6 Comments

Summary:

Tech blog SemiAccurate sped up a slow news Friday with a so-crazy-it-might-be-true rumor that Apple will be switching CPU architecture. Again. Right now, it seems impossible, but given time, could Apple really use in-house designed ARM-based chips to provide the processing power behind Mac computers?

rumor_apple_arm_cpus

Could this be a WWDC Keynote sent from 2012?

Tech blog SemiAccurate sped up a slow news Friday with a so-crazy-it-might-be-true rumor that Apple will be switching CPU architecture. Again. According to SemiAccurate “moles,” Apple is planning to transition its laptops to ARM architecture in “2-3 years,” and “presumably” desktops, too, though without a time frame. The plan is a “done deal” according to the site.

This seems hard to believe. While it’s arguable the Apple A5, derived from the ARM Cortex-A9, can keep up with Intel’s Atom in a netbook, there’s simply no comparison between ARM chips and the Intel Core CPUs used in Mac laptops. Of course, in 2013 it wouldn’t be the A5 in a hypothetical ARM-based MacBook Air, but a CPU derived from the ARM Cortex-A15.  Based on a 32nm fabrication process, with up to 8 cores and clock speeds up to 2.5 GHz, as well features like virtualization and access to the 64-bit ARM instruction set, the A15 certainly packs enough keywords to qualify for a Keynote presentation on paper. Unfortunately, the most salient benefit of ARM architecture, power efficiency, will diminish as computing potential increases. Is the diminishing difference worth another major architectural transition for Apple, one that would undoubtedly require a vast amount of engineering effort?

Even should the computing potential of the A15 be sufficient for mainstream use in 2013, software written for current x86 Macs would either have to be rewritten or run in some kind of emulation mode. It’s hard to imagine Rosetta 2 running x86 applications on ARM being the engineering miracle the original was when running PPC applications on Intel Macs. That would leave Mac users waiting months, possibly years, for ARM versions of applications like Office for Mac or Adobe Photoshop. Considering the incredible success Macs have enjoyed since transitioning from PPC to x86 architecture, this rumor seems ridiculous.

And yet one can’t discount the possibility of ARM extending beyond mobile; certainly Microsoft isn’t. As much as Apple has grown in traditional computing over the last five years, x86 PCs running Microsoft Windows sell about 19 times as many computers in aggregate. That’s why Microsoft announcing Windows 8 will be available for ARM-based systems matters. If Microsoft sees value in the platform, then Apple, with its growing mobile focus, no doubt does as well.

Also important, if true, is the rumor reported by EE Times that Intel is seeking to become a foundry for Apple fabricating ARM chips. One could argue that Intel doesn’t care what kind of chips it sells Apple, as long as they come from Intel.

Regarding the problem of applications, while current Mac apps wouldn’t run on an ARM MacBook, it could open the library of apps that run on the iPhone and iPad. Imagine a touch-screen MacBook running Angry Birds. Considering iOS devices outsell Macs by a ratio of six to one and growing, there’s a good argument that iOS developers could quickly take up the app slack for ARM-based Macs. Apple is reportedly making the Mac App Store the “preferred” method of distribution for OS X software beginning with Lion, so that could give it greater control when it comes to handling a chip architecture transition with developers.

Finally, there’s something to be said for Apple’s desire to control its own destiny. It banned Flash from iOS, for example. Rants about security and power management aside, Apple doesn’t want Adobe controlling any part of their mobile operating system. Considering Apple’s painful CPU history with Motorola (now Freescale) and IBM, and most recently the GPU spat between Intel and Nvidia impacting Macs, underestimating Apple’s corporate obsession with controlling the “whole widget” would be a mistake. A transition to Apple CPUs would give the company much more control over its own product, in the same way the A4 and A5 chips have done for its mobile devices.

SemiAccurate has proven itself fairly reliable with similar reports in the past, including the prediction that Nvidia would eventually disappear from Mac computers, and that Light Peak would make its way to Apple products. TUAW points out the original story on SemiAccurate is tagged “humor,” but a quick look at how that tag is used in other stories doesn’t indicate that site isn’t serious about the claims made regarding Apple’s intended use of ARM. We contacted Apple for comment, but have yet to receive a response back.

What do you think? Is this latest rumor crazy or conceivable?

You’re subscribed! If you like, you can update your settings

  1. Apple has been pretty pleased by Intel. It would be foolish to alienate a good business partner especially when they came out with the the 3D transistor technology, which will most possibly be the next step to the evolution of the microprocessor, and show why they are the leaders in their field. Apple will not switch anytime soon.

  2. Nope. Shut up.

  3. If they do this maybe they will include both ARM chips AND intel chips. That would ensure compatibility of all Apple applications (mobile and mac).

  4. Lucian Armasu Friday, May 6, 2011

    I don’t know about the others, but I know in 2013, Nvidia should have Tegra 5 as a 2.5 Ghz 8 core Cortex A15 chip (quad core one in 2012 – Tegra 4). The thing is it won’t use more power. Every single ARM chip maker is doubling the performance every 12 months while keeping the energy consumption the same every year.

    To me this rumor makes perfect sense. Apple will transition its other computers to ARM. The keyword here is “transition”, though. It will only use it in a low-end notebook at first, like the Macbook Air (or perhaps a new notebook), and then transition ARM to Macbook Pro, and maybe even iMac by the end of the decade.

  5. It’s not so hard to believe. ARM were producing desktop class chips in 1988. I personally owned a desktop machine powered by one of these early chips and lusted after the ever more powerful ones that came later. Apple is the perfect partner to bring ARM back to this market.

  6. The name “SemiAccurate” vastly overstates Demerjian’s reliability. Predicting that some GPU brand will disappear from some product line is like predicting sunset; this happens periodically with complete certainty. On anything less inevitable, SemiAccurate’s reliability lies somewhere between that of the National Enquirer and the Weekly World News.

    This story is ridiculous. There is no architectural advantage to ARM, as there was to Intel when Apple switched away from PowerPC. ARM’s chip makers have no plans to compete directly with Intel on performance, so switching would kill Apple’s premium product lines. Apple doesn’t make these decisions years in advance anyway; the company is well known for keeping its options open, but the decision is made when the hardware is ordered, and that can’t happen for years anyway.

    The reactions to this story from those who are expected to know better have been utterly pathetic.

Comments have been disabled for this post