KGI: Apple is designing its own processors for Mac

“A KGI report predicts that Apple will begin using its own processors for Macs ‘in the next 1-2 years,’ with a specific prediction of a Samsung-fabbed A10X chip powering at least one Mac made in 2016,” Ben Lovejoy reports for 9to5Mac. “The wording appears to suggest an entry-level machine–possibly a future model of the 12-inch MacBook Air.”

Apple may launch Mac products that use own AP [Application Processor] in next 1-2 years. This prediction is based on the assumption that Apple’s self-developed AP performs at a level between Intel’s Atom and Core i3 and is good enough for Mac. Using self developed AP can help Apple better control the timing of Mac launches and Mac product features. – KGI Research

Lovejoy reports, “With performance between an Atom and Core i3, the chip would not be suitable for mid- to high-end Macs.”

Read more in the full article here.

MacDailyNews Take: iOS devices and OS X Macs inevitably are going to grow closer over time, not just in hardware, but in software, too:

Think code convergence (more so than today) with UI modifications per device. A unified underlying codebase for Intel, Apple A-series, and, in Apple’s labs, likely other chips, too (just in case). This would allow for a single App Store for Mac, iPhone, and iPad users that features a mix of apps: Some that are touch-only, some that are Mac-only, and some that are universal (can run on both traditional notebooks and desktops as well as on multi-touch computers like iPhone, iPad, iPod touch, and – pretty please, Apple – Apple TV). Don’t be surprised to see Apple A-series-powered Macs, either.MacDailyNews Take, January 9, 2014

[Thanks to MacDailyNews Reader “Dan K.” for the heads up.]

Related articles:
Apple A9-powered MacBook Air? – December 16, 2014
Why Apple will switch to ARM-based Apple A-series-powered Macs – August 27, 2014
Intel-powered Macs: The end is nigh – August 4, 2014
Intel’s Broadwell chips further delayed; not shipping for most Macs until early-mid 2015 – July 9, 2014
Apple will inevitably drop Intel for their own A-series processors in the Mac – June 26, 2014
How long before Apple dumps Intel from MacBook Air? – June 26, 2013

34 Comments

  1. This rumor has been going around for about six years now and it looks bogus as ever. One of the major useful things about Intel Macs is that you can run Linux, Windows and Mac OS X and they huge software libraries available for them at native processor speeds. Besides, Microsoft already tried this and look at how well that turned out.

    1. The issue with its rumour is that even already released Apple A8X is already as powerful or even more powerful than Intel i7 per cycle now.

      So if Apple would want to produce CPU for Macintosh, they could have done it already last year by just re-designing the low-level layout of basically the same cores to spread things out a little bit so the eventual chip could run on higher clock speeds — say, 3 GHz instead of 1.3 GHz — such chip would be as fast or faster than any Intel’s chip.

      However, it would not make sense for Apple since Intel has exclusive advantage in manufacturing norms. While Intel’s competitors such as TSMC and Samsung can only offer 20 nm process, Intel already uses 14 nm.

      So if Apple would come up with its own chip for Macintoshes, due to worse manufacturing norms it would consume more energy than Intel’s Broadwell chip with the same actual performance.

      (This all does not matter much in mobile market since Intel has no real footing there. But it is critical for notebooks and desktops.)

      1. Intel x86 systems power 80% of the top 500 supercomputers in the world. IBM Power and AMD have most of the rest. How many of the top 500 supercomputers in the world run ARM? Zero. Nada. Zilch.

        Please show me the benchmarks showing that the A8X is more powerful than a Core i7.

        ARM chips are built on a low-power process. That’s why you don’t see ARM chips running at 4 Ghz – stuff that’s routine for chips with high-power processes.

        Say Apple improves their chips. Do you think that Intel is standing still? They have improved their chips according to their tick-tock cycle and it’s been a furious pace.

        My professional background is in software engineering working on browsers, databases and operating systems. I also dabble in computer architecture and I’ve done many ports to different operating systems and different architectures. I chat with lots of EEs that do chip design work and they tell me what can and can’t be done. They have to really dumb it down for me though, because I don’t have anywhere near the education and experience in their field so I rely on experts. With EE degrees, patents, papers and recognition.

        What’s your background?

        1. Any benchmark for CPU that runs both on iOS and on linux/Windows shows that A8(X) is as fast or faster per cycle as Intel i7.

          Besides, my argument is that due to manufacturing norms difference between TSCM/Samsung and Intel Apple’s high-clocked version of A8 would consume more energy than Intel’s Broadwell CPUs in similar performance no matter what Apple does. So I am not sure what you are arguing about.

          How many ARMs in top 500 is irrelevant to discussion as I discuss non-existing hypothetical version of Apple’s CPU that some rumours claim could/should be made by Apple instead of CPU, and why it makes no sense.

          My background is also irrelevant (though I have practice with more than couple of dozens of programming languages in thirty years).

          1. A CPU benchmark should compensate for operating system factors. So ideally you’d want the same compiler, browser, optimization levels, operating system and diagnostic modes. Otherwise, you’re comparing apples to oranges. Intel has a bunch of high-performance software libraries that greatly accelerate a variety of operations. They also have arguably one of the best performing compilers around. Do benchmarking sites use these packages? Do they use these compilers? These development tools aren’t cheap so a casual benchmarker probably wouldn’t have access to these tools.

            If Apples A8 chips are less efficient than Intel’s, then why should Apple bother?

            You can make any claims that you want to about hypothetical systems. But they are fairly useless without broad expertise in hardware and software.

            Programmers are a dime a dozen these days. How many Computer Science degrees do you have? I work in a building where the vast majority has a Phd or MS in CS. I know a lot of people with advanced degrees in EE too.

            1. Geekbench uses all ways possible to optimize its testing tasks for every platform, including Intel. Though besides this test there are many others which also show about the same thing: A8(X) is about as fast or faster than even the very best and latest Intel CPU — per cycle.

              A8(X) design is not less energy efficient by itself, it is manufacturing process that limits up-scaling it to desktop class CPU without making it less competitive to Intel. Intel has unique advantage and it does not share it with anyone, it only uses it for itself and does not allow contractual manufacturing of microelectronics on its best process. This is the reason why I wrote that there is little sense for Apple to make CPU for Macintoshes itself.

              And, by the way, I still do not quite understand what you are arguing against.

    2. Michael, you’re thinking all or nothing. But they’re not going to shift all the computers all at once off Intel. If history is a guide they will make one blur-the-lines device that draws from iPad and MBA. It’ll be desktop class with a hardware keyboard, but also a touch screen. It will run iPad apps, but will also open up a huge door to taking those apps further. Swift encoders may also be updated to allow compiling of the code for iOS or Intel, to encourage people to begin updating their desktop class apps.

      Look for a hardware must-have, like a security or login tie-in with Watch to sweeten the pot (like the watch automatically logs you in as you approach the computer, and with an iCloud synced desktop so that wherever you go the layout is the same and the same apps are available).

      This definitely is in the cards. In fact, there’s no way this doesn’t happen.

      1. Touch is here today as still is the mouse.

        Siri is the next UI for Apple – its been said.

        The real change comes from IBM and APPLE with artificial intelligence and Siri. Once this happens the convergence of iOS and OS X shall occur also. Desktops will no longer require the mouse and portables will no longer use touch.

      2. Apple likes to keep their product lines small and having to support two different architectures is inefficient. You could argue that it would be easier for Apple to go with Intel for iOS because you develop and test Apps on x86, then port to ARM.

        Have you ever ported complex software, say 10 million lines of code from one architecture to another and then had to support both? Not something that you really want to do. The Apple-Intel partnership seems to be working just fine these days – why rock the boat?

        iOS currently doesn’t support the concept of a mouse pointer device and I’ve seen no plans for said support. You’re not going to get professional class tools that are touch only because a lot of work requires precision and working with fine detail.

      1. Yes running Windows has been a huge boon to adoption of Macs, however I think that in about 3-4 years it should be possible to create a massively parallel multi-chip A-series Mac, that can out perform Intel chips even in virtualization.

  2. I would expect that Apple would release ARM powered Macs alongside Intel powered Macs and then derive reliable figures about demand, rather than listen to noisy on-line comments for and against the idea. I would still expect Intel CPUs to be used in Macs for many years to come, but that ARM Macs would also employ software emulation, just as when Apple previously changed CPU.

    While a lot of users need Intel CPUs, a lot of users certainly don’t and an A series CPU could provide a very attractive and cost-effective solution for those customers.

    For many office workers, such an approach could offer a great computer at a sensible price as their computing needs would not require a particularly high performance CPU.

    1. ARM chips are slower than Intel chips and you want a 5-1 performance penalty on top of that? That would work about as well as Windows/RT.

      If ARM chips had the advantage, why are there so many Chromebooks with x86 processors?

  3. This would be dumb as Windows RT. Now if Apple wants to make iOS device w/ a keyboard and trackpad fine. A15 or Apple version A8 ARM CPUs are just now in quad core platform hitting Intel i3 levels on most mainstream processing. The same i3 CPUs Apple will not use because they are too slow.

    RISC vs CISC is over. ARM has adopted some CISC sorts of stuff like NEON (aka SSE or MMX) and Intel has RISC based core. I don’t get why any user thinks moving the Mac Platform to ARM is a good thing. Macbook Air’s already have 12 hour battery life. Plus you loose the x86 compatibility. I don’t care if you hate Windows or Linux they are here for long time. Even setting that aside why buy a macbook air with 10 – 20% more battery life and 1/3 the speed?

    Bottom line is ARM has yet to hit Intel/AMD in absolute performance. The desktop is different than mobile or even server. I think ARM will be competitive in the Data center for some applications. The benefit being able to up your rack density with fewer heat issues. Intel may yet be competitive there.

    Since all iOS Apps are developed on a mac. I can see a decision like this have a reverse halo effect. Hurting the iOS devices too as developers won’t want a single use development machine.

    1. Yes, CISC vs RISC is a dead horse. Read up on the architecture manuals from either AMD or Intel to see why. Intel has started designing custom x86 processors for a few large tech firms like Oracle and Amazon and that may be a trend if the customer is large enough.

      Rackspace recently reported that they were going to buy a bunch of Power systems from IBM for hosting but I’d guess that a customer specifically requested them. The server world is still massively x86.

      1. Why are you telling to read up on CPU arch. I have read the Intel PII, AMD Athlon and Athlon64 manuals cover to cover and know the difference.

        Although I am not a big fan of Rackspace. I have had clients use Power Arch in certain things. As I briefly said ARM in certain cases may have a use in DC enviros. I have Odroids and Raspberry Pis. What the Arm arch can do on mW or W is impressive, but not in terms of absolute performance. It comes nowhere near the i5, i7, A6(amd) or A8(amd).

  4. One of the main reasons why apple are selling 5m macs a quarter is because of windows compatibility. It allows me like many others to have our home computer able to do both personally and work activities.
    So I am all in favor of non-Intel macs but they have to have real compatibility with windows and I am not talking about emulation either.

  5. It doesn’t really make sense if it’s just for low end Macs. Apple would be wasting a lot time and energy just to turn Macs into something that competes with Apple’s iOS devices.

    To start replacing Intel in Macs would only make sense if the chips could scale to high end computing. Macs are the trucks in Apple’s post-PC landscape. If all you need is a cheap or low power computer, then Apple would likely prefer you use an iOS device instead of a Mac.

    iOS is huge computing platform, with apps and hardware accessories that allow them to do just about anything. It’s just not worth trying to make Macs compete directly with them – iOS computers are already best in that category.

    Macs need to maintain its own unique reason for existence – to keep doing its own thing – that’s not cheap low powered computing.

  6. I can see it happen. Macintosh is no longer anywhere near the major breadwinner for Apple. It will always be important, increasingly more sentimentally than strategically, but it simply isn’t a crucial component for the company. Being a hostage to Intel’s roadmap for a product line that isn’t really your primary revenue driver is becoming an increasingly significant problem. Leaving Intel for home-brewed processors would eliminate that problem.

    Apple knows exactly how to make, market and support non-Intel Macintosh; they had done it for over 20 years (before switching to Intel). They can easily do this again. While there may be some users who occasionally use Windows (I used to be one), many don’t need to, and losing the ones that occasionally do may be the cost worth incurring for Apple. Regardless of whether we’re talking RISC or CISC, if Apple chose to make their own chips, they would be designed for performance that is better than what they get today with Intel (otherwise, it would make zero sense to migrate).

    Ultimately, I am still convinced that Apple is on a path to merge the two platforms (iOS and OSX) into a single platform, leaving the old keyboard-mouse-display paradigm behind and making all computing multi-touch driven — form iPhone, all the way to a 30″ iMac. The UI will be designed to know the difference between accidental touch (when resting your arm on the display) and an intentional command, the devices and their displays will be used horisontally, and keyboard will only be needed when we need to type large amounts of text (and can’t dictate it via voice).

    This is not quite out of the realm of possibility.

    1. Hostage is a strange term to a partner that has given you special parts over the years so that you’d have an advantage over your competitors.

      Apple left IBM because IBM wasn’t responsive to their needs.

      You are always hostage to someone. It could be TSMC (they’re going to have a conference to discuss yield issues I hear), Samsung (Apple’s best friend) or maybe Global Foundaries (they’ve had a lot of FinFet issues I hear).

      I think that some guy talked about integrating disparate things into a monstrosity in terms of a toaster and a refrigerator. That can go for operating systems too.

      It’s hard to talk about integration with those that don’t understand software abstraction layers (and hardware abstraction layers as well). There’s a really good video at MIT OpenCourseware in one of their EE courses that gives you the whole thing in one lecture. It helps to know a bit about electricity and magnetism.

  7. Apple will ONLY do this, IF using an “A-based” CPU allows Apple to design a Mac that is not possible using the latest Intel CPU. For example, to give it significantly better battery life, while significantly reducing size and weight.

    An A-based version of MacBook Air is the most obvious “guess,” but Apple could also create a new Mac mini that has the footprint of an Apple TV (and maybe about twice the thickness). It would be the “Apple TV that runs apps” that some people want, because it’s actually a “Mac with an Apple TV app.” The ideal “media center.” The Apple TV (which is only an Apple TV) continues priced at $99. New Mac mini starts at $299. That’s something not possible using Intel.

    Apple will NOT do this just because an “A10X” (or whatever iteration) has become powerful enough for use in a Mac, if the resulting Mac looks and performs mostly like the Intel-based Mac it replaces. There must be a clear and obvious reason… Otherwise, just keep using various Intel processors in all Macs.

    If Apple starts using its own processors in some Macs, they will still be Macs. When Apple moved from PowerPC to Intel (and from 68K to PowerPC before that), the user interface did not suddenly change. The Mac user experience is NOT defined by the flavor of processor.

  8. Just a guess – but – I would say the A8 M8 chipset, which is claimed as being the 64bit desktop class for a cellphone is already in-between, the Atom and the i3.

    For Apple to want control over its entire line – makes sense. However wishing collaboration with Samsung for this Chip is insane. Why why why?

    Beyond that, talks of A9 M9 are already well in development, perhaps a A10 M10 can run OS X.

    And lastly, good enough, thats not the Apple I knew.

Leave a Reply to alanaudio Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.