How apple took the “RISC”
On Jun 22, 2020, at Apple's annual Worldwide Developers Conference (WWDC), they first unveiled the transition from Intel-made x86 chips to in-house ARM-based processors.
This wasn't Apple's first time jumping from one processor architecture to another at the 2005 Worldwide Developers Conference (WWDC) when then Apple CEO Steve Jobs announced to transition away from the use of PowerPC microprocessors supplied by Freescale (formerly Motorola) and IBM.
Before I move on further, I want to say that this isn’t a recommendation to buy any of Apple’s products.
Simply, I perceive this to be the biggest moment in computing history in the last decade or so.
It seems that everyone hates Apple these days, and there’s some rationale for that. Recently they’ve made their products impossible to repair the increased prices of their products and accessories to a ridiculous level and enjoy excessive control, but what if Apple actually did something to push the industry forward and spur competition? Well, it turns out that they may just have done that.
Want to read this story later? Save it in Journal.
Phones and computers are never seen as comparable when it comes to power. They both live in different worlds. Back in 2012, there were comparisons between the two of them and see if a phone could replace a PC.
The rate at which phone power was improving year over year was absolutely unbelievable.
The main difference between phones and desktop computers is that they use different types of processor technologies, phones use a type of processor design called “ARM” while desktop computers generally use another type called “x86” like Intel or AMD.
ARM is a simpler design than x86, making it more efficient but, x86 traditionally used to be able to do more but this is changing. The interesting thing to note here is that in recent years our phones have been getting faster at a much more rapid rate than computers have been improving.
So, here’s a question what happens when the chips in our phones come close to the power of laptop and desktop computers?
What if you took this amazing performance and efficiency of a mobile chip and put it in the body of a laptop that has access to a larger battery and more space to breathe? It's going to a revolution in the history of computing.
So I’ve recently got my hands on an Apple M1 Macbook Air;
Animations were buttery smooth as you’d expect, and even opening and closing some of those bigger apps like chrome and lightroom are just as fast if not faster than on my current Intel Macbook Pro, which is sick it’s just a really responsive feel.
The surface pro x was the windows version of this but, there was extreme lag moving and doing anything there.
But, here in Apple MacBook Air it is extremely smooth, and this is the base model MacBook air, the cheapest one you could get, and even without a fan.
The new MacBooks and the M1 chip that powers them resulted in an energy-efficient chip as powerful as some desktop CPUs. What most people are missing here is that this isn’t an ordinary MacBook the M1 chip represents the culmination of a decade of planning and a radical shift in the consumer computing industry on the surface putting a mobile chip in Apple’s best-selling computer.
It seems really stupid, but when you look deeper, it makes a whole lot of sense. As you’ll see later, it’s actually a classic case of disruption as defined by clay Christensen in 1997.
So how did Apple do this well?
To truly understand what apple has done we first must go through some quick history on central processing units or CPUs which are the brain of computers they do billions of calculations per second to make our devices run.
CPUs run on what are called instructions these are the fundamental tiny bits of data that tell our computers what to do, for example, add these two numbers together or fetch this number from memory kind of thing. When you group these commands they’re called an instruction set.
In the emerging battle between mobile and desktop CPUs, the variations in how these instruction sets were handled would make all the difference.
Intel since its founding in 1968 by Robert Noyce and Gordon Moore has been the gatekeeper of mainstream computer CPUs they would gain the name x86 after a series of popular chips in the late 70s and early 80s that ended in the number 86.
The x86 instruction set architecture would go on to dominate the industry from the 1980s chip manufacturers and would begin to add more and more complicated instructions
These instructions made the chip’s functionality bloated because the limited physical space on the chip was being taken up by fancy instructions that was hardly ever used by the rest of the system.
Meanwhile in the UK in 1983 a company by the name of acorn computers decided to go down a different path instead of making things more and more complicated they took a simpler approach.
This model of thinking was called RISC or Reduced Instruction Set Computing as opposed to the bloated CISC from companies like Intel and some people were saying that RISC technology is the wave of the future.
So, RISC gives us a new starting point we can say okay here’s a simpler instruction set faster, less code space. If you talk about RISC, it certainly is a philosophy or a style and in fact, it’s a style that differs depending on what you’re trying to do. The main idea is not to add any complexity to the machine unless it pays for itself by how frequently you would use it so, out of this came the acorn RISC machine a project also known as the ARM CPU project.
This was the birth of these type of chips that would be used in all of our smartphones in the coming decades.
While developing the first arm chip during testing something unexpected happened.
“As we were set to designing the chip we didn’t really expect to pull it off okay firstly we thought this RISC idea is so obvious that big industry will pick up on it and will get trampled underfoot we expected to go into this project finding out why it wasn’t a good idea to do it and the obstacle just never emerged from the mist we just kept moving forward through the fog until eventually, we had a fully working ARM chip in our hands”
“I don’t think it was day one but a few days later we decided we’d better measure the power consumption so I turned the arm chip on and ran some code on it and looked at the ammeter, but it was reading zero, and I knew we designed a fairly low power chip, but this was a bit remarkable and it turned out that actually, inserting the ammeter into the power supply I failed to connect the power supply and so no current was flowing through the ammeter, but the chip was still running.” — Professor Steve Furber
“So hang on, it wasn’t just you didn’t connect the ammeter. It was actually that you didn’t connect the chip to the power supply; yes, there was no power supply connection to the chip, and yet It was running.”
By 1985 they had their first prototype, and by 1987, the first computer-based on arm was produced, the Acorn Archimedes, and from here arm would license its designs to other manufacturers, being low power was a nice side effect but it didn’t have much of an advantage for desktop machines at the time but as portable computing devices started showing up arm became the first choice. hence, they were used in everything from Nokia to Ericsson phones.
So, how are these ARM chips so efficient?
When we talk about computer CPUs we measure their clock speed in gigahertz(GHz) and that means clock cycles per second. Simplified CPUs such as ARM generally will do one simple instruction per clock cycle while desktop chips may use many cycles to complete one complex instruction this means more power consumption less efficiency and more heat produced multiply this by hundreds of millions of times a second and the issue starts to become clear.
In some desktop CPUs, some of the complex instructions can take a dozen clock cycles or more to complete though contrary to popular belief both kinds of processes were flourishing.
During the 1990s in the early days, simple instruction set designs were great for engineering and graphics so much so that computers using this technology helped create films such as Jurassic Park, toy story and more.
Simple instruction CPUs were also used in the 3D graphics chips of the Nintendo 64 and other early game consoles but the desktop market safely belonged to intel.
In 1991 Apple, IBM and Motorola combined forces to create the PowerPC chip interestingly PowerPC was a desktop chip but used simple instruction sets. Apple would proudly put them in their computers during the 90s.
Apple’s use of the arm chip and the iPod was a microcosm for ARM’s popularity in portable devices from mobile phones to mp3 players by 2002.
There were 1 billion arm chips and devices despite using arms simple CPUs and their portable devices
Apple’s PowerPC collaboration with IBM and Motorola wasn’t going so well at the time their ARM CPUs were falling behind Intel.
It was looking like intel and their complex CPU way of thinking was the way to go for the desktop market so apple left PowerPC behind.
Yes, it’s true we are going to begin the transition from the PowerPC to Intel processors when intel signed the deal with Apple in 2005.
But Apple was already thinking of what comes next and they went to intel to see if they could power the upcoming iPhone now at this time, a moment occurred that intel would never forget Paul Otellini CEO of intel at the time who didn’t saw such a deal working out as it was far outside their speciality. He didn’t believe that intel could earn enough money by building mobile chips for Apple’s new iPhone So, He said revenue wasn’t going to cover the research costs. Well, he couldn’t imagine the iPhone selling well and this obviously proved to be a huge missed opportunity.
And then the iPhone launched in 2007 using a simple instruction ARM CPU.
If intel’s CEO Paul Otellini had made the deal happen they would have had a massive stake in the iPhone business.
But, unfortunately for intel Apple was moving forward without them.
In 2008 apple would sign an architecture license with ARM, so they could design their own chips from the ground up
And they would also buy P. A. Semi a company whose founder previously specialized in high-performance arm chip development and this was the key strategic acquisition for apple which gave them the expertise and potential to design some of the best arm chips and put them on the road to disrupt intel and their x86 processors.
In 2012 Apple releases their first fully custom-designed CPU and was code-named swift which was used inside the iPhone 5.
This was in contrast to using off-the-shelf designs from ARM as they had previously done this new apple built chip was two times faster than the previous chip. Even for being a first-generation chip, it had an impressive performance when compared to the mobile competition but it wasn’t head-turning. The real shock to the industry came the next year 2013 with the Apple A7 in the iPhone 5s.
Apple’s early adoption of a 64-bit architecture stunned everybody, they even beat ARMs own CPU teams such that Arm’s own 64-bit design wouldn’t be seen until late 2014 in the form of the galaxy S4 Apple famously called their 2013 chip design a “desktop-class architecture” most people thought it was just apple bragging. But this wording actually was a hidden clue as to where the company was heading as you’ll soon see their latest custom chip does actually hold up with desktop pcs.
By 2014 there were 50 billion arm chips in the world. Apple kept selling a bunch of iPhones and making a heap of money and do you know what all of this revenue you can buy? A ton of research and development.
The chips inside of the iPhone kept evolving, improving and getting faster while using less power. By the end of 2018 with Apple’s A12 Bionic chip, it was looking like apple was catching up with intel.
In this chart the grey dots are Apple and the blue squares are Intel in the past five years.
Intel managed to increase its flagship performance by about 28% and at the same time, Apple managed to improve their performance by 300%.
This is where things start getting very strange the intel chip at the top of this chart the core i9 10900K which is a desktop chip that uses 125 watts of power the kind that you have to plug into a wall.
Meanwhile, apple’s A14 Bionic chip in the iPhone 12 the one in the top right-hand corner performs better than the Core i9 10900K chip I just mentioned and uses only 5 watts powered by a phone battery.
YouTuber Jonathan Morrison does a test of exporting a heavy-duty video file on a desktop versus an iPhone 12.
“Check this out this is a fairly capable machine it’s a 10 core it’s got a beefy GPU and what I have here is a sony fx9 hdr 10-bit footage…”
— Jonathan Morrison
The M1 chip is even faster than the A14 Bionic and Apple’s chip performance was accelerating.
Apple pretty much was forced to ditch Intel and their x86 platform. Intel was just too slow to innovate they were struggling to keep up with the manufacturing technology.
Whereas apple could outsource manufacturing of their custom ARM designs to others like Samsung or TSMC.
The performance lead of intel’s desktop CPUs was being evaporated right in front of everyone’s eyes.
By 2019 the iPad was more powerful than the Intel MacBook Pro.
And now the 2021 iPad Pro got the M1 chip inside which is a 50% CPU and 40% GPU performance bump comparing to its predecessor.
“You don't see these kinda upgrades every day” — MKBHD
The speed of all of these advancements would come back to shock people in 2020.
Arm chips are being used for low power devices like in mobile phones.
Intel refused to make chips for the iPhone and then apple decided to do it themselves by acquiring a company and then with this expertise apple has pushed the state of the art technology forward until mobile phone chips were comparable to PCs and now it was time for ARM and Apple to take on the PC head-on in.
In November of 2020 Apple reveals its new processor to take on the PC market the M1 chip which is the company’s first custom CPU designed with Mac in mind.
It features 16 billion transistors 35% more than the A14 inside the new iPhone 12 and because it’s arm-based and shares the same DNA as the iPhone with transistors on a five-nanometer process.
Also, the new MacBook can run iPad and iPhone apps.
Early testing of the chip is blowing the online tech reviewer community off their feet proceeded to annihilate the other tests posting 1744 on single-core which is faster than every intel mac ever made. we can certainly believe that apple and the m1 will be able to achieve that claim AMD has shown lots of progress lately however it’ll be incredibly hard to catch up with apple’s power efficiency “if apple’s performance trajectory continues at this pace the x86 performance crown might never be regained” when you are aware of the wider applications of ARM in the real world it was only a matter of time
The fastest supercomputer in the world the Japanese fugaku is three times faster than IBM's Super Computer and it uses ARM architecture though this is the clincher needing a more powerful computer isn’t relevant to everyone but having more efficiency and less power consumption is and it’s something that everyone can notice.
“So arm chips have existed since the 1980s but in the current day, they’ve traditionally only been used for a lower-end device like smartphones for example though putting arm chips in a laptop has existed for a long time windows has done this for years but having an arm chip that is so powerful that it can reach the top end of the consumer computer market is new” — Christiansen. — Clay Christensen Author of “innovator’s dilemma”
Apple is planning to release more powerful apple silicon macs later this year and my guess is that they might announce their next arm based processor at the upcoming WWDC event in June.
And the most recent 2 devices which got the latest M1 chip was the iPad Pro and the new iMac.
Still, they are Mid Level or maybe Low power devices
I'm excited to check out what kind of plans apple have for their Pro devices like the Mac Pro and the iMac Pro.
That's the end of this article thank you for reading.
📝 Save this story in Journal.
