• Explore the magic and the mystery!
  • The Tech Night Owl's Home Page
  • Namecheap.com





  • Revisiting Mac on ARM

    April 6th, 2018

    I have lived through all the major Mac processor transitions. Makes me feel old. First it was the Motorola 680×0 series, followed by the PowerPC and, by 2006, Intel.

    Overall, the last one went pretty well. There was a way to run PowerPC software for a few years, courtesy of something called Rosetta. It was pretty decent from a performance standpoint, unlike the 680×0 emulator, which suddenly put you a couple of generations behind in terms of how well the apps ran until they went PowerPC. But until the new apps arrived, the all-new RISC architecture didn’t seem so impressive.

    So is Apple planning yet another processor switchover? Well, consider how Apple has managed to deliver its A-series processors with huge performance boosts every year, very noticeable with most apps.

    Compare that to new Intel processor families that might be measurably more powerful than the previous generation, but the performance advantages are often barely noticeable without a scorecard. Apple’s advantage was to create an ARM-based processor family that took direct advantage of iOS. It wasn’t bogged down with legacy support for things that never existed on an Apple platform, making for more efficiency.

    So does Apple have a Mac on ARM in its future? Microsoft tried Windows RT (on ARM) without a whole lot of success, but perhaps its second try will fare better.

    Using Apple’s Xcode, it shouldn’t be such a big deal for developers to go with the transition to ARM, and allow developers to build flat binaries for that and Intel. Recent rumors have it that you’ll be able to run iOS apps on Macs, and vice versa, more or less. The Touch Bar on the latest MacBook Pros runs with a second processor on that computer, an A-series system-on-a-chip. A similar scheme is used for low-level functions on the iMac Pro,

    So Apple is clearly taking you partway already. How long will it require for a full shift, and should you such a possibility seriously?

    It’s a romantic ideal, that Apple has full control of more and more of the parts that make up its hardware. It would also allow the Mac to offer far more differences than just a higher-priced PC in a fancy box.

    According to recent reports from reporter Mark Gurman of Bloomberg, the prospective shift may happen beginning in 2020. Take it with a grain of salt for now.

    But can an iPhone or iPad chip really power a Mac with equal or better performance than current models? Consider the benchmarks that show Apple’s mobile hardware exceeding the performance of most notebook PCs and coming up real close to the MacBook Pro. No doubt those CPUs are not running full tilt to lower the drain on resources and battery life. What will those benchmarks be if Apple allowed them to run full bore?

    What about the chips shipping two years from now? Remember, too, Apple already has control of graphics hardware, so what happens to its existing partners, AMD and NVIDIA? Apple probably wouldn’t care if its taking these steps.

    It wasn’t so easy for Apple to persuade developers to adopt PowerPC, but far easier to go to Intel, since there was so much legacy software on the Windows platform. That meant that many developers knew how to optimize their Mac apps for Intel. As I recall, it wasn’t such a difficult move.

    But there was one key advantage of Apple going Intel, other than being assured of regular improvements, more or less, in the chips. It was the ability to run Windows natively with Boot Camp, and at pretty good speed with virtual machines courtesy of such apps as Parallels Desktop.

    If Boot Camp and virtual machines have to run in emulation on one of these new fangled Macs, how much would performance deteriorate? Or would Apple devise ways to work around this, such as licensing some Intel chip functions using the graphics hardware to reduce the performance bottleneck? I would be loathe to predict how it could be done, but if the ARM chips end up significantly faster than Intel counterparts, maybe most people won’t notice much of a difference.

    It wouldn’t take the infamous performance hit of running Windows under emulation the PowerPC. That was just dreadful. I remember opening a document would often take a full minute or two.

    Some suggest that Apple, which has often ditched older technologies without apology, might just give up on the concept of running Windows on a Mac. But I suspect lots of users still need that feature. I also suspect that Apple is quite capable of devising a solution that wouldn’t hurt performance in any particularly noticeable way.

    But this all needs a reality check. That Apple could make this change doesn’t mean it will. It might very well be that Intel’s existing hardware roadmap is a viable solution, without saddling Apple with the development costs of a new processor transition. But there are good reasons for consistent hardware across its major platforms. If the annual improvements in Apple’s A-series CPUs continue to provide healthy two-digit performance boosts, maybe it will happen after all.

    I’m skeptical, but with Apple, never say never, particularly if Intel confronts any serious headwinds in improving its chips going forward.



    Share
    | Print This Article Print This Article

    3 Responses to “Revisiting Mac on ARM”

    1. Max march says:

      Well intel has been stuck at 14nm fir a loooong time. 10nm doesn’t even have a timeframe at Intel which is crazy- they usually had 5 years out. It could be a bifurcation where iOS aficionados get A hardware, and no Windows, pro level peeps get intel with A series, so more expensive to emulate. Cheaper mags are A series only, and reliance on windows is reduced. I don’t know many people who run emulation any more.

    2. James Katt says:

      Apple DOES NOT need to do an ARM Mac CPU. This is what some people fear. But Apple can irrevocably and perpetually license the IP necessary to build an Intel Compatible CPU just like it does with ARM. It can either 1) BUY AMD to do this 2) INVEST in AMD and get the license or 3) get the license from INTEL itself – in exchange for Intel building the CPUs for Apple – essentially becoming a foundry for Apple just like TMSC and Samsung are.

      What is intriguing is that with a license to both ARM and INTEL CPUS Apple can custom build its own BLEND of the CPUs for its Macs. These would of course be state of the art and faster than any existing Intel or AMD CPU given Apple’s chip designer’s hellbent penchant for high performance.

    Leave Your Comment