Early benchmarks show that the third-generation, 64-bit ARM-based Apple A9 SoC powering the new iPhone 6s offers performance in line
with the x86-based Intel Core M processor used in the 12-inch Retina MacBook. This is not surprising, as the second-gen 64-bit Apple A8X that was the heart of the iPad 2 was not far behind in our GeekBench 3 cross-platform benchmark testing as part of our 12-inch Retina MacBook review
from earlier this year. What this means is that the iPhone 6s is capable of running Apple's full OS X, which powers its Macs -- but just because it can, should Apple release mobile devices powered by OS X?
A complete OS X on a device you can stick in your pocket is certainly an interesting proposition -- but before you get too excited about the prospect, keep in mind that Microsoft has already tried to do something similar. Windows RT was Microsoft's attempt to port its x86-based Windows 8 platform across to ARM-based tablets, and the original Surface RT
was powered by Nvidia's ARM-based Tegra 3 SoC. As you may be aware, Windows RT was unofficially ended by Microsoft when its third-gen (non-Pro) Surface 3
was released this year, running Intel's latest x86-based Intel Atom chip and full Windows 8.1, which is now upgradeable to Windows 10. Windows RT on ARM failed, in part, due to Microsoft's inability to clearly articulate its reason for existing, and RT's inability to run regular Windows 8 applications.
It has always perplexed me as to why Microsoft went down the Windows RT path. Certainly, at the time the Surface RT was launched, the ARM-based iPad had popularized thin and light tablets. Intel's more powerful x86 chips, necessary to support full Windows 8, needed more thermal space and active fan cooling, which is why the original Surface Pro was as thick and heavy as it was by comparison with the Surface RT, and of course, the iPad.
However, Microsoft already had Windows Phone 7 (and later Windows Phone 8) which could have been readily adapted to work on tablet form factors in the same way that Apple tweaked iOS (originally iPhone OS) to work on the iPad, in order take full advantage of the additional screen real estate. I think it would have made for a very decent experience. However, it would have been a strategy that could have potentially undermined its Windows OS cash cow.
If Apple released a mobile device with OS X adapted for ARM, it would potentially face similar issues to those experienced by Microsoft with the Surface RT: a lack of apps. Mac/OS X developers would be required to port their x86-based OS X applications across to the platform it is was to be truly successful.
However, Apple would benefit from all the ARM-based iOS apps already out there that probably wouldn't need as much recompiling (if any) to work in an OS X on ARM environment. In this instance, there already numerous desktop-class iOS apps that would help to make such a transition more feasible, including, of course, Microsoft's own Office for iOS suite that Apple was keen to show off running on the new iPad Pro.
For those Apple fans who yearn for OS X on an iPad-like device, the possibility of OS X on a mobile device like an iPhone or iPad is the stuff of dreams. However, Apple has made it clear at this point in time that iOS is intended for a touch-first experience, and OS X is for use on desktop Macs and MacBooks with trackpad and/or mouse input. The main question to ask is why would Apple want to move in that direction?
Apple prides itself on the totality of its user experience; its mobile devices already play very nicely with its desktop devices, in large part because Apple has total control over the software running on those devices. However, it has a special edge when it comes to its iOS devices that no other manufacturer mobile maker enjoys -- it has complete control over both the silicon and the software on the iPhone and iPad.
This has turned out to be a tremendous advantage for Apple, for whom we can thank current Google employee and Nest co-founder Tony Fadell. It has been reported that Steve Jobs originally wanted iOS devices to run on Intel's x-86 based silicon. Tony Fadell insisted that ARM-based chips were the way to go, and apparently threatened to quit Apple if Jobs persisted with his thinking on Intel.
In the end, it was agreed that they would give Intel six months to see if they could come up with a workable solution for the original iPhone. As we now know, it took Intel many more years before they were finally able to shoehorn an x-86-based Intel Atom into a smartphone. Under Jobs, Apple not only went with ARM-based chips (a company it had originally helped to found), but also went on to acquire the capability to develop custom ARM-based in-house designs, thanks to the acquisitions of Intrinsity and PA Semi.
The real fruits of Apple's custom ARM-based chip designs were born when it was the first to market with an ARM-based 64-bit chip in the form of the Apple A7, which launched with the iPhone 5s in September 2013. This caught both the chip industry and the smartphone industry completely off-guard. According to the official ARM product roadmap, 64-bit designs were not scheduled to reach the market until 2014. In fact, Qualcomm, one of the leading suppliers of mobile smartphone chips only announced its first 64-bit design, the mid-range Snapdragon 610 and 615, in February 2014.
Devices with this chip did not ship until late 2014. It was not until the launch of the Samsung Galaxy S6 earlier this year that Samsung shipped a high-end smartphone with 64-bit architecture. Qualcomm's high-end Snapdragon 810 64-bit chip that first shipped in devices earlier this year, some 18 months after Apple, have since suffered overheating issues as it struggles to keep pace with Apple's more power-efficient and advanced designs. Apple, in the meantime, is already on to its third-generation 64-bit chip with the new A9 and A9X SoCs.
With custom chip architecture, Apple has been able to power ahead of the competition with thin and compact iPhones that have had substantial performance-per-watt advantages over the competition. To keep up with the performance Apple was achieving from lower-clocked, dual-core chips that deliver both excellent performance and battery efficiency, the competition was forced to use 32-bit quad-core or octa-core chips, clocked much higher and with larger batteries in tow.
When you add into the equation the additional performance optimizations that Apple is able to achieve by compiling its own firmware and software for the iPhone and the iPad, Apple is able to offer devices that are truly cutting edge and exceptionally powerful. The A9 and A9X are so powerful, as we previously pointed out, that Apple could be shipping iPhones and iPads with OS X if it wanted. The advantage in doing so would be that it can further differentiate the performance and portability of its devices over the Intel Core M-powered competition, just as it has with the iPhone and iPad through controlling both the chip design and software experience.
If you want to get a sense of how an iPhone running a ported version of OS X for ARM might perform, it is worth taking another look at the direction Microsoft has taken since its failed Windows RT on ARM experiment. Microsoft will have spent tens of millions of dollars in developing Windows RT, but if you think they have ditched it altogether, you will probably find that you are mistaken. Microsoft is expected to announced at least two new high-end Windows 10 smartphones at a special event on October 6, probably running high-end ARM-based Qualcomm 64-bit Snapdragon SoCs.
One of the key features of the new smartphones is Windows Continuum. Windows 10 smartphones will be able to connect to a monitor, and pair with a Bluetooth keyboard and mouse, and automatically scale to function just like a desktop computer. Its interface will allow for navigation by a mouse, and apps like Microsoft's Office suite will look and function just like their x86-based full Windows 10 PC counterparts. You can bet that Windows RT technology is powering this capability.
Should Apple pursue a similar strategy, if not porting full OS X across to its ARM-based A-series chips, port across the frameworks necessary to give iPhones and iPads desktop-like scalability similar to Windows Continuum on Windows 10 smartphones? The main problem that Windows 10 Continuum faces, like Windows RT before it, is that even though it will offer desktop-like functionality with mobile apps that will scale to a desktop monitor, it will not support x86-based apps designed to run on full Windows 10 machines -- but it does give us an insight into just how powerful mobile chips have become.
A desktop experience from a smartphone is realistic possibility. After all, the x86-based Intel Core M-powered 12-inch Retina MacBook, which runs OS X, performs very well for everyday tasks running full desktop applications like Microsoft's Office suite. An iPhone running OS X on an A9 would perform just as well, given its performance parity, while an iPad Pro with A9X running OS X would actually outperform the Intel Core M-powered 12-inch MacBook.
Apple has long been rumored to have MacBooks in its labs running OS X on ARM. If it was going to go down that path, the 12-inch Retina MacBook would have been the ideal time to do that. Intel should, however, be feeling somewhat nervous about the direction that Apple might take with its ARM-based chip technology.
Intel will sell a lot of additional Core M devices thanks to Apple's decision to build the 12-inch Retina MacBook around its x86 technology, but the reality is that Apple's A-series chips now have the performance capabilities to supplant this particular Apple-Intel combination. The iPad Pro is another device where Apple could have opted to launch OS X on ARM, but clearly, that didn't happen. So for now, Intel probably doesn't have anything to worry about.
Unlike Microsoft, Apple is intent on keeping its mobile and desktop experiences distinct. The iOS platform will run strictly on devices with touch-first interfaces, and OS X will run strictly on devices that have a non-touchscreen, mouse/trackpad interface. Further, given that the iPad Pro would also have been an ideal candidate for an ARM-A9/OS X combination, that missed opportunity all but confirms that Apple is not interested in bringing OS X to touchscreen devices, even though it could. Time will tell which company has the right approach -- with Windows phones failing to make any impact in the market to date, it will be interesting to see if features like Continuum help Microsoft to gain market traction.
What I do expect to see happen, thanks to the introduction of the iPad Pro, however, is a new generation of mobile apps that offer even more powerful, desktop-class capabilities. In particular, these will probably be targeted at the iPad Pro, which offers the additional screen real estate required to take full advantage these types of capabilities. Apple is no doubt well aware of this, which is why the iPad Pro also features 4GB of RAM, the most ever seen on an iOS device.
It does make you wonder where it will all end. Will iOS and OS X eventually converge, as the capabilities of Apple's A-series chips continues to rapidly accelerate? Will Apple eventually kill off OS X and the Mac altogether (iOS devices already dominate Apple's sales)? Or would a dual-booting iPhone that runs iOS when used as a smartphone, but powers into an ARM-based OS X desktop UI be the culmination of our tech dreams?
Let us know your thoughts in the comments below.
-- Sanjiv Sathiah