Apple ponders Macs sans Intel chips

Apple ponders Macs sans Intel chips

Summary: Apple is pondering a move to non-Intel Macs. If Apple follows through it would give ARM a nice computing win.

SHARE:

Apple is reportedly pondering a move from Intel-based Macs to its own processors.

According to Bloomberg, Apple is exploring ways to replace Intel in the Mac. The aim is to use the technologies used in its iPhone and iPad products.

The report is based on unnamed sources, but the move isn't all that surprising. Apple designs its own processors.

But the big push here is that the Mac could become ARM based. ARM is an architecture that competes with Intel's x86 architecture. If the Mac were to go ARM, it would give the architecture a big boost in PCs. ARM chips from the likes of Qualcomm and Nvidia are typically found in smartphones and tablets.

Should Apple go with its own chips the winners and losers would look something like this:

  • Intel would lose a high-profile account with Apple.
  • ARM would likely be a winner.
  • Apple could either win or lose. Should chip processor speed begin to matter---right now good enough seems to be the motto of the day---Apple would find itself on a processor development treadmill.

Keep in mind that Apple used to run on chips manufactured by Motorola and IBM, but then switched to Intel. Those transitions---should they be necessary---can hurt.

Topics: Processors, Apple, Hardware, Mobility, PCs

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

65 comments
Log in or register to join the discussion
  • Oh...

    This again. Weren't there rumors this was happening over a year ago?
    Big Sparky
    • This Bloomberg article is derived just from Timothy Cook's comments about

      ... Robert Mansfield staying with Apple and managing all of wireless and SoCs electronics development teams, naming the plans for the latter as incredibly ambitious for the future years.

      One of the ways to interpret those words and this Mansfield role is that Apple is going to produce incredibly complex, but still power efficient (thanks to PA Semi unique patents) SoCs, CPUs, GPUs to replace Intel's chips in the future.

      However, with information that is known for today, there is no way to confirm whether this is correct interpretation or not.
      DDERSSS
    • Spot on...

      It was the same all through the 90's; apple was going intel. That didn't happen until the multi-core era.

      I think I've heard this rumour each year since '09. However I think that this has some weight. What went largely unreported during their reshuffle last week was an increased importance in their chip development.

      The A6 has made it very clear that apple intends to perfect it's arm platform. They've also been very clear about merging ios and mac os... OS 11 is likely to rub on both I'd imagine. Likewise arm does give them bennifits in the area of heat production and power use for their ultra low profile desktops. As for the render-hog mac pro's RISC architextures can easily offer intel topping performance, but arm is unlikely to be the solution. To give you some figures the A6X chip gets benchmark scores around 20% lowwer scoring than a 2006 core 2 duo. For mobile devices, this is an excellent comparision. For a desktop it's abysmal. It's no small step to take that A6 technology and scale it up nearly four fold to replace current ivybridge chips.

      All in all it makes sence for them to do it if they can; they're a proprietary platform. Being able to design the cpu and os makes a lot of sense.. It just remains to be seen if they can do it better than the biggest chips makers on earth can.
      MarknWill
      • Intel is now a dead duck, and so is Microsoft.

        Apple will move to ARM.

        Yes, you can cite better performance from an Intel chip at the moment, but there are two factors that you must consider.

        First, is that Apple is designing its own ARM-based processors, and will likely release a desktop-optimized version that beats current power-saving processors found in phones.

        Second, ARM offers more processing power per watt of electricity. ARM is scalable, and will soon be used in vast server farms. Apple can create a special version for desktop computer use that contains many more cores than the phone version has.

        Third, with the world switching to ARM, there'll be less money invested in Intel x86 processors, so development of x86 will slow down. Flushed with cash, ARM development will speed up.

        The balance of power is changing, and that is going to be the ruin of both Intel and Microsoft, as the world switches to ARM (Microsoft's Windows RT on ARM will be irrelevant, has no apps and will not sell).
        Vbitrate
        • Sorry, I may not have been clear...

          I wasn't promoting intel, x86 or microsoft. Similarly I wasn't bashing the Arm architexture. I was simply outlying the challenges involved in the proposition.

          I think it's good to note some key definitions here; intel, microsoft and apple are companies. X86 and arm are processor achitectures and risc and cisc are approaches to chip design.

          As this article is about Apple moving it's desktop chips away from intel I was highlighting some bennifits and challenges in the move. As a closed ecosystem, apple have demonstrated on more than one occasion that they are able to transition their ecosystem across architectures. This isn't easy - all of your software becomes obsolete. Now for android and windows apps this isn't a problem as the software runs in a type of vm so it runs on anything the vm will run on. On a desktop this isn't such a sensible aproach as you sacrifice performance for compatibility. This means that Apple would most likely be looming to maintain a common architecture across mobile and desktop platforms to get the best performance out of their machines.

          As a previous powerpc and sparc user, I'm a big fan of RISC on the desktop; it does provide typically smaller chips, producing less heat on lower power. However it is not without compromise; CISC chips such as the x86_64 platform do perform certain complex processing tasks better than RISC chips as they can complete several of these tasks with less cycles. This is where I referenced the mac pro. It is important for apple to provide a machine with top class calculation power. This is why the majority of the most powerful computers run CISC chips even though they don't run windows.

          My point was that it is definitely posible for the risc chips to achieve this and try again to take the desktop market. However it may not be as simple as growing arm. Arm is the mobile solution of choice because of it's power and heat - centric design. On the desktop both of these hold less relevence, so I would expexct to see apple continue to develop away from the arm norm to maintain a common architecture that can do both.

          Of course if they plan to remove focus on the pro line and stick to imac and below desktop computing, it becomes a lot more appealing to them. One of the reasons gaming consoles use powerpc currently is that they onow exaclty whats in the box and what it's for. This makes it much easier to overcome the downside to RISC and embrace all it's strengths. Similarly ape controls it's entire ecosystem. I'm for this move, but I doubt it will happen over night, and would expect to see apple stick with intel troughout haswell's socket life.

          Just to highlight a couple of misconceptions; current plans to impliment arm in serverspace is actually in low processor power tasks where the energy and heat savings yeild large cost cutting (in addition to inexpensive outlay) it's not quite ready to replace x86 for the intensive server tasks.

          Additionally x86 is developed direct by amd and intel, they make the chips and sell them. Arm is developed by arm holdings and licenced. This does allow for a far greater interpretation ans editing - A chips, tegra chips, snapdragon, etc. but it does mean that you don't get the direct to market aspect and it's more cost appealing to chip makers not neading the latest and greatest to buy previous generation licences.

          Finally just to make it clear Ms and intel are companies. MS make software. They are unlikely to leave themselves exposed to the death of x86 should it happen; they are already embracing arm on their budget tablets. Similarly intel are a chip giant. They sell chips and make a lot of money from their platform. However again, they've experimented with RISC based chips countless times over the years, and have even licensed Arm thrmselves. If they find themselves unable to defend the x86 platform, you can bet they'd quickly impliment a new way to make money.

          Personally i like the idea of more choice - more chip makers making a common platform (probably arm based) and more software platforms running on it. I don't really want to see a return to the days apple run one architeture, sun run a similar but different one, ans all home pcs run another; as a multi os man, it's a pain!

          We're seeing some very interesting and exciting advances in the arm architeture and the prospect of x86 finally being replaced is very exciting. However it's not wuite that simple, and we aren't there just yet.
          MarknWill
        • That's inaccurate...

          1. ARM at its core is not designed to match the performance of x86. A "desktop-optimized" version is simply not feasible. I think Apple knows better than to move on to ARM. The switch to ARM will cause Mac OS to lose compatibility with all previously released software. That's just simply not an option for virtually every owner of a Mac, especially schools and businesses. And no, Apple will not be able to create a new version of Rosetta to emulate x86 on the ARM architecture. As it stands, x86 emulation on ARM is extremely poor and can barely run Windows 98. The transition from PPC to x86 worked smoothly because the x86 architecture is powerful enough to emulate PPC, however ARM, no matter how many Apple logos you want to put on it, will never be powerful enough to smoothly emulate modern x86 chips.

          2. The age old myth of ARM getting more processing power per watt of electricity is false. Myth busted! You just simply have to look at the newest Intel Atom SoC's (codename Medfield). With Intel Atom, a single core x86 chip is able to match the performance of a quad-core ARM chip while keeping power consumption below 1 watt. I think you should be aware that more cores doesn't equal better performance. If you were to make a modern ARM chip with 12-cores, it still wouldn't be able to beat the performance of a Core 2 Duo and it'd barely match a single core Pentium 4. Your average ARM chip today has about as much processing power as a Pentium 3.

          Also, at higher performance tiers, Intel x86 is way more power efficient than ARM. If you were to scale ARM to the same performance as a Core i7 Ivy Bridge, the ARM would use about 2 times more power than the Core i7. ARM has no benefits in the high end desktop market other than its low cost. And even in the mobile device market, ARM is starting to lose what few advantages it has against Intel. You see, when it comes to desktop computing, especially in the professional world and high end gaming world, ARM is basically a toy. RISC architectures just don't cut it for a lot of high performance applications. CISC is able to perform complex instructions a lot more efficiently using less cycles than RISC chips can. The following example shows how each CPU would solve this simple math problem. Which is faster and more efficient?

          CISC (x86):
          2^3 = 8

          RISC (ARM):
          2^3 = 2*2*2
          2*2 = 4
          4* 2 = 8

          Also, ARM servers are only used for simple things such as file transfers and simple hosting. They cannot be used, however, for demanding server-side applications and virtualization as well other things like Hyper-V and so on. Again, ARM just doesn't have the performance to be useful in higher tiers of computing.

          3. The world is not switching to ARM. That's just an illusion. ARM is simply the only viable option around. Intel never really bothered to make any effort in the mobile market until now. Intel has the money and resources to achieve whatever goals it needs to achieve. That is how they were able to prove that x86 could indeed beat ARM in performance while maintaining the same level of power consumption.

          Once Intel starts making more of an aggressive push with their mobile Atom chips, it'll be the end of ARM... again. Don't forget, ARM is almost as old as x86 itself. Intel beat ARM once during the 80's and 90's. It can do it again.
          Generalkidd
          • A lot of what your are saying is true and I agree but..

            I agree with the general thrust but some of your calculations regarding power and heat seem problematic. The highest speed Xeons still use 130 TDP. That translates to 52 A15s at full power and full clockspeed. 52 A15s would have a higher benchmark than the intel for highly parallel loads. For single process heavy weight loads the Intel would still kill it.

            Also the A15 is faster than the fastest Atom. The Atom you listed is not the fastest because the fastest atoms consume way more power than 1 Watt.

            Typical operation of A15 has it consuming considerably less.

            ARM still has Intel beat in this category. It isn't a Myth.

            That said, I like you don't worry about Intel. Intel will compete and at the high performance end it will be way ahead for a few years at least.
            DevGuy_z
          • You forget that Apple was on RISC chips before Intel.

            That considered, when I first heard of Apple buying that chip design company, I forecast even then that Apple would eventually return to a proprietary chip and I still believe so. Even the modern ARM chips are as powerful--more so really--than the original 286/386 Intel chips. That means that given time, ARM chips will meet or exceed modern x86 chips and relegate the x86 to larger, heavy-duty processors scaled into super-computers. At that rate, a 2Ghz ARM could be twice as fast as a 2Ghz x86 simply because it only has to translate half as many different instructions.
            DWFields
          • Apple can't even do bugfree firmware

            and you want them to make proprietary CPUs?
            they cut & pasted cortex A15 to make the soc in the iphone5 and suddenly everyone thinks Apple has mastered nanotech. LOL
            warboat
          • What do they care about compatibility?

            They dumped support for all of their old hardware and software when OSX came out, why would it bother them to do it again? More money for them, when all the apple faithful respond by buying new systems. And they will line up for them around the block.

            Hey, then they can go back to the old "Wintel" knock-name that they had to abandon when they sheepishly had to move to Intel chips.
            garyleroy
          • RISC vs CISC

            But your CISC processer is still doing

            2 ^ 3 = 2 * 2 * 2
            2 * 2 = 4
            4 * 2 = 8

            It's just micro-coded to hide the intermediate steps. All other things being equal, RISC is faster than CISC because its always faster to do things in (comparable)hardware than in software.
            deanders
        • Putting more cores won't make it faster for many desktop work loads.

          On the server, ARM works for lightweight work loads. ARM doesn't compete at all where heavy horsepower is required. Adding more cores does nothing if the workload doesn't take advantage of it. And to make ARM compete you have to both increase cores and clockspeed which means more power. Clearly the advantages begin to weaken both from a power and heat and complexity standpoint.

          Next, the A15 (quad core) isn't yet being delivered in quantity. 64 bit (necessary for addressing large amounts of memory) probably won't be available in quantity for two years. Intel is 3 years ahead of the industry in manufacturing and their lead is manufacturing is increasing . They are both increasing speed and at the same time reducing power.

          I do think ARM has a chance but Intel's strength is manufacturing. Can the other's: Samsung, Texas Instruments make the same profit margins, probably not. Apple doesn't need to but they can't go faster than ARM in releasing new technology. They can deliver better SoC designs but they have to wait on ARM. And the designs are easier right now but after they start going to 64 bit and ECC DDR3 and more interconnects and such I imagine the time delay from when ARM releases a new design until it is implemented will get longer.
          DevGuy_z
        • intel

          intel used to produce ARM chips and stoppd. intel holds license to manufacture ARM chips again if it works out to their benefit. Remember that intel has been producing 22nm chips longer than some companies have been promising to produce 22nm chips.
          mswift1
      • Benchmarks don't matter.

        They're like polls during election season. Snapshots, interesting maybe, and worth having. But they don't drive innovation in a meaningful way.

        I'm not trying to be combative, I agree with everything you wrote, except I think a mobile SoC that is comparable to 2005-2006 era desktop chips is far from abysmal when you consider the complementary advances in storage and battery technology.

        What's stopping Apple from rolling out an ARM MacBook Air and letting the market decide?
        bamurphymac
        • Where would an ARM MBA fit in Apple's strategy?

          "What's stopping Apple from rolling out an ARM MacBook Air and letting the market decide?"

          Well for one thing, since the coloured iMacs, Apple has been very dilligent at pushing their product tiers (consumer/professional, portable/desktop). The current Macbook AIR already filles the consumer portable tier very handily and is selling quite well.

          In the current market, an ARM based MBA would not run any software sold at your local Apple store or Mac dealership, and would require a unique Mac OS or would have to run iOS. So basically it would be an iPad with a keyboard and no touch screen, and slower than the current MBA.

          In other words, you want Apple to build a machine that takes the worst aspects of both the iPad and the current Macbook Air and combine them?

          I can see a comparison to the Surface RT happening here, but what you're proposing would be even worse. You'd have the Surface, but without the touch screen, with less portability, and without the developer toolset in place that would allow useful applications to be built.
          daftkey
          • Where would an ARM fit in an MBA?

            Let's see.
            * Lower power demand: Increases battery life without having to make the batteries themselves any larger.
            * Cooler: Doesn't cook your lap when you actually do put it there.
            * Faster: Being a Risc-based processor gives you effectively twice the processing speed as an x86 Cisc processor. Considering how multiple-core chips are coming out for ARM as readily as for x86 now, a 4-core ARM probably gives at least as much total computing power as a single-core x86 while still using less energy. That means while you may use more cores, at lower prices and lower operating costs you'll still do just as much work more efficiently.
            DWFields
    • I'm willing to bet that if Apple does anything,

      they'll end up releasing their own discreet graphics chip based on PowerVR tech, considering that 1) it'll eventually kill the Hackintosh (watch for 10.9 to kill off Core 2 and first-gen Core i-series support), and 2) they wouldn't have to wait for NVIDIA and ATI to develop drivers for them, considering that buggy drivers seem to be a recurring issue with both vendors.
      Champ_Kind
  • Articles came out last year saying Apple would move to ARM by 2013

    For $2500, a puny ARM processor is not going to pique the interest of any technically-conscious customer that needs the power...

    It's a shame Adobe won't open up to Linux, or make it easier for OS X users of their CS6 products to migrate back to Windows. Yes, it can be done but it's a time-consuming chore compared to Lightroom and its licensing policies...
    HypnoToad72
  • And "good enough is good enough", followed by

    everyone whining how everything is cheaply made, breaks down easily, etc, etc, etc...

    Gotta love our "culture" nowadays...
    HypnoToad72
    • But then, I didn't _____ so I could profiteer from it

      Where ____ =

      * using lower quality parts that I told my workers to buy
      * offshoring the jobs
      * engaging in price wars to drive out competition

      etc
      HypnoToad72