It is not always that such a drastic change is announced: after much speculation, the Apple confirmed that the new Macs will adopt their own processors, based on architecture ARM, still in 2020. It is the end of a 15-year era that began in 2005, when Steve Jobs revealed the transition from PowerPC to current Intel chips.
THE Apple Silicon can impact the entire industry. Intel reigned alone in the PC market for many years, without real competition from AMD. And Apple has shown that it knows how to design powerful chips on iPhones and iPads, with an ARM architecture more optimized than any other company.
But nobody is in a comfortable situation. If Apple succeeds and manages to innovate in this market, Intel and other PC chip makers will be pressured to pursue it. On the other hand, this is the end of such a close relationship between Apple and a giant that has been a leader in the semiconductor market for 24 consecutive years. And now?
Intel transition to Apple Silicon on video
Macs with Apple Silicon: what we have in concrete
The first Macs with Apple Silicon will be launched by the end of 2020. According to Tim Cook, it is estimated that the architectural transition will take two years to complete. This means that, if everything goes as planned, the entire line of Apple computers will already be equipped with the new chips in mid-2022.
Despite the change, Apple is still developing Intel-based computers that will hit the market soon. In addition, Macs with x86 architecture will continue to receive updates, including new versions of macOS, for an unspecified time. MacOS Big Sur, for example, will be compatible with Macs launched seven years ago, such as the MacBooks Air and Pro 2013.
As for the ecosystem, the new Macs with Apple Silicon will be able to run:
- Old apps through Rosetta 2, which translates the x86 code to ARM at the time of installation;
- Native apps of macOS compiled for Apple Silicon, with Microsoft and Adobe (keep those names) already confirmed that they will develop professional software for the new architecture, and Apple’s Final Cut Pro and Logic Pro programs are ready;
- And for the first time, iPhone and iPad apps, which can be installed through the Mac App Store.
The current iPad chips will be the starting point for future Macs with Apple Silicon, as the first development kit is a Mac mini with a processor Apple A12Z Bionic, the same that equips the iPad Pro launched in March 2020.
The transition from PowerPC to Intel
It is not the first time that Apple has faced a similar challenge. The latest architectural change was announced at the WWDC 2005 developer conference, when Steve Jobs took the stage and showed the phrase “It’s true”, with the “e” dropped, in reference to the Intel logo from time. That was the official starting point for the transition from PowerPC to Intel x86, but rumors have been circulating since 2002.
It was a reverse change to what is happening today, as the PowerPC was created in 1992 in an alliance between Apple, IBM and Motorola. The architecture of Mac OS X was no longer in good shape in the 2000s because Intel was doing a better job, both in product and in marketing. I remember the number war very well: Intel made a noise when it crossed the 3.0 GHz barrier on the Pentium 4, a frequency that no one could reach, including the IBM PowerPC chips.
Apple does not seem to be afraid of killing technologies: it was like that with the floppy drive, the physical keyboard on the smartphone and the Adobe Flash Player. With PowerPC appearing to be at the limit and Intel chips being faster with less power consumption, the transition was expected to happen.
And it happened very quickly. Steve Jobs announced the transition in June 2005, with a forecast for completion by the end of 2007 (that is, the same two-year estimate for the current transition). But the first MacBook with Intel arrived in January 2006, and in August of that year, the entire line of desktops, notebooks and even an Apple server had already abandoned PowerPC.
The result is that the transition was completed in exactly 210 days, more than a year ahead of schedule.
In the announcement of the transition, Steve Jobs confirmed rumors that all versions of Mac OS X were secretly being compiled for both PowerPC and Intel – that is, since 2001. It was also important to have support from developers, and the first two companies to announce support for Macs with Intel were Microsoft and Adobe (what a coincidence, right?).
In the operating system, the story was similar, too. Apple announced for Mac OS X 10.4 Tiger software called Rosetta, which translated PowerPC programs for new Intel-based Macs, giving developers more time to adapt and compile their software for the new architecture. Rosetta lasted until the 2009 Mac OS X 10.6 Snow Leopard, which was also the first version exclusively for Intel.
Now that PC and Mac had the same processor architecture, Apple made the decision to launch the Boot Camp, a tool that allows you to install Windows and then Mac OS X in dual boot. This still exists today and works very well for those who still need to run exclusive Windows software (basically, games).
I was a Windows XP user at the time and did not have this experience, but I invited Rafael Fischmann, editor in chief of MacMagazine, to tell how it was to feel the transition from PowerPC to Intel. He tells everything in the video above. 😉
Going back to the present: what we discovered most
Of course, after the official announcement at WWDC 2020, more information emerged about the transition from Intel to ARM.
The first is that new Macs with Apple Silicon will no longer have Boot Camp, which was confirmed by Apple’s senior vice president of software engineering Craig Federighi in an interview with The Talk Show. It will still be possible to run virtual machines, something demonstrated in Apple’s own presentation, but it is the end of the story of dual booting with macOS and another operating system.
And yes, there is a version of Windows 10 compiled for ARM, but that does not mean that it will be possible to install it on Macs. First because Microsoft has confirmed that it does not sell licenses of this version to end users, only to manufacturers. Second because the simple fact that Apple’s processor is based on ARM does not guarantee compatibility with operating systems compiled for ARM.
The reverse is also true: the fact that an operating system, such as macOS, is compiled for ARM, does not mean that it can be installed on any computer with this architecture. Therefore, the practice of hackintosh, installing macOS on a regular PC, can be very difficult or even impossible when Apple stops supporting Intel processors.
This because ARM is not a processor in itself, but a foundation of an architecture on which other licensed companies (such as Apple, AMD, MediaTek, Nvidia, Samsung and Qualcomm) can design their chips, with their own optimizations and extra instructions. With the exception of Samsung, none of these companies manufacture processors: everything is outsourced to companies like TSMC.
The flexibility of ARM allows devices to be very different from each other, which is one of the reasons why you are unable to install iOS on your Android, or Android on your iPhone. Likewise, it is not because your Wi-Fi router has an ARM processor that you can install Android on it.
Also in the week of the announcement of the transition, a statement was made by François Piednoël, one of the main ex-engineers at Intel, who worked at the company for twenty years, saying that Skylake processors (6th generation Intel Core, launched in 2015) were the trigger for Apple to decide on the architectural change.
According to Piednoël, Skylake’s quality control was “abnormally bad” because Apple found many flaws in the architecture. It is normal for there to be an exchange of knowledge between partner companies, but the former executive says: “Basically, our colleagues at Apple have become the number one rapporteur on architectural problems. And that was really, really bad. When your customer starts to find almost as many errors as you do, you are not going the right way. ”
In a developer session, Apple explained that, unlike Intel x86 architecture, Macs with Apple Silicon will have asymmetric processing cores – that is, a group of high-performance cores and a group of low energy consumption, as already happens in ARM chips for cell phones and tablets. One possibility is to keep all the heavy processing in the powerful cores and leave the rendering of the interface to the economic cores, generating a greater sense of fluidity.
What about the performance of Macs with Apple Silicon?
The first Macs with Apple processors didn’t arrive until the end of 2020, so any information on performance, price or compatibility is pure speculation – which does not mean that we cannot analyze the change based on hard data (and then come back here to see the when I got it right).
One of the doubts is, of course, the power of Apple chips. I like Geekbench because it is one of the few multiplatform benchmarks calibrated by processor. According to official documentation, an Intel Core i3-8100 scores 1,000 points in single-core mode. A processor that reaches 2,000 points has twice the performance of that chip.
By Geekbench, a 2017 iMac Retina 5K, with Intel Core i7-7700K processor, makes 1,119 points in single-core and 4,558 in multi-core. A 2020 iPad Pro with an Apple A12Z Bionic chip has 1,118 points in single-core and 4,625 points in multi-core.
I always had good expectations with the transition from Intel to ARM because Apple is doing an excellent job. The proof of this is that there is still nothing like it in the competition: Qualcomm’s Snapdragon 865 doesn’t even come close.
Of course, there are some points to consider. First, we have to differentiate between peak performance and sustained performance. The simplest way to explain it is with the Intel processor numbers for notebooks: the Core i9-9880H has a maximum frequency of 4.8 GHz. But the base frequency is 2.3 GHz. This is because it is not viable, with the architecture current, make the chip work at 4.8 GHz at all times: you would need to have a power plant at home and a very powerful air conditioner.
Therefore, the same score does not necessarily mean the same performance in long-term heavy tasks, such as 4K video editing. The point is that… Apple is already doing better. The 2017 iPad Pro, with A12X Bionic, rendered 4K video four times faster than Windows and Core i7 notebooks. Remember that the iPad is at a thermal disadvantage: it does not have a fan to cool the processor, it is not usually used directly in the socket and is one third of the thickness.
So, considering that an Apple ARM processor on a MacBook will have an active cooling system, more physical space for the transistors and more electricity to work with, the scenario looks very promising. Even more because Apple did not even try to beat Intel (after all, the iPad is a tablet and does not need to compete directly with a notebook).
How is Intel in this story?
Little should change for Intel, at least in the short term, because the company has a very diverse range of customers – and even the biggest one, Dell, only represents 15% of the company’s revenue. It is different from a small business like Imagination Technologies, which almost closed its doors after Apple decided to design its own graphics chips.
In addition, Intel executives must have received the news calmly, not least because the transition was already expected: trying to do things “inside the house” is a strategy for decades at Apple. Also because Intel itself tries to detach itself from the image of a “processor company” to sell itself as a “data-centric company”. And Intel knows that it has a more than consolidated position in servers, a market that grows faster and has better profit margins.
Apple’s chip cost advantage
But if Apple is going to rely on the ARM architecture, what prevents another company, like Qualcomm or even Intel, from licensing ARM technology and starting to design better chips? Nothing. But Apple has an advantage that, perhaps, only Samsung can get close to: the vertical integration and the dissipation of costs.
In recent years, iPhone processors have been better than those available for Android phones. Don’t believe the myth of software optimization: the iPhone is faster mainly because the hardware performs better raw. Obviously, this does not come from the sky: Apple spends more on research, development and technology to develop a better chip which, consequently, is more expensive.
The advantage is that Apple can afford to spend more to make a better processor: it only sells the chip to itself, and can dissipate development costs elsewhere on the iPhone (or earn more in other areas, such as services or accessories). Qualcomm, Intel, MediaTek, or another company whose sale of processors is a core activity does not have the same freedom: if they make a very expensive chip, they may not be able to get customers to cover the costs.
Will it be too expensive? (Go.)
All of this leads us to another point: how much will all this joke cost? I saw some people betting on Tecnoblog Community that the prices of Macs could decrease because Apple would design the processors themselves and eliminate an intermediary, which today is Intel. In fact, this is illogical for several reasons.
Apple is a company that prioritizes margin over volume. Many companies, if they were in the same situation, would make their own chips and sell PCs in all segments with values lower than Intel’s equivalents, starting a price war. In this case, whoever wins the largest market share wins: if the XPTO company starts selling 80% of the chips and Intel only 20%, XPTO won.
But what we’ve seen at Apple in recent years is that, instead of throwing chips in all price ranges, the company prefers to focus on increasing its share in the more expensive segments, where the margin is higher. That’s why Apple has only 12% of the smartphone market share, but it eats two-thirds of the profit for the entire industry.
(In fact, in an economic system known as capitalism, every company aims to profit more and sell more to have a better margin, although not all of them achieve this in the same proportion: if XPTO starts to launch more expensive products than Apple just to make a bigger profit, maybe the most you can get is a bankruptcy for lack of buyers.)
Anyway, it is illusory to expect a drop in the price of Macs (and impossible in Brazil, with uncontrolled exchange). On the other hand, we may have Macs that are more powerful than the current ones due to Apple’s increased commitment to processor development.
But best of all, if the move goes well, Apple can pull the competition up and make even other companies’ computers, in all price ranges, look better and more powerful in the future – and the cliché “Who wins is the user” would become true once again.