Making macOS run well on ARM processors isn’t the hard part
Less surprise will mean more delight
Mark Gurman at Bloomberg has the confirmation we’ve all be waiting for: Apple will reportedly use a 12-core 5nm ARM processor in a 2021 Mac. There’s plenty of time to work out the details, but getting both the rollout and the technical side of this transition right won’t be easy. We watched a generation of pretty bad Windows 10 ARM laptops come out. Then we got the excellent Surface Pro X, which still has very aggravating software compatibility gotchas.
The forthright and direct way Apple handled the last Mac processor architecture switch — from PowerPC to Intel — went really well. Though I’ll admit it’s easy to say that now that the complications of that switch are so far behind us. Still, as I remember it everybody knew what to expect, knew it would take a minute, and was so eager for the switch that they were willing to deal with the hassles it caused.
If you haven’t ever watched Steve Jobs’ 2005 announcement of the Intel switch for Macs, I highly recommend it. He makes a strong case for the transition’s necessity, lays out the benefits for users, details how it’s going to happen, and makes some jokes in the process. He’s not trying to hide the rabbit, he just explains that Apple hadn’t been able to make the computers it wanted to make on the old PowerPC chips.
Making these new ARM chips that Gurman has detailed must be a huge, multi-year effort, but it will all be for naught if the software doesn’t run well — or at all — on them. And even then, the truth is that just porting macOS and Apple’s own apps over to ARM isn’t the hard part.
The hard part is clearly communicating to users and developers what the change will mean to them — and providing them with tools to deal with it. What software will work, won’t work, and will work slowly via emulation? What will developers need to do in order to port their apps over? Will porting an app to ARM even be worth the effort and cost?
Apple does not like pre-announcing anything, but I’m not sure how you effectuate a whole damn processor transition without giving developers a heads up. In fact, I think it would be utter madness to not give developers a heads-up as early as possible. Apple was willing to pre-announce and share some basic information on its Mac Pro plans well ahead of that release, so there’s recent precedent for pre-announcing.
This year’s WWDC would have been a great time to do it, but who knows if the online-only version of it will change that plan (if it was, you know, the plan). Certainly many developers would benefit from one-on-one time with Apple’s engineers once the transition is official.
That’s just the communication and release strategy. When it comes to the actual technical solutions, I am sure that there are no easy answers, either. Windows on ARM has performance issues with emulated apps and straight-up availability issues with apps that don’t work with its emulation. It’s entirely possible that Mac on ARM could face similar problems.
And while I am sure Apple was hoping Catalyst apps would be a piece of the puzzle, to date they’ve been quite disappointing. Even with a massive turnaround, they would be need to be just one of many strategies for getting fast apps on the new ARM-based macOS. There will surely need to be some sort of emulation layer for Intel-based apps. And I have to assume that the many developer tools Apple has been pushing lately (like Swift) will smooth the transition for app makers.
Even so, there’s a lot of work ahead for Apple and also for app developers, who will have to contend with this new processor architecture at some point. Hopefully that work will also come with new opportunities. I would love nothing more than to not hear anybody (including myself) complain that Adobe apps are either unavailable or painfully slow because there are so many great, native-ARM alternatives.
There are more questions than answers, and until we get a better sense of what Apple is planning for software compatibility it’s hard to even say what the right answers would be in the first place.
So the best I can do is offer some some very unsolicited advice: don’t be afraid to Osborne your current Macs, Apple. You’ve got the cash. Announce as early as you can and go all-out to support developers big and small. If you want to avoid the stigma Windows faced (and still faces) with its ARM version, make sure that macOS on ARM absolutely flies. Then take whatever investment you’re making in developer tools and developer relations and double it.
There’s a marketing term called “surprise and delight.” I’m sure you’ve heard it. When it comes to switching the Mac to ARM, I’d suggest forgetting about the surprise part — it’ll make it that much harder to get everybody to the delight part.
Source: Making macOS run well on ARM processors isn’t the hard part (The Verge)