In November 2020, Apple unveiled its new M1 processors, powering its latest laptop computers. The initial reception was positive — Apple is pretty good at these product launches — but what followed in the weeks and months since has been incredible. Benchmark after benchmark have shown just how great a leap forwards the new Apple silicon really is:
“The new MacBook Air has better single-core performance than every Intel Mac that exists. Not just every Intel MacBook Air, but every Intel-based Mac”
“In this first-time view of the popular Cinema4D based benchmark, we see the Apple M1 toe-to-toe with the best-performing x86 CPUs on the market, vastly outperforming past Apple iterations of Intel silicon.”
It is not only the groundbreaking performance of the chip that stands out. It also runs so cool that it doesn’t even have a cooling fan, and its battery life gives it an incredible 14-16 hours of use.
How is all of this possible? There isn’t one answer to this: a lot of the credit must be given to Apple’s ingenuity with microprocessor design and vertical integration of its hardware and software, but the foundation the Apple M1 chip is built on is the famous “ARM” architecture.
Simply put, ARM is fundamentally different from architectures such as Intel’s x86: it has a much smaller set of instructions (an instruction is a fundamental computing function like “add these two numbers together” or “jump to this location in the code”), and this allows it to operate more efficiently. The industry term for this is “Reduced Instruction Set Computing”, or RISC for short. RISC isn’t a new computing paradigm – it’s been around for more than 35 years – but it’s only since the advent of mobile computing that it’s really started to become a big competitor to other chip architectures.
So what’s the big deal about ARM?
Apple laptops using ARM might seem like just a nice incremental improvement in performance, efficiency and cost that benefits consumers the same way other hardware improvements do. Every year, everything gets a bit faster, cheaper, or both. But this new product cycle actually represents a potential tipping point. If all the software developers using Macbooks (many/most of them, depending on which survey you read, with a population in the tens of millions) upgrade to M1, then this will have a knock-on effect on the server computing market: the whole software ecosystem needs to support ARM architecture, and it pushes public cloud providers like AWS and Azure to support ARM as a first class architecture in their cloud computing offerings.
Actually, if you dig a little, you’ll find that this has already started to happen. AWS, always happy to be ahead of the curve and relentless with its product development, offers a custom cloud computing instance type called “Graviton“:
AWS Graviton processors are custom built by Amazon Web Services using 64-bit Arm Neoverse cores to deliver the best price performance for your cloud workloads running in Amazon EC2. Amazon EC2 provides the broadest and deepest portfolio of compute instances, including many that are powered by latest-generation Intel and AMD processors. AWS Graviton processors add even more choice to help customers optimize performance and cost for their workloads.
Here, we see a crystal clear translation of ARM’s superior performance and efficiency to “Best price performance for your cloud workloads” in AWS. All those companies running their services in AWS will be able to run it for cheaper on ARM computers, in turn encouraging Amazon to offer more and more of them, passing the cost savings down.
But what does this mean for the x86 titans like Intel and AMD?
Well, AMD is rumored to be working on an Arm-based Apple M1 rival – in fact it had a kind of ARM prototype called the K12 core. Perhaps AMD are starting to see which way the wind is blowing.
And Intel? It looks like Intel might be in some serious trouble strategically. Apple dropping them for its personal computing products is one thing; AMD starting to beat it in the desktop and server markets is another; being years behind with its semiconductor manufacturing technology is a third; but an ascendant architecture that is both fundamentally different and at the same time, better, than x86? That seems like a truly existential threat to the US chipmaker’s survival. At the moment, it looks like ARM’s rise is unstoppable, and every chipmaker that wants to remain competitive needs to adopt it — or something similar to the RISC paradigm — or be pushed out.
What does this mean for professionals in IT?
At the moment, it probably just means you should buy an M1 Macbook and enjoy working with a laptop that never gets hot in your lap, even when processing heavy workloads, and has a battery life that’ll last longer than an international plane flight (remember those?!). If you work in the area of DevOps or cloud infrastructure — particularly if your company trusts you to do things like “manage and reduce our cloud spend” — you should give the AWS Graviton instances a serious look; if you’re compute-heavy there might be significant cost savings out there for a modest infrastructure migration investment.
If you work on commercial software tools, compilers, or other native software then you’re probably already well aware that you need to make sure everything works properly on ARM. It’ll be a continuous investment going forwards, because we’re only going to see more computers with ARM chips in the future.
And what about NVIDIA buying ARM? A lot of people (including corporations and the governments of quite a few countries) have quite significant and understandable concerns about this acquisition, and there’s a significant chance it will be blocked. Even if it goes through, the chances are high that NVIDIA won’t be allowed to abuse its ownership of ARM to stifle the many licensees. In other words — the future looks like it’ll be an ARM one, irregardless of the fate of the company that owns the IP.