘secret sauce’ into the design of their products. This business model allowed the creation of a rich variety of products that no single company (even the largest ones) would have been able to put into their product catalog. It also made it increasingly difficult for processor manufacturers using other architectures to compete with Arm because they had to compete with a whole ‘ecosystem.’ Many of them progressively decided to stop wasting money on processor architecture development and realized that it was much less expensive just to license state-of-the-art IP from Arm.
Arm-based processors are used in virtually all applications requiring processing capability: as the company says, “wherever computing happens.” Over the years, the company has developed a range of products that address very different needs, from the tiniest processors for embedded applications (the Arm Cortex-M processor portfolio) to the largest application processors that are used in high performance servers, or that ones that power 95% of the mobile phones in the world (the Cortex-A processor portfolio). There is more than a factor of 100 in complexity and size between the smallest and the most performant cores.
However, central processing units are not the only IP offered by Arm: a diverse range of IP has been developed or acquired by the company to address the needs of many applications. This is the case of what is called ‘System IP’: all the elements that enable processors to connect to the rest of the system, transfer or store data between those elements, manage security, enable the debug of the software, and manage power. Another very important line of products relates to media processing, and the Arm Mali series is now the world’s ‘most shipped’ commercial GPU IP.
You can download the System-on-Chip Design with Arm Cortex-M Processors reference book for free. Readers can also pay for a print on demand version of the reference book sometime in September.