It’s not a novel idea. Apple has done so much to enhance the performance if iPhones and iPads using customized ARM chips. When it comes to Macs, Apple remains at the mercy of a third-party company. When the Motorola/IBM alliance failed to deliver the goods nearly a decade ago, Apple went to Intel. But the next generation of Intel chips, better known as Broadwell, has been delayed until the latter part of the year, which means new Macs will probably be delayed as well.
Sure, there was a modest update to the MacBook Air, using a slightly more powerful Haswell chip. The performance change is insignificant in the scheme of things, but paying $100 less could surely jumpstart Mac sales for the spring and summer. Or at least till something better comes along.
Meantime, an old rumor has become new again, that Apple is testing Macs using multicore ARM chips, perhaps as a future replacement should Intel’s roadmap not deliver the goods. This isn’t to say Intel’s processors are bad. But recent generations have emphasized power efficiency over improved number crunching. Sure integrated graphics are far better, but the overall benchmark improvements aren’t significant, or even noticeable, for the majority of Mac and Windows users.
In recent years, Apple has spent an estimated billion dollars in recent years acquiring chip design technology. You’ve seen it bear fruit in the A-series processors, particularly the A7 with 64-bit capability. Apple boasts it’s a “desktop class” processor, and that surely fuels speculation that it might turn up in a future Mac.
In the scheme of things, however, the A7 is nowhere near as powerful as a Haswell mobile processor. It’s getting better, but it may take a while for ARM technology, with Apple’s help, to catch up. If it does, maybe Apple might consider moving the Mac platform, aside from the Mac Pro, to ARM.
But it won’t be a casual move. Apple has done the processor-switching thing twice before and it wasn’t easy. Yes, porting OS X to ARM is not an issue, since the iOS led the way. But what about the tens of thousands of apps optimized for Intel? How well do they translate? This is a key issue that the rumors barely touch.
Apple’s solution would, in the near-term, probably be some sort of on-the-fly translation capability. Although the feature expired beginning with OS 10.7 Lion, Rosetta was the tool that allowed PowerPC apps to run on Intel Macs with decent performance. Not as good as native apps, but good enough for most purposes until developers had time to upgrade their products. Of course, some never bothered, but that’s another story.
In moving to ARM, Apple would have to offer a similar capability, and one that’s processor based would present the least performance hit. Indeed, the first ARM-based Macs, should they appear, would likely have to offer performance that represents a fairly decent improvement over Intel, partly to overcome the speed loss for translated apps.
But Apple wouldn’t make this move out of hubris. The A-processor roadmap would have to demonstrate a reasonable boost in performance and power efficiencies over what Intel is doing. The ARM architecture has come a long way, but nobody can say an iPad, snappy as it seems with those slim iOS apps, can approach even the lowly MacBook Air or Mac mini.
There is also the Mac Pro factor. Even of Apple can build a chip that offers good performance on an iMac doesn’t mean it will be powerful enough to match a Xeon, a cost-no-object chip for workstation and server use.
So what’s behind those Mac-on-ARM rumors?
Perhaps some of it is wishful thinking. It may also be true that there are indeed ARM-based Mac prototypes out there undergoing tests. I’d be surprised if there weren’t. But a prototype doesn’t necessarily represent a finished product. This may simply be an ongoing process with Apple considering options for the future.
After all, OS X on Intel was tested for several years before the processor switch was announced in 2005. During his WWDC keynote that year, in fact, Steve Jobs famously showed a satellite photo of the Apple facility in which the OS was under development. The revelation confirmed rumors that had been around for several years.
To be sure, Apple’s processor team is certainly working on chips that will be released for several years, and it may well be that, in a lab setting at any rate, there is an ARM processor that is fully capable of matching what is expected to arrive from Intel. Moving a prototype to a finished product, however, doesn’t happen overnight, and this is the sort of complicated move that would require careful thought, careful testing, and a long-range plan to actually occur.
Regardless of how it turns out, I have no doubt Apple could accomplish a successful ARM transition. The Intel example, where Macs migrated to the new processor faster than even Apple predicted, is an example of how well these turnovers can be accomplished.
Right now, though, I expect any testing with the ARM architecture is done more to have an alternative than as a potential final move for the Mac platform. It may happen, but it won’t occur that quickly unless Apple has achieved a design miracle or two that we know nothing about. Sure, one can speculate about the ARM switch as a potential “one more thing” revelation during a WWDC for this year, or the next. That doesn’t mean it’ll happen, so don’t make too much of the latest rumors.