A lot of things have already been said. Id like to talk about the most common things a little more.
Intel x86 will always be more powerful than ARM.
Sure, at least in near future. RISC-based architectures like PowerPC, Alpha, MIPS and SPARC were more powerful than CISC-based x86, and probably would still be today. However you dont see much of them around anymore - because x86 is
nearly as good, but significantly cheaper. The same goes for ARM, a 2.5GHz quad-core Cortex-A9 CPU might not be a wise choice for a MacBook Pro, but it most definitely blows the MacBook Airs 1.6GHz Core2Duo out of the water, while consuming 20% less power. This means also, that an ARM core could be enhanced to use those 20%, likewise by adding two additional cores, which should bring it to a performance level comparabile to a Sandy Bridge CPU in that class. The Quad-Core Cortex-A15 is already reaching
nearly the raw DMIPS performance as a Quad-Core Phenom II running at 3GHz - while using about 7% of the power.
The same nearly as good, but significantly cheaper argument goes for AMD K10 vs. Sandy Bridge as well, if you forget about the fact that synthetical benchmarks are most likely to
favor Intels architecture anyways. A i7-970 costs 2.5 times as much as the Phenom X6 1100T, though. Will most people even notice the difference in day-to-day use? Probably not.
Ivy Bridge with its Tri-Gate technology wipes the floor with ARM.
Sure it does, it brings advantages in either performance or power consumption of
up to 37% compared to second-generation i7 chips. However, youre comparing apples and oranges here. The smaller the manufacturing process, the lower the power consumption, and that translates to higher processing power at the same power usage by adding more cores or increasing the frequency of the CPU. The current generation of Apples ARM implementation, the A5, is manufactured in 45nm with classic transistors. Given the fact that this matches the Penryn variant of the Core 2 architecture, you have to compare these to the Cortex-A9 - and as stated, ARM exceeds the performance of the lower-end Core2Duos, while consuming 20% less power. A shrink to 32nm will probably be superior the low-end first and second generation i7 chips, while shrinking again to 22nm and implementing the Tri-Gate technology will do the same thing with Ivy Bridge. ARM states that Cortex-A15, the next generation of ARM CPUs we will most probably be introduced to as the Apple A7, which is still based on the 45nm manufacturing process, will be offering a 30% overall performance increase over the Cortex-A9 with other factors the same.
If Apple decides to go the ARM route for low-end MacBooks, they intend to ship in high volume. They are predestined to become the third large player in the CPU business behind Intel and AMD. Given the fact that Apple has incredible financial possibilities, there is nothing wrong with the idea of Apple teaming up with Samsung or any other chip manufacturer and developing the manufacturing process needed to build Tri-Gate silicon in an 22nm process. For the engineering of the CPU itself, Apple has already some knowledge in house (
PA Semi+
Intrinsity), its just a matter of extending it. Lastly, the Apple A5
costs $25, NVidias Tegra 2 comes in at $15, while the Core 2 Duo SU9600, found in the 11 MacBook Air has a hefty
$289 written on its price tag.
That leaves enormous room for Apple to invest in this area, and even if Apple does
not know how to make a $500 sub notebook that is not a piece of junk, which is reasonable given the fact that the Intel CPU alone costs nearly $300, their DNA might allow them to ship a $799 ARM based MacBook Air while enjoying the same 40% profit margin, solely for the fact that they dont have to pay off Intel anymore. Not to mention that this most probably results in increased sales. I consider this plausible for the simple fact that Apple calls the MacBook Air the notebook of the future", they need a cheap and powerful CPU which is low in power consumption - and that's exactly ARMs strength.
If we just look at all the hardware associated topics: Its plausible - if not even reasonable.
But Mac OS X doesnt run on ARM
Wrong. Mac OS X and iOS are the same thing, iOS just being a derivate of Mac OS X. They both distributions of Darwin, which is the underlying operation system - just like Ubuntu is a distribution of GNU/Linux, and kubuntu is one of its derivates. This comparison is really stating the point: Only the things that you actually see have been changed. Mac OS X incorporates the traditional ideas of windows and point'n'click, while iOS is based on the touch idea. However, this doesnt make them a different OS, the OS is still Darwin. Both Mac OS X 10.6 and iOS 4 are derivates of Darwin 9.0, both are the first versions that support Grand Central Dispatch for example. Even higher level API changes in AppKit and UIKit corespondent with each other(compare
iOS API to
Mac API), making the abstraction from Mac OS X only at the needed parts.
So, given the fact that there is a Mac OS X derivate that runs on ARM, porting Mac OS X is less then marginal. You can expect that there is already an iPad 2 somewhere deep in Apples labs running Mac OS X with the help of the VGA adapter, a Magic Mouse and a bluetooth keyboard. There is nothing much to state here, its not even solely possible, but something that works by default, because Mac OS X is build to be platform in-dependent already, otherwise it wouldnt run on PPC and Intel. ARM support consists literally of changing the target platform in Xcode to ARM and recompiling. Hard to believe, but true.
BootCamp doesnt work anymore
Not with your x86_64 version of Windows 7 - but with the
ARM version of Windows 8. If you need it for business applications, youll be as fine as you can be with Microsoft supporting two platforms simultaneously - you cant blame that on Apple, though.
ARM CPUs are intended (at first) for notebook usage - MacBook Air and MacBook. If you run games, you dont want to do it on such a machine in the first place. It will take longer to see an ARM-based Mac Pro, and even if you think of ARM as the architecture that powers phones - if Apple takes ARM serious, it will not even be more powerful than the mobile chips talked about in the first two paragraphs, but also exceed the performance of a Xeon. Although this is a hard road, Apple likely will stick with x86 in the iMac and MacPro lines longer than in the MacBooks and the Mac Mini. There is nothing wrong with supporting two architectures simultaniously and slowly phasing the high-end lines to ARM when possible, or not at all. We had that with the introduction of the Intel Macs, which coexisted peacefully with the PPC Macs - where a PowerMac G5 was faster than a Intel MacBook, too. Apple could have kept it that way, and it wouldve been fine - But the PPCs that IBM was able to deliver werent fast enough for any Mac anymore, from the MacBook to the MacPro.
Third-party applications dont work anymore
For this matter lets assume the worst case scenario - no x86 emulation whatsoever. However complicated one application might be, Apple has done a great job solving all those problems with the Intel transistion already. Apple usually introduces such a big step in advance, so developers have time to recompile their applications before they lauch the consumer hardware. Usually, application only have to be recompiled to run on ARM, and this happens automatically when a new version of the application is released. Unlike Microsofts approach with ARM support on Windows 8, Mac OS X comes with
Universal binaries, and runs one version of an application on PPC, x86, x86_64 and ARM. You dont even notice the underlaying architecture anymore, Mac OS X became the architecture in itself.
If you use applications which are not under current development, you might consider to find a replacement anyways.
Conclusion
After all that has been said, it is clear that transiting part of the Mac family to ARM is not only possible, but also reasonable and the best thing that Apple can do. Given the facts that ARM is nearly as good - and potentially even better - , but significantly cheaper than Intels chips in the near future, it wouldnt make sense to go another route in the sense of profitibility.
The fact that people associate ARM with slower mobile phone and tablet devices is a minor problem, as they will be proved wrong by raw facts anyways. If youre given the choice between a 13 Quad-Core ARM MacBook at $799 and a 13 Dual-Core i3 MacBook Pro at $1199, the question isnt anymore between ARM vs. x86, the question is whether you can afford it. The lower the scale you go down, the more money becomes an argument. The MacPro is anywhere between $2.499 and $12.000. Add $500, everybody rants, but nobody cares. But removing $200 or 20% from a MacBook while increasing the performance compared to previous generations is a chance for Apple to increse the marketshare in the areas where money matters. This alone justifies a switch to ARM.
