• Explore the magic and the mystery!



  • So What’s Apple to Do About the Slipping Intel CPU Roadmap?

    April 4th, 2017

    When Apple released the Late 2016 MacBook Pro, one major criticism was about using an older processor, Skylake, instead of Kaby Lake, the latest and greatest Intel silicon. Apple also got dinged for its decision to limit memory to 16GB instead of 32GB.

    There were other criticisms, but Apple was constrained by Intel’s roadmap and its inability to release new CPUs on time. So quad-core Kaby Lake chips didn’t ship until weeks after the MacBook Pros went on sale, and it’s not as if Apple just swaps out hardware on a whim. Parts are ordered far in advance to meet production schedules, and, no doubt, to get the best prices. It’s not the same as some of those Windows boxes, where new model releases often do not make much sense.

    So why not 32GB? Well, aside from the fact that no previous Mac notebook supported that amount of memory, Apple said it would have to use a RAM controller that consumed more power, and there would be less space for the battery. Consider, also, that the MacBook Pro has already been criticized for subpar battery life for some users, and this move would surely make matters worse.

    Indeed, neither Kaby Look nor its successor, Coffee Lake, offer support for mobile LPDDR4 memory, the power efficient kind, which also argues against upping the RAM limit this year.

    So when will Apple be able to make the switch? According to published reports, the 10-nanometer Cannon Lake processors, which will support LPDDR4 and 32GB, may not ship until the end of the year, and possibly not till the beginning of 2018. What this means is that it these parts won’t be available for this year’s expected MacBook Pro upgrade.

    That won’t stop the critics from suggesting that Apple simply offer an optional model for those who need 32GB, even if battery life suffers. While it is certainly a legitimate criticism, that Apple ought to offer more models or model variations to better suit the needs of customers, that ship sailed long ago. Back in the 1990s, the Mac lineup became so complicated that Apple’s own executives reportedly had difficulty figuring out precisely what was what, and which market a product was meant to serve.

    Indeed, model proliferation continues to inflict much of the rest of the tech industry, where even trying to buy a new TV set can become a confusing and befuddling process. Take VIZIO, which offers TVs in five separate lines, each of which provides multiple sizes. Do you want a 50-inch set or the 48-inch counterpart? Why should there be such fine distinctions? Unless you place them side-by-side, would you even notice?

    Now when looking over Apple’s product lineups, there’s a healthy level of simplicity.

    The real problem here, though, is Intel. Apple is constrained by Intel’s ability — or inability — to get its chips out on time. So if Apple wants to refresh a Mac, it has to use older chips, as they did with last year’s MacBook Pro upgrade, or just wait. But customers who grow antsy waiting for their favorite Macs to be refreshed may not always give Apple the benefit of the doubt here. And I understand the concerns about the fate of the Mac Pro, where Intel’s parts, and the GPU, have been upgraded during the past three years.

    Now it’s not as if Apple can just browbeat Intel into getting its work done faster. Rushing chips into production may result in buggy silicon or other issues. Better for Intel to attempt to get it right the first time.

    So what’s Apple’s choice?

    One might be to consider AMD’s Ryzen, a new processor family that promises to match or exceed the performance of Intel silicon at roughly half the price. It probably wouldn’t involve a whole lot of development effort to make this change, since AMD is designed to be x86-64 compatible and will run the same operating systems and apps. But this new chip family may have problems that AMD is working on, such as possible glitches in lower-resolution gaming performance.

    But if Apple could move to AMD in place of Intel, it could mean a decent cost reduction. Imagine a high-end chip, priced at $1,000 from Intel, being available for $500 from AMD. Apple could pass the savings onto customers. But remember that Apple is buying chips by the tens and hundreds of thousands at much lower prices; the cost reduction would be far less, but still significant.

    Another solution for Apple would be the wholesale move to ARM-based hardware, those A-series chips that have begun to approach the performance levels of mid-range Intel-powered notebooks. In the meantime, Apple has given notice to its provider of mobile GPUs, Imagination Technologies, that they are going to stop using their intellectual property soon. With a growing staff of GPU engineers, including some recruited from Imagination, Apple is likely taking development in-house.

    Apple uses an A-series SoC to power the MacBook Pro’s Touch Bar, and might be expanding support to control other functions, such as Power Nap. That’s not the same thing as a wholesale processor switch. Would it even be worth putting developers through another processor transition? And don’t forget the many Mac users depend on the ability to run Windows at full speed with Boot Camp, or with good performance via a virtual machine. Could Apple build an Intel emulator with its A-series silicon that would provide minimal loss in processing power? Could it be done by passing the conversion through the GPU, which will evidently now be developed by Apple’s in-house team?

    I wouldn’t care to guess. I’m assuming that, so long as Intel can continue to improve its chips at a fairly steady pace — even if it runs late — Apple isn’t going to switch to its own chips. Using AMD parts, however, doesn’t seem to be out of the question.

    But with Apple, it’s hard to predict what it might be testing behind the scenes.



    Share
    | Print This Post Print This Post

    2 Responses to “So What’s Apple to Do About the Slipping Intel CPU Roadmap?”

    1. DaveD says:

      Hard to believe that it was nearly twenty years ago when AMD was giving Intel good competition. Not only did Intel adopt AMD 64-bit instruction set, it had to rethink the Pentium. After Intel got a good foundation with the Core processors, AMD had been stumbling for a long time. I do hope AMD’s Ryzen is successful enough to shake Intel. If so, it would be nice to see the Mac mini have it.

      • gene says:

        Well, whatever Apple does, it won’t happen probably till fall.

        From Phil Schiller’s announcements, it appears the next iMac is months away, although it seems they will be incorporating the changes/additions I’ve suggested.

        Peace,
        Gene

    Leave Your Comment