Dieter Bohn, wrote a great article about processor transitions published on The Verge. The article titled “What Windows can teach the Mac about the switch to ARM processors” illustrates some issues Apple may face.
Looking back after the Apple M1 Mac announcement, we can see how they addressed each and every topic in that piece. It’s a great example of how Apple tends to put all their wood behind a single arrow, making sure as much as possible that they get everything right.
ARM-based Windows computers are slower
Apple could have made this transition earlier if it was content with Apple silicon Macs being as slow as the Surface Pro X. I’m sure that they had MacOS running on ARM for many years already, gauging the performance compared to Intel. They waited until performance was good enough, which, based on their previous successful transitions, means about double the performance on at least some key tasks. With the M1 chips which have screaming performance, Apple has clearly overcome this hurdle.
Windows computer with an ARM processor actually requires a higher level of technical expertise
Dieter is talking about how Windows on ARM can emulate 32-bit Intel code but cannot run 64-bit code, and so users have to know which apps are written in which binary. Even when Apple transitioned from Motorola 68k to PowerPC, they made it very seamless with a great emulator. This time, Apple has Rosetta 2 which is apparently at least as good, so they cleared this one too.
“Here’s a cool new thing you can get if you want, but the reliable old thing isn’t going anywhere.”
Here Dieter is talking about whether Apple should allow ARM and Intel to coexist, or whether it should transition the whole lineup. Windows is supporting both x86 and ARM, and the emphasis is still clearly on x86 — ARM is a niche platform for those people who prioritise ultra-portability and long battery life ahead of performance and compatibility. As a result, as Dieter mentions, Windows on ARM is not getting hardly any developer attention. Even Microsoft has failed to port Office to ARM. Apple on the other hand has been very clear. Although they will continue to support Intel for many years, they will transition their whole Mac lineup to ARM within a couple of years. ARM and Intel will not coexist and there will be a clear transition. This is how Apple has managed every other transition.
What Apple got right
Apple is one of the rare tech companies that does not constantly throw spaghetti against the wall to see what sticks. In the case of this transition and all previous ones, instead of spitting out hardware with the new chips before they were ready, they waited (intentionally or otherwise) until the new processor architecture provided a multiple-fold performance advantage, prepared a great emulator, and then made a full transition in one sweep. Because of this, developers prioritised re-writing their code and supporting the new Macs.
In hindsight, Steven Sinofsky’s tweet storm also shows why Apple’s strategy of doing everything in discrete stages makes a lot of sense and why Windows RT (the ARM port of Windows 8) failed. Apple is forceful in committing to change, but also careful to change as little as possible with each transition. In the 68k to PowerPC transition and also in the PowerPC to Intel transition, Apple changed only the CPU architecture while maintaining the software APIs. On the other hand, when making the transition from the Classic Mac OS to MacOS X, the hardware was kept on the PowerPC. Apple never made a transition where both the hardware and software APIs were changed. In contrast, it is clear that Windows RT tried to do too much in one go. With the current Intel to ARM transition, Apple is changing only the CPU architecture while keeping the software APIs and the user interaction model exactly the same.
To put this in one simple word, I would choose “focus”. Apple focused on the right timing. They focused on getting all the parts ready for that timing. They also focused on doing just one thing during each transition. Dieter’s article and Steven’s tweet storm remind us that this was the very thing that Microsoft lacked.
Naofumi Kagami 