If you’ve paid any attention to smartphones and tablets you’ve likely heard of the term “ARM” used to refer to the hardware inside. It’s thrown around left and right, often as a point of differentiation from laptops and desktops, which use Intel x86.
But what is ARM, exactly? A processor architecture? A brand of hardware? Or something else? That’s rarely explained. Let’s cut through the confusion and get down to the basics.
The Key To ARM Is RISC
ARM is an acronym which stands for Advanced RISC Machine. In the name is another acronym (yes, nested acronyms!) which stands for Reduced Instruction Set Computing.
RISC is, in its broadest form, a design philosophy for processors. It stems from a belief that a processor with a relatively simple instruction set will be more efficient than one which is more complex. The term originally came into use back in the 1980s with a research project called Berkeley RISC that investigated the possibilities of this approach to design and then created processors based on it.
All ARM processors are considered RISC designs, but this doesn’t mean much because RISC itself is simply an approach to design rather than a technological standard or processor architecture. Still, a basic understanding of RISC properly frames ARM.
ARM refers to itself as an architecture, which can cause a misunderstanding when compared to Intel. Intel gives every new chip design its own unique code and talks about each as a new architecture – even when there’s often many similarities and they all use the same instruction set (x86). ARM, on the other hand, treats its designs as an unbroken family. Updates are still a part of the ARM architecture. They’re just given a new version number.
The trait that’s most relevant to consumers is not the micro-architecture (the physical design of the chip) but instead the instruction set. The instruction set is the basic set of capabilities and features a processor makes available to software. It determines what arithmetic can be used, how cache should be allocated and the order in which instructions should be executed. Software designed for one instruction set can’t be used on another unless it’s revised.
Micro-architectures and instruction sets can’t be separated because the architecture is a physical expression of the instruction set. This is why ARM-based processors tend to be small, efficient and relatively slow. The simple instruction set requires a small, simple design with fewer transistors. Transistors consume power and increase die size (which increases production cost), so having as few as possible is ideal when selecting a processor for a smartphone or tablet.
ARM’s Business Is Different
Talking about ARM processors as a whole can be difficult because there are so many available and their performance varies. It’s counter-intuitive. How can Apple have ARM processors that are quicker than the competition when it’s using the same architecture?
This happens because of how ARM Holdings, the company that is responsible for ARM, does business. ARM Holdings is only a design company. They manage the instruction set and design new versions of the core architecture and then license it to other companies. Those companies can then improve it and pair it with whatever hardware seems appropriate.
It helps to understand that ARM’s core architecture is only a processor. It doesn’t handle wireless connectivity. It doesn’t handle graphics. It doesn’t handle USB or other forms of wired connectivity. All of that is the responsibility of other hardware licensees pair the architecture.
That’s why there are so many variants of ARM on the market and why they perform differently. Apple has an entire in-house engineering staff that works on its ARM processors. Other companies, like Qualcomm and Texas Instruments, act as middle-men. They take the ARM architecture, pair it with a variety of hardware, and then re-sell it as a “system-on-a-chip” for smartphones and tablets.
What Does ARM Mean For Consumers
To a consumer, ARM can be thought of as an ecosystem. Software designed for ARM will only work on ARM. Windows RT apps, for example, don’t work on a PC with Windows 8. Modifications must be made to a program to jump from ARM to x86.
Operating systems that work on one ARM device should work on others. This is why there are so many Android modifications and why Android can potentially be loaded onto tablets from HP and BlackBerry. Apple does mess with the ecosystem a bit, however, because the source code of iOS is not available. Attempting to port iOS to other ARM devices is almost impossible without it.
ARM also means lower power draw and lower performance relative to x86. This is not set in stone, however, because both architectures are changing over time. Intel is working hard to create versions of its processors with extremely low power draw. And ARM Holdings is working hard to improve the performance of its designs.
Will ARM Be In Your PC?
There have been a few attempts to sell traditional devices with ARM processors that work like traditional PCs. Motorola sold a keyboard dock for the Atrix smartphone and advertised it as a laptop replacement. ASUS sells a line of Android tablets with keyboard docks. And Samsung now sells a Chromebook that runs on ARM.
Such attempts foreshadow a storm of potential surrounding ARM, yet predicting the storm’s path and intensity is impossible. Current ARM architectures are significantly behind the performance Intel’s slowest processors (nevermind its mainstream line of Core processors). Nvidia says that it is working on a processor using ARM’s architecture that will compete with Intel, but it’s not clear how this is being accomplished or when a finished product might be released.
For now the potential threat of ARM remains a dark cloud on the horizon of the potential PC space. It looks threatening, but a concrete threat has yet to manifest. Is ARM a fearsome storm or simply a few shadowed clouds that will eventually disperse? That remains to be seen.