BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Could ARM-Based MACs Arrive On Thursday?

Following
This article is more than 7 years old.

There have long been rumors that Apple would port the macOS to its ARM-based “A” series processors, replacing Intel CPUs on some Macs. These rumors keep surfacing year after year, without any real evidence. While I’m not going predict that Apple will unveil A10-based Macs this Thursday, there’s evidence that Apple is moving towards that goal. One of the most fundamental challenges for Apple will be the performance of Apple’s in-house A series processors compared with Intel’s Core processors. According to respected newsletter Microprocessor Report, Apple’s own processors are now hitting that goal.

Microprocessor Report recently ran an article describing a few performance benchmarks and codenames for the new Apple A10 Fusion processor (as reported in Forbes).

The most recent A10 processor added an unusual configuration when Apple designed it with two sets of processors- one very complex and powerful; the other is smaller and lower powered. ARM’s own implementation of this technology has been around for some years and is called “big:LITTLE.” Linley Gwennap reported in the October 20, 2016 Linley on Mobile newsletter:

“Apple’s investment in custom CPU design continues to pay off, as the new iPhone 7 delivers better performance than any other flagship smartphone and outscores even some low-end PCs. The phone uses a new processor chip, the A10 Fusion, that contains not one but two custom CPU designs, representing Apple’s first foray into the Big.Little approach that many other mobile processors employ. The massive Hurricane CPU improves performance by 35% over the previous-generation Twister, boosting both the clock speed and the per-clock performance. The smaller Zephyr CPU helps the iPhone 7 extend battery life compared with its predecessor.”

Apple was not only giving other ARM processors run for the money, but in addition Hurricane is also very competitive with individual CPU performance from Intel’s Core mobile processors. And while the ARM processors are all running well over 2 GHz, Intel’s more complex mobile Core M processors, like the one used in the 12 inch MacBook, run at lower frequencies.

For example, Intel’s Core M products run at only 1.3 GHz in order to sneak under the 5 Watts limit for fanless operation. Only under-clocked Intel Core M processors can fit into thin and fanless MacBooks, but it is possible to use Hurricane at clock speed over 2GHz in fan-less designs like the Apple iPhones. In addition, Hurricane could lower Apple’s costs, make for thinner MacBooks, and give longer battery life, while still offering competitive performance for ported applications.

Apple is a company that has long been known to prefer to control its own fate, which is why it has designed its own ARM cores for iPhones and iPads since 2012. But since it migrated from PowerPC to x86 processors from Intel back in 2005, the company has relied exclusively on Intel as a supplier for its Mac products.

Apple’s relationship with Intel has not always been smooth. Intel took what it the learned from Apple’s MacBook Air products and helped to create the UltraBook category for Windows PCs. Apple also pushed Intel to improve its on-chip graphics (which was often pretty basic in the past – just good enough for Microsoft Office). While Apple could have used AMD as an alternative at some point, it has chosen to stick exclusively with Intel, likely due to Intel’s stable supply chain and deep technical support (Intel helped with engineering resources to port the old macOS from PowerPC to x86). Still, the dependence on Intel probably makes Apple uncomfortable.

Since those days, Apple’s primary business and revenue has migrated from Mac products to iOS products, which includes iPad tablets, iPhones, and iPod music players. To put its own processor into the Mac line would require some significant software work. Any macOS software that has x86 source code will not run on ARM processors, and so far software translation has been slow. Apple last made this transition when it went from PowerPC processors to the x86 processors, and at the time it relied on its “Rosetta” software translation technology based on technology licensed from Transitive Corporation, a company that has since been bought by IBM.

Running x86 code on ARM is more complex than running PowerPC on x86 because the x86 instruction set is more complex and more difficult to emulate. That is why there have been few widely distributed x86 software emulators. Many of these software emulator companies were snapped up by companies like IBM and most recently Intel (Soft Machine). So this begs the question: how will Apple support legacy Mac applications on an ARM-based Mac? Porting all the Mac applications to ARM would also take a long time and emulation is slow.

There are possible alternative solutions. One, a hybrid approach where iOS applications run in a virtualized window on top of a macOS base operating system. In this case the applications would mostly come from the Apple Store and would bridge the application between iOS apps and Mac applications. But by putting a macOS underneath iOS you would add the management capabilities that professionals look for. In this case you would have iOS at operating system running as a guest OS on macOS, were you can still manage the platform like a Mac, but with the lockdown applications of iOS. In this case, Mac applications would develop that could run a both x86 and ARM over time.

Another interim solution would be to run any x86 Mac application is a cloud application using a virtual display interface (VDI), which is common offering today from many cloud vendors like Amazon AWS and Microsoft Azure. From my own experience using NVIDIA’s GRID, those legacy applications can run quite well when supported by a virtualized GPU in the data center. This would require a minimum of software support from Apple, but Apple would have to figure out an acceptable subscription strategy during the conversion from x86 to ARM.

While these interim solutions might not be appealing to many professionals, Apple would still maintain x86 Mac product versions over a period of time, most likely using x86 at the very high end of the Mac product line until Apple’s own processors are good enough to take over at the high end.

I believe the first ARM-based Mac product will likely be a MacBook clamshell notebook. This new MacBook would have a keyboard and add a touch screen, and bridge the gap between the iPad Pro and the MacBook Pro.

Like any Apple conjecture, I could be completely wrong. But that said, there is certainly something here beyond typical rumor and technical and business reasons for Apple to make the change. An ARM-based MacBook is also much more likely to happen in the near term than an Apple car or an Apple television.

Kevin Krewell, Principal Analyst, TIRIAS Research

--  TIRIAS Research tracks and consults for companies throughout the electronics ecosystem from semiconductors to systems and sensors to the cloud.