Macintel: The end is nigh

Monday Note:

When Apple announced its 64-bit A7 processor, I dismissed the speculation that this could lead to a switch away from Intel chips for the Macintosh line for a homegrown “desktop-class” chip. I might have been wrong.

I don’t know enough about this end of the industry but it’s hard to argue with Gassee’s line of “Secondly, the Mac line is suspended, literally, by the late delivery of Intel’s Broadwell x86 processors.” The “end of MacIntel” is certainly a possibility to be on the look out for.

  • Nicholas Chan

    Also, Metal for OS X might be interesting.

    • matthewmaurice

      Not sure that we’ll see Metal for OS X so much as it being the core graphics component of OS 11 that merges the best of OS X and iOS for a unified experience that scales from iPhones to iPads up through whatever replaces MacBooks and iMacs.

      The irony is that Apple is moving towards the same place Microsoft, with it’s “One Windows”/”three screens and a cloud” strategy has been stumbling towards for half a decade. Of course Apple is doing it at a measured, controlled pace. Let’s not forget that part of what Apple got from the NeXT crew was incredible porting expertise. It’s been said that Apple had OS X running on Intel hardware in its labs almost from the beginning. I’m sure Apple has had “ArmBook Airs” with A5, -6, and/or -7 CPUs in the lab running OS X versions, probably as far back as Mountain Lion but certainly Mavericks and Yosemite. I doubt they were usable for much, other than as engineering studies, but you’re nuts if you think Apple hasn’t seriously considered a day when Apple developed ARM chips run a next-generation Apple OS across the entire range of Apple hardware.

      • Mac OS X, as Rhapsody, was actually originally running on Intel PCs while it was being ported to PowerPC Macs. I was working on it at Apple at the time (summer, 1997). It didn’t have the “blue box” compatibility environment to run Classic Mac OS applications, though. I’m sure that only ever ran on PowerPC Macs (though I left Apple before Rhapsody shipped, as the original Mac OS X Server).

        As for “unifying” OS X and iOS, I don’t think Apple will ever try to combine the mouse-based UI of OS X and the touch-based UI of iOS, like Microsoft tried to do with Windows 8. There’s just not enough benefits of it for Apple, and lots of detriments. It’s more likely they’ll just keep a common core, and continue to cross-polinate elements back and forth. Heck, as it is Apple already has two more separate UIs, for Apple TV and CarPlay. They may soon add another for wearables, too.

      • The irony is that Apple is moving towards the same place Microsoft, with it’s “One Windows”

        i dont think so. Schiller and whoever have already gone on record that they think a single OS for all devices is a dumb idea.

  • I’ve been skeptical about OS X on ARM too. I figured it could happen someday, but not anytime soon.

    Two things came together to change my mind.

    First, I was thinking about the original MacBook Air, an amazing machine that marked a new era in Mac notebooks. It also really kind of sucked (mediocre battery life, thermal problems, mediocre I/O performance, even with an SSD, only supported for two OS updates), and it did so because Apple did the best they could do with the best Intel was willing to do for them. It was released just two years after the first Intel Mac, and, fortunately, Apple managed to get Intel to up their game, for a while at least.

    A week or so after this insight on the Air, I heard about the Broadwell delays, and it became clear that I was wrong about the time table for OS X on ARM. Intel was, again, holding Apple back, but Apple is now better positioned to do something about it.

    That isn’t to say that the A-series is up to the task yet. Much is made about the cost of x86 compatibility in terms of die area and power consumption, but at this point, the die area cost is relatively minor, and the power consumption of x86 desktop chips probably has more to do with their overall performance than the x86 overhead.

    Mobile device CPUs have very tight power and thermal constraints. They also have the advantage of running software that is typically lighter weight and more optimized. As a result, they can get away with lower levels of absolute performance, which has an important added benefit: they don’t have to do as much work to contend with the gap between computational throughput and memory bandwidth/latency. An A-series chip that can keep up will Intel’s main CPU line have to contend with that problem, and will see performance/watt drop even as absolute performance rises.

    Apple has already done part of the work to deal with memory latency in the A7. It is a very “wide” chip, allowing a lot of execution in parallel, which is one of the approaches used to deal with the memory/core mismatch (if one instruction has to wait for memory, other pipelines can still continue executing). To compete with Intel’s “desktop” chips, they’ll have to do more than that though, but we should expect them to, because the A7’s design doesn’t make a lot of sense for a chip line that is just intended for tablets and phones. It makes a lot more sense viewed as a step towards developing the processor IP needed to hit much higher levels of absolute performance.

  • Moeskido

    How interesting would it be for Apple to have made two major processor changes to its flagship computer line within a decade? What other company would have the patience and foresight to execute such a thing?

  • rb763

    Would we lose Windows on Mac? For me it is a necessary evil and I’m using the word evil loosely. More like a necessary pain in the butt.

    • Colin Mattson

      Along those lines, I could see switching the MacBook Air line (and possibly the iMacs) to ARM.

      But the MBP and Mac Pro? There’s a tremendously huge chunk of that market depending on x64. Those buyers can forgive a lot of weird decisions and temporary inconveniences… when they still have the x64 processor they need to do their work.

      Unless AMD’s going to suddenly start knocking things out of the park (and monkeys will fly out of my butt), Intel’s the only game in town for the high-end machines and high-end users.

    • BenRoethig

      Win 9 will basically treat ARM v8 and x86 as equals.

  • Michael

    Maybe I’m being dense, but I just don’t see this happening. Asking devs to re-engineer their apps again for a move to another CPU platform would be a daunting ask, especially considering Apple’s stated goal to move to the “post-PC” era. What possible reason would devs have to jump on board with this with the looming desire Apple has to mothball traditional desktop computing?

    • matthewmaurice

      I’m not sure how daunting that task would be for developers. I’m not saying it would be easy, but I think it’s a much easier sell now than when Apple moved to Intel. First, I see lots of Macs in the hands of established and wannabe developers. Second, I think the mood in, at least parts, of the dev community is that between Apple’s history of “skating to where the puck is going to be” and its willingness to cannibalize its own products before someone else does makes the Apple platform a good choice to “hitch your wagon to.” So where they go, you’d best follow.

    • You are ignoring the fact that Apple has already done this twice in the past 20 years. First to move from the 68K line of CPUs, and then to Intel. Furthermore, iOS already represents another partial step in this direction in that their ARM compiler infrastructure is mature.

      One of the most important lessons they learned from those earlier transitions is that they need to be in full control of the primary development environment for their platform(s). The PowerPC to Intel transition was slower and rockier because there were a bunch of companies like Adobe who depended on the mostly dead CodeWarrior dev environment. Now everything is consolidated on XCode.

      Plus, some change is good for ecosystems, clearing out the underbrush creates opportunities for developers.

    • What? This has little to do with mothballing traditional desktop computing, but (if it’s true) with providing a better experience.

      We developers are users ourselves. Damn right we’re excited by a better experience.

    • you misunderstand “post-PC” era — post means “after”, it does not mean “replace”.

      can you link to Apple’s stated desire to eliminate desktop computing?

  • Are we really sure this is a transition rather than two architectures run in parallel?

    • It’s not a transition at all. Just supposition.

  • Terry Maraccini

    I don’t think this will happen any time soon. But the future is uncertain and the end and always near

  • jwoodgett

    In a small office in Cupertino, a suppressed laugh is perhaps rising above the sound of keyboard clicking. The second shoe (after Swift) won’t drop just yet, but Mac on Intel ran for 5 years (or more) before being hauled onto center stage.

  • Colin Jensen

    Makes no sense: there is no upside to moving to ARM. Let’s remember our history: when PPC was a bit faster than Intel, it was a “bit” faster and not a big deal. When PPC lagged, Apple looked like fools for using it.

    All Apple has to do is stick to Intel and no one will ever complain about the processor speed.

    • but they will continue to complain about battery. battery is now #1, processor speed is not.

  • theothergeoff

    wasn’t the Mac Line Suspended by the inability by IBM to reduce the Heat/Energy requirements of the G- series chips. Apple, living at the Mid-to-high end of the laptop/Consumer-PC line, is hamstrung by their CPU vendor hitting performance/packaging requirements (as opposed to hacker/enthusiast CPUs where they can slam it into freon cooling units and attach NASA wind tunnel fans for cooling).