Apple’s historic announcement of moving away from Intel’s x86 based microprocessors for its Mac line of personal computers was a long time coming.
Apple’s historic announcement of moving away from Intel’s x86 based microprocessors for its Mac line of personal computers was a long time coming. It is the most important milestone as yet as Apple aims to master its own destiny when it comes to processors.
Recommended ArticlesView All
Four-day work week is here but will it work for Indian companies?
IST7 Min(s) Read
Decoding multi-year health insurance policy — What is it and what are key benefits?
IST3 Min(s) Read
View | Pakistan Election: Will Imran Khan's changed tack from long march to resignations to snap poll work?
IST5 Min(s) Read
View | G20 Presidency: India can shape global Web3 narrative
IST6 Min(s) Read
To use a word recently made popular by PM Narendra Modi, Apple wanted to be Atmanirbhar (self-reliant).
Here’s the story of how Apple evolved into a powerhouse for processors.
More than a decade in the making
In 2008, Apple’s legendary founder and then CEO, the late Steve Jobs, pulled off a masterstroke. He acquired an Israeli semiconductor maker called PA Semi, which used to make custom chips for the CIA.
This was three years after Apple’s previous transition for the Mac, when it moved from IBM’s PowerPC processors to Intel’s then world-class chips. PA Semi was bought in the wake of the disruptive launch of the iPhone.
As disruptive as the first iPhone was, Apple felt the chips it sourced from vendors was keeping it from creating unique experiences.
PA Semi’s acquisition also coincided with the arrival of Johny Srouji to Apple - now its lead lead for hardware technologies who revealed on Apple’s silicon plans for the Mac at WWDC, two days ago.
Srouji was a rising star at iconic semiconductor companies like Intel and IBM. He was recruited to lead this new silicon division by longtime Steve Jobs lieutenant Bob Mansfield.
After the PA Semi buyout, Apple’s new skunkworks unit started developing technologies quickly. The first fruit of this unit was the A4 processor, which was released with the iconic iPhone 4 in 2010. Till date, many believe the iPhone 4 is the most iconic iPhone outside of the original and perhaps so, the best iPhone release Apple has ever had.
In 2013, the chip team made a huge technological leap. Back then, companies including Apple used Intel’s x86 chips for computers, they had adopted ARM’s designs and architecture for mobile phones.
The thinking was that ARM’s architecture was more suited for mobile devices as they were a simpler instruction set, which also made them more power-efficient than Intel’s processors. But the 32-bit ARM-based chips were inferior to Intel 64-bit chips, which are able to handle more data.
Then, during Apple iPhone 5s’ launch in 2013, Apple said its A7 processor would be based on 64-bit architecture, leaving behind other ARM microprocessor vendors like Qualcomm and Samsung who were a couple of years away from the feat. The launch of the A7 processor was foreshadowed by the launch of Apple’s new SWIFT programming language for its iPhone and iPad operating systems.
There was more. In 2014, Apple unveiled the Apple Watch to the world, showcasing the S1 processor that went into its heart. Ever since, the Apple Watch has become the gold standard fast and efficient smart watches, making it the world’s bestseller.
In 2016, this team was at it again. When Apple announced the “courageous” demise of the 3.5mm jack on the iPhone 7, it showed off a new type of wireless earphone called the AirPods. They were not just earphones, they were miniaturised computational speakers powered by a custom W1 chip. Today, AirPods are the best-selling earphones in the world.
Intel starts floundering
Earlier in 2016, Apple realized Intel’s ascendency in semiconductors had started to come off. Intel failed in its attempts to enter the mobile market with its Atom chips, but a bigger setback was awaiting its core PC business.
The effect of the “Moore’s Law” proposed by Intel co-founder Gordon Moore – which loosely means that processing power would keep exponentially increasing -- had started to wane.
In addition to this, its famously agile chipset foundries had started falling behind competitors like TSMC and Samsung. Its latest Skylake line of processors were being built on a 14nm node.
Come 2020, and TSMC, the manufacturer of Apple’s A-series of processors, built the A12Z Bionic in the latest iPad Pro on a 7nm process.
Today, the most advanced MacBook Pro, the 16-inch model is based on Intel’s latest IceLake processors which are on a 10nm process. TSMC will go one step further later this year and it is widely expected that the processor on the next iPhone and the first generation custom Apple chips for the Mac will be based on a new 5nm process.
The manufacturing process is a very critical part of how efficient a new chip is. The rule of thumb is that a chip’s performance and battery efficiency will improve if you pack the transistors closer. The lower the nm count, the better it is.
Of course, Apple saw the writing on the wall for Intel. In 2015, it had released a super-slim 12-inch MacBook. Its design was lauded the world over but it was a failure as Intel’s Atom processors were grossly underpowered. Apple’s notorious obsession with design and frugality of battery life was at loggerheads with the limitations of Intel’s silicon.
This problem reared its ugly head again in 2016 with the fourth-generation MacBook Pro, Apple’s flagship notebook. Intel’s chips were lacking their iconic performance and reliability. Apple also realised that it couldn’t add some new features like “Touch ID” from the iPhone. Apple’s chip team was called in to create a new processor called the T1 only for the tasks of handling the new touch bar, running its interface and managing the security.
In 2017, a new T2 chip was added to the Mac, which took a large part of the system on its head. This T2 chip was no slouch. It was a reworked version of the A10 Bionic chip that went in the iPhone 7. Apple was slowly learning how far it could go with its ARM-based silicon.
To make matters worse, Intel also struggled with shipments of its chips.
Intel wasn’t the only problem
The 2016 refresh of the MacBook Pro also came in the backdrop of Apple’s fallout with iconic graphics company Nvidia. By 2015, all Mac products had swapped out Nvidia GPUs with AMD’s inferior designs. Nvidia’s discrete GPUs were a key part of the Mac experience as these devices were used by creative professionals like graphic designers, video editors, photographers and industrial designers. Nvidia’s secret sauce was their CUDA cores, which were optimised for applications like Photoshop and Premiere Pro.
The bone of contention between Nvidia and Apple were the software drivers. Apple had released its Metal graphics framework and it wanted Nvidia to adapt its cards for it. Nvidia, on the other hand, wanted to release software drivers from its end. Apple wasn’t willing to relent control of a core subset of the macOS experience. Nvidia wasn’t motivated to play by Apple’s rules as Macs accounted for less than 10 percent of the PC market and it could do whatever it wanted on Windows side.
At that time, Apple needed an even bigger push from Intel on the discrete graphics side alongside AMD. While AMD held its own, Nvidia’s graphics technology remained the industry benchmark to aspire to.
Alongside, Apple was scaling graphics on mobile using its custom chips under graphics licensed from Imagination Technologies. While it struggled with finding the perfect processor match for the Mac, the iPhone and iPad began to carve out their names in the gaming industry, emboldening Apple was emboldened to try developing its GPUs in house as well. It started hiring, rather controversially, a lot of Imagination Technologies employees alongside stalwarts from the graphics industry.
At one time, renowned names like Jim Kellar and Raja Kodouri played a key role in the early graphics architecture of the iPhone. Kellar had stints at AMD as their chief architect, then moved to Tesla and Intel before resigning recently. Kodori, a legend in the graphics technology industry, was at Apple for 4 years till 2013, then moved to AMD where he led the graphics, He was recently hired by Intel to help with its weaknesses in mobile graphics.
Since 2017, Apple has built its GPU team, which has developed industry-leading graphics. The first Apple-designed GPUs came on the iPhone 8 and iPhone X.
Today, GPUs have become very important for on-device machine learning. Apple has also progressed to a point where a device as slim as the iPad Pro -- measuring just 5.9mm -- can have the graphics of a huge gaming console like the Xbox One S. Apple is finally ready to take matters in its own hands.
Apple’s tryst with modular ARM architecture
The more Apple used ARM, the more it realised that it was the way to go for its future. Apple had been using ARM-based chips since the 90s when it first introduced the Newton PDA. Later in the 2000s, even the iPod was powered by chips based on ARM’s designs. In recent years, Apple carried out a lot of customisation to the ARM design and added things like the neural engine, which allows features like Face ID. Recently, Apple also added machine learning accelerators and its custom ISP -- this became the secret sauce behind the iPhone camera.
Johny Srouji, who recently was also in the running to become Intel’s CEO, said at WWDC that Apple will be able to customise these chips following its product roadmap. Apple never had that level of synergy with its partners -- Intel and Nvidia. AMD has been more yielding as it releases chips that are exclusive to Apple’s product line but AMD’s strength isn’t in the discrete GPU space.
Apple recently also acquired Intel’s modem business, which will likely be integrated at some point with its custom chips. Intel was developing 5G modems for the iPhone for years, but Srouji pulled the plug on the deal as he wasn’t satisfied with the results. The modem space was nearly single-handedly dominated by Qualcomm, which forced Intel to sell its division to Apple.
Under Srouji, Apple will likely have its 5G modems in a couple of years. The iPhones in 2020 are expected to use Qualcomm’s modems after both the companies settled a longstanding court case. But for the Mac, where 5G may not be critical, Apple could come up with its own 4G solution.
Apple’s developer transition kit a precursor of things to come
Apple’s developer transition kit – intended to assist developers in the transition away from Intel’s chips -- is a Mac Mini with a twist. Instead of an Intel Core i7 processor, it has the A12Z bionic chip coupled with 16GB RAM and 512GB of SSD storage. Mind you, this is the same chip as the new iPad Pro based on an 8-core CPU and 8-core GPU, which is coupled with 6GB RAM and up to 1TB of SSD storage. This is a chip built on TSMC’s 7nm process.
Thanks to the extra breathing space granted by the Mac Mini, Apple can afford to run this chip at higher clock speeds given that the setup can manage the extra heat generated. The additional RAM will also remove any bottlenecks that one would have faced in a 6GB iPad. And this is the machine that’s running macOS Big Sur. This is the same machine that’s running hardcore pro software like Final Cut Pro with three streams of 4k seamlessly. It could even handle a massive 5GB PSD file running on Adobe Photoshop.
Apple has built a virtualisation subset on the SoC-level (system-on-chip), which allows virtualized instances of Linux and Docker. Apple and Microsoft should be able to even get Windows on this if the two giants are in the mood. This shouldn’t be a problem as even Microsoft has recompiled its Office software for these new Apple processors.
Even for software that’s not compiled for these new ARM chips, Apple has a bridge software layer called Rosetta 2 which is very performant. Apple showed 3D modelling software Maya and the latest Tomb Raider game running at 1080p on this iPad processor.
This hints towards a tectonic shift in performance towards the Mac. Reports have suggested that Apple is working on a version of the A14 processor based on a 5nm process. This processor is said to have at least 12 CPU cores, which would translate to mind-boggling CPU performance gains. Apple will likely also deliver industry-leading battery life with these new chips.
Apple is also saying that all software compiled for the iPad and iPhone will work out of the box on the Mac. Also when the software is compiled specifically for the new semiconductor architecture, one can expect the Mac to regain its prime position in graphics while retaining the slick and sexy of Apple’s hardware industrial design. Simply put, these devices will be significantly more powerful than anything running traditional Windows, Intel and Nvidia OS, CPU and GPU combinations.
Macs haven’t been known for gaming, but given Apple’s recent push in the gaming space, custom GPUs could make the Mac a platform that gamers like. Already the unity and unreal engines work very well on the Mac. Apple could supercharge this by adding a prodigious GPU.
Apple is also saying that little known plugins will also be supported via Rosetta 2 but it advises all developers to recompile their software natively for the new chips.
There will be a pain for some
Little known software developers will be impacted the most by this. Even though Apple says recompiling apps via the Xcode tool will just take a few days to work, this may not be true or feasible for some. The performance will also not be optimal on apps that are running via Rosetta 2. This will impact games the most alongside pro software like digital audio workstations that are dependent on little applets called plugins.
Tim Cook says this transition will take 2 years, but this could take longer for some app developers.
Apple is also not dumping Intel fully. Cook said that it has several Macs powered by Intel in the pipeline. This will be particularly true for the most powerful Macs like the Mac Pro, which Apple unveiled last year and the iMac Pro.
Intel’s Xeon line of server-grade processors are still the benchmark for the industry and it is highly unlikely that Apple will scale its chips for something as powerful as the Mac Pro in just 2 years. One can expect the Mac Pro and iMac Pro to be based on Intel for at least 3-4 years.
In the end, Apple wants to be the master of its destiny by creating a fully vertically integrated stack. It also helps Apple to transition the Mac to its processor architecture since the iPhone, iPad, HomePod, Apple TV, Apple Watch and AirPod are on it already.
For most users, the difference will be a positive one. From a user experience perspective, things shouldn’t change. On the contrary, most things should become better.
Macs will pull an iPhone SE on the PC market
Apple will likely make more profits too as Intel’s chips are very expensive. Making its own chips will be cheaper for Apple.
Apple has never been about playing the affordability card. So, don’t expect the Macs to become cheaper. But one can expected them to be significantly more useful at the same price points, if not slightly lesser. It could play the value performance card.
This would make entry level Mac notebooks significantly more desirable for users to need “pro” or “gaming” tasks at less than the Rs 1,00,000 price point. Apple will eat more market share into the PC market. Currently, Macs have less than 10% of the PC market, but if Apple’s market share increases then it becomes a more desirable platform for developers playing well with its push into services.
Apple has launched services like Apple Arcade and Apple TV+ and the increased capabilities of the Mac with these new processors will attract more users, in turn developers turning its ecosystem into a virtuous flywheel.
Apple has done the same with the iPhone SE, which gets the same chipset as the iPhone 11 Pro at half the price. At Rs 42,500, the iPhone SE is objectively faster than any Android smartphone that’s sold, including the high-end OnePlus and Samsung phones. That’s the value performance card Apple is known to play ruthlessly well. And it will do it again with the Macs.
This is a master class in long term thinking and surgical execution that’s unprecedented in the consumer technology sector. At the end of the day, Apple’s shift comes because it couldn’t make the products it wanted to make for its users — not will be able to and it wouldn’t be able to pass on the blame on any partner.
It’s going to be on its own, but in Apple’s universe, that’s always been a good thing.
First Published: IST