Obstacles to an ARM-based MacBook Air
By: Brian Webster |
There has been a fair bit of talk over the last couple years about the possibility of Apple creating a version of their MacBook Air that uses an ARM processor rather than the current Intel processors. This has become even more prominent over the past few weeks, after Apple announced their new 64-bit A7 chip that powers the iPhone 5s. I don’t think this is going to happen anytime soon though, and here are some of the reasons why.
Why make an ARM MacBook Air at all?
There are a couple different upsides that Apple could see from switching the MacBook Air over to an ARM processor. The main advantage would likely be increased battery life, since ARM processors have largely been used in mobile products, and low power consumption has always been a top priority in their design. Intel has made a lot of progress over the years in making their mobile chips much less power hungry, but they will always have a fair bit of legacy design to overcome in this area, since their architecture was originally designed for desktop, not mobile use. One of the MacBook Air’s main selling points is its battery life, so an ARM processor would seem to be an ideal match for the machine.
Another advantage, which is a bit less concrete, would simply be that Apple would gain control over yet another component in the supply chain that’s necessary to build the MacBook Air. On the iPhone side of things, they now design a fair number of the components themselves, so it would make sense that they might want to do more of this on the Mac side too.
What would the hardware look like?
On the hardware side, Apple would obviously need to make a SoC (system on a chip) that could handle everything that a Mac would need, including the processor, graphics chip, and interfacing with a logic board with USB, Thunderbolt, and other such connections. This would be a fair bit of work, but nothing especially out of the ordinary for Apple here.
They would need to make sure that the performance of the machine at least matched what they are getting from their current Intel processors. It’s fairly difficult to compare the performance of platforms as different as the Mac and iOS, but we can get a ballpark estimate by looking at numbers from the Geekbench suite of benchmark tests. On the iOS side of things, the current iPhone 5s gets a single-core score of around 1400. On the Mac side, the current base level MacBook Airs score around 2560-2590.
So, to match their current performance level, Apple would need to crank up their A7 chip to nearly double its current output. This could certainly be possible, since I’m betting the A7 is set at a fairly moderate clock speed in the iPhone, to try to find a “sweet spot” for performance vs. battery life. If they did this, then the question becomes, how would that affect the battery life of this hypothetical MacBook Air? There’s no way to know for sure, but I think it’s not a sure bet that an A7-based Air would necessarily get significantly better battery life than the Intel ones.
How about the software?
The first order of business here would be to get an ARM version of OS X up and running. Apple almost certainly has such a thing lurking on some machine down in a basement somewhere, but there’s a big leap between a skunkworks project and a shipping product. Apple is no stranger to such transitions, having done it twice before (68k => PPC, then PPC => Intel), but it is still far from a trivial task to accomplish.
The whole OS needs to be checked and rechecked for resulting bugs, especially those stemming from going back from a little-endian Intel processor to a big-endian ARM processor. All of Apple’s apps would need to be updated, and they would need to provide tools for developers to update their own apps. Since people will still have quite a large number of Intel-only programs, a Rosetta-like solution would need to be provided for emulating those apps during the few years of transition to the new processor. And of course any device drivers would also need to be rewritten for the new architecture. This is all achievable, but I think a lot of people dismiss how big of an effort this really is, since Apple has made it look relatively easy from the outside the couple times they’ve done it so far.
OK, so we have an ARM MacBook Air! Um wait, doesn’t Apple make some other stuff too?
Oh yeah, it turns out Apple actually has a whole line of Macs, not just the MacBook Air, going all the way up to the yet-to-be-released “trash can” Mac Pro. So what’s going to happen with those machines? Are they also going to be updated with ARM processors? How will Apple be able to make such a processor that matches the performance of the current Intel chips? I guess it might theoretically be possible, but I doubt that Apple could produce anything that would challenge Intel at the high end of the market.
One way they might work around that is to keep the rest of the line (or at least the desktop machines) running on Intel, and use ARM just for the laptop side of things. That’s feasible, but then you’re not just talking about porting OS X from Intel to ARM, but maintaining two separate versions of the OS going forward. This would basically mean Apple would now have three platforms going forward instead of two. Yeesh.
Does it need to run OS X?
OK, so if porting and maintaining OS X on ARM turns out to be too much work, what if there were some other operating system this MacBook Air could run instead? Like, say, iOS!
I’ll admit this is certainly an intriguing thought. I don’t think such a product could really be called a “Mac” though, since the main difference between iOS and the Mac is what software they can run, so such a hypothetical product would probably be named some variant of “iPad”. But, given the number of third party setups you can buy that wrap an iPad up with a bluetooth keyword, so you can use it as a pseudo-laptop, it’s not out of the question that Apple could make some sort of iPad with a built-in keyboard, making it basically a laptop form factor. It couldn’t possibly be a replacement for the MacBook Air though, since the whole point of the Air is that it can run Mac software.
Or Apple could merge iOS and OS X, and make one operating system that provided a first class interface with both a touch screen and mouse/keyword input. Just like Windows 8.
…
…
What, you’re still not done laughing? That’s OK, go ahead and get it out of your system. I’ll wait.
Of all the possibilities here, I think this would be the most plausible, though that is far from saying that it’s something Apple would actually want to do. There would certainly be some questions to figure out (Would it support a mouse/trackpad? USB/Thunderbolt ports? Detaching just the “iPad” part from the rest?), but it’s a feasible idea at least.
To sum up
So, I wouldn’t go so far as to say Apple would never produce an ARM-based Mac, but given the current state of things and the amount of effort it would take, I really don’t see how the work it would take would pay off in the near term for such a project.