Is Intel doomed on the Mac?

Is Intel doomed on the Mac?

With apologies to Mark Twain, the reports of Intel's death on the Mac are greatly exaggerated

Former Apple executive Jean-Louis Gassée recently offered his opinion on the future of the Mac. In his blog post Macintel: The End Is Nigh, Gassée predicts the imminent demise of Intel-based Macs. Is he right?

Intel's dominance of the Macintosh platform, in place since Apple migrated away from PowerPC in 2006, will end with a future generation of Apple-made silicon, Gassée surmises. Is it possible? Yeah. Is it going to happen tomorrow? No.

The chips Apple uses in iPhones and iPads today certainly aren't suitable to replace the silicon inside the Macintosh. iOS devices are radically different in their design and execution from the Mac, and there's a whole host of technology inside a Mac that don't concern an iPhone or iPad. Instruction sets are very different between the two chip architectures, too, and that raises a whole host of other issues.

Having said that, if there's anyone that can manage a major architecture transition, it's Apple. They've done this repeatedly. They moved the Macintosh from Motorola's 68K architecture to PowerPC, then later from PowerPC to Intel. And while there were some growing pains along the way, each move has, on balance, yielded positive results for Apple, for developers and for customers.

Why does Apple switching the silicon inside the Mac keep coming up, from Gassée and others? A lot of the speculation is driven by a drought that Apple has seen this year from Intel, which has had trouble getting a new 14 nanometer manufacturing process to produce acceptable chip yields — something it needs to do to get its Broadwell microprocessors into production.

Broadwell's behind schedule, but it is coming. And it's coming fairly soon.

"I can guarantee for holiday, and not at the last second of holiday," Intel CEO Brian Kraznich said to Reuters in May.

So later this year we'll begin to see new computers featuring Broadwell microprocessors, and I have very little doubt that Apple will be one of the first, if not the very first, PC manufacturer to offer a new computer with one of the chips inside.

Broadwell promises to be more power efficient and to have some distinct performance improvements in areas like integrated graphics — something of key importance to Apple, of course, since graphics hardware, whether integrated or discrete, is used so thoroughly throughout the entire operating system.

In that same interview Kraznich also said that Intel wasn't going to get Broadwell into production in time to meet the back-to-school schedule, so it's no surprise Apple's refreshed its most popular student machines (and realigned prices on some of them) already, rather than waiting.

One of the reasons Apple's transition from PowerPC to Intel went as smoothly as it did in 2006 is because the operating system itself had been running on Intel processors for years. NeXTStep, the OS upon which OS X was based, ran fine on Intel processors, and Apple kept the technology working on Intel hardware even as it sold PowerPC-based Macs, and published PowerPC-based development tools, operating system and application software.

And that transition has yielded big benefits for Apple. One of them, for example, is the Macintosh's innate ability to run Windows at native speeds using Boot Camp. Running Windows on the Mac wasn't unique to Intel-based systems: You could run emulators like Virtual PC in the PowerPC days. They just didn't scale to native hardware speeds at all. Even now, virtual machine software like Parallels Desktop and VMWare Fusion run leagues better than Virtual PC ever could have dreamed of, thanks to the Mac's Intel underpinnings.

Windows compatibility — and the Mac's ability to operate as a host to many other x86-friendly operating systems — is a selling point both with consumers and with enterprise that shouldn't be underestimated, and something that likely would be sacrificed if Apple transitioned away from Intel.

I have very little doubt that somewhere in the bowels of Apple's Cupertino skunworks, sit Mac devices that already use some generation of A-series ARM processors. Apple would be crazy not to leverage its skills at owning the entire process not to have such a machine in operation. But there's a huge difference between having a prototype showing a proof of concept and a machine that's ready to unleash on the world.

In all likelihood, at some point in the future, Apple silicon will be able to make the transition to desktop use. But I don't think Intel's delays with Broadwell have been enough to put Apple off from using Intel hardware in the Mac.

Contrary to what some impatient Mac fans might want us to believe, Intel does have its act together, and they remain an incredibly important manufacturing partner for Apple. We're yet to see the best of what's to come from Intel and Apple, that's for sure.

Peter Cohen

Managing Editor of iMore, Mac and gaming specialist and all-around technologist. Follow him on Twitter @flargh

More Posts

 

6
loading...
0
loading...
40
loading...
0
loading...

← Previously

Best iPad apps to store and manage PDFs: Organize, edit, annotate, send, and search better!

Next up →

How to troubleshoot and fix problems with iMessage

Reader comments

Is Intel doomed on the Mac?

36 Comments

I think the os9 to osx transition was the toughest and definitely not the smoothest transition

Posted via iMore App

Not me, though I do use it. Until Apple starts putting decent gaming graphics cards in their systems, I have to have a desktop to run windows for gaming. Still prefer Apple's laptops though.

I can see at the lower end (Macbook air) the ARM processors being good enough to be used in laptops and desktops, but I just don't see the point. The current iOS and OS X boundary sits between tablet and laptop and that's a good place for it. It would take a huge amount of work to get lower end laptops working with ARM processors and all you end up doing is moving that boundary one level up from where to sat before. You're still going to need Intel processors for the mid and high level PCs.

"Why does Apple switching the silicon inside the Mac keep coming up, from Gassée and others?"

Because of that extreme Intel Tax (tm) that Apple is forced to pay for every x86 chip they buy. Anywhere from what, $200 to $350 per Mac. Compare that against $19 or so for the 64-bit A7. Then do the math.

Sure, there would be a transition period for legacy 3rd party apps. And Apple might need to use the old "fat binary" technique for universal x86 / ARM apps for a few years. They did that quite successfully during the transition from 68K to PowerPC and again in the transition from PowerPC to x86. And they could easily do it again in the transition from x86 to ARM.

But how many users would actually need to wait for 3rd party apps? The average consumer might not need to wait for anything. They'd buy a new MacBook Air with ARM SoC, with OS X pre-installed, and with Apple's suite of free apps. All compiled for ARM, all updatable through the Mac App Store (along with any paid Apple apps they have), all able to use or migrate data from their legacy x86 versions.

Yes there's a pro market, and no they won't jump to buy ARM-based consumer-grade Macs. For them, Apple could keep on shipping x86-based Mac Pros and MacBook Pros, with whatever pro apps Apple still sells. Eventually, the pros could migrate to ARM. They might need to for performance. All that decades-old x86 cruft, all those obscure instructions that Wintel compilers depend on, and all that CISC complexity all slow Intel down in terms of performance improvements and die shrinking.

In a few years, it's possible that ARM-based AX chip performance could surpass Intel's. I've heard arguments that ARM simply isn't able to provide desktop class performance. But that's because ARM chips are currently all optimized for mobile. Just imagine if the battery usage, chip size, and cost constraints are eliminated for desktop use.

Perhaps the reason why Intel chips cost so much more than Arm chips is simply because they are an order of magnitude quicker. You're not comparing like for like. One is intended to power simple hand held computers which do very little, while the other powers Photoshop, Maya, Zbrush, Renderman, Final Cut Pro, Lightroom and a hundred other pro apps.

Good luck running Zbrush on a mobile CPU.

If by "quicker" you are referring to the Megahertz etc. then you don't know much about chip design and how performance works. Future chips for all computers made by all chip makers will tend to be the same speed or slower than the ones we have today. The idea is to get better performance out of the energy being used, not to up the ante for the amount of energy needed.

But how many users would actually need to wait for 3rd party apps? The average consumer might not need to wait for anything.

I think you're underestimating the average consumer's expectation. When people buy Macs they expect to be able to use software on them. Many of them don't download a ton of stuff from the Mac App Store but they know it's there. What's more, many others — and many institutions buying Macs — expect software like Microsoft Office to work on Macs.

Invalidating the Mac App Store and Apple's existing base of registered developers seems like a smart way to kill the Mac all together, not push it forward for the sake of an architecture change.

To add to that, the earlier transitions SockRolid mentions were successful because Apple eliminated that "average consumer" nightmare with compatibility shims -- Classic environment for OS9->OSX, and Rosetta for PowerPC->x86. These techniques worked because they were moving from a slower system to a faster one. That option is simply not available for x86s-ARM, because the reverse is true here; although the A7 is more power efficient, it is ultimately less powerful than x86. Any runtime compatibility/translation layer is going to be unusuably slow, unless Apple first comes out with silicon several times more powerful than the current Ax line. Without that, those "average consumers" are going to get confused and frustrated.

Does Apple have the engineering talent to pull it off? Perhaps -- but it seems questionable that it would be worth the effort, or diverting the focus of their top talent.

I was shopping for a notebook in 2004 to take to college. I had been planning on getting a PowerBook until the rumors of the switch to Intel started sounding more certain. I didn't want to be stuck on an outdated architecture so I went with a PC notebook instead. A decade and a few PCs later I'm again considering getting an Apple notebook. Fitting that rumors of an architecture switch would be swirling again...

...but did you read the article? Stating that any such switch is unlikely at this point?

Sent from the iMore App

Oh good! Some blogger said it probably wouldn't happen real soon because there would be some disadvantages for customers. I guess I can hang my hat on that!

I read the article referenced and there are good arguments either way. I don't expect Apple to jump ship anytime soon but I gotta think it's going to happen. I don't think the persistence to switch Mac chips is as bad as iOS chips because of the relationship with the players involved. Apple won't do it until everything is right.

Apple moved away from Power PC because Motorola weren't pushing the tech and were more interested in lucrative console contracts. Intel are obviously not in that position. Plus the speed difference between say, an i7 and an A7 is staggering. Apple would gain nothing by ditching Intel but would lose a lot. It won't happen for a very long time, if ever.

Great article, but I think you left out consideration of the possibility that the change might not be for ALL of Apple's computers at once.

For instance, it's unlikely that the ARM chips will get to the level of performance that the Mac Pro needs for many years to come. This means that if they change at all, they have to keep intel OS X around for a while just for the Mac Pros.

While this could be considered as merely fuel for your suggestion that it won't be happening anytime soon, it could also be taken as indications of the need for Apple to be running multiple versions of OS X on different chip hardware.

The Broadwell chips that Apple needs in MBPs are not coming out this year, at the very least they'll be available by early spring next year.

The only thing that might be coming out later this year is the Core M, which isn't the chips that Apple uses in MBP. Apple might be able to put it in MBA but unlikely.

Umm, truly obvious? No, it isn't. Why would Apple do that when they have zero experience dealing with x86/x86-64? You do know that Apple was one of the founders of ARM and that Apple bought 3-4 ARM-based companies in the past five years?

It is truly obvious to build and customize their own ARM chips than it is x86 that is not going to happen. Apple cares more about power efficiency than they do about speed, you want that, you go ARM.

Also, Apple does not own fabs, they will still have to rely on Intel to build and share fabs with Apple to build those x86-64 chips. Why would Apple do this when Intel's pretty much the best on the planet and ahead of everybody by at least 3-4 years, of which they'll remain so for the next decade.

It is also very unlikely that Intel and AMD would sign a massive cross-agreement with Apple to let them build x86 chips. You cannot build an x86-based chip without a license. While that is true for ARM as well, Apple's an initial founder and have arch license for it.

Because any of this is sooooo going to happen.

NOT!

I mean it could... Just like we could all wear cowboy hats and spacesuits as normal fashion by 2015. Anything could happen.

I don't believe that anything discussed regarding Apple switching away from Intel is going to happen.

One reason they are pushing Swift language so hard is that it would simplify a switch over to their own architecture. Simply by recompiling developers wouldn't need to worry about the chipset change. Apple wouldn't need to require a license to build a chipset based on the Arm architecture that they already license.

As MikhailT below (or above) has stated, ARM was a business that was started with Acorn Computers in the UK and Apple joined them into designing SoC architecture. The whole idea being that by building such systems they could create chips that did functions.

Acorn started using SoC when they introduced the follow up to the BBC Micro, the Acorn Archimedes which was the first computer designed around a RISC system on a Chip design. Apple bought into the business and while Acorn ultimately vanished ARM was born and the rest is history.

I also doubt that Apple would waste time and effort building their own version of chips using the instruction set from the x86 chip set from Intel (no need to license from AMD, they license from Intel) it would be far easier to make the switch from Intel Chips straight to ARM based chips, it may cause a ruckus for those that have Mac devices until many switch to the newer devices and eventually just as previously the older Intel based systems would be left behind.

I think we are though a long way off from that change over, if at all, the more times Intel suffers issues with a newer version of chips forcing Apple to do interim updates with chips that are just interim updates to the same old chip set the more likely Apple will be swayed. Then we have to think about cost. If Apple can produce their own chips cheaper and have them perform better it may be more desirable but again I think this is a long way off.

One reason they are pushing Swift language so hard is that it would simplify a switch over to their own architecture

I don't follow your logic here. Objective C and Xcode already produces code that will work on either ARM based iOS devices or Intel-based Macs. How does Swift change things in that regard, specifically?

Swift is just "Objective-C: Cleaned Up."

The middle part of your post: Yes, yes. ARM designed for the Newton, licenses for CPU arch isn't some new thing.

Last paragraph: Exactly. If it ever occurs, it will be quite a while.

Intel is the only chip manufacturer that can consistently provide 14nm chips, and drive the technology further - they are by far the most technologically advanced chip manufacturer. If they are running into issues at 14nm, it's any wonder how foundries like TSMC or GlobalFoundries can provide sustainable quantities of 14nm silicon. And if Apple ditches Intel and decides to go for home grown processors, they need to be manufactured by one of those foundries, like Samsung. That would immediately limit them to those foundries'' capabilities

Sounds like a stupid prediction to me. Apple would be mad to move away from Intel as they are the largest mainstream chip manufacturer in the PC business. So Intel who has years of experience is behind in implementing a new smaller manufacturing process doesn't mean that Apple who has no experience in this field can just step up and do it better. Never gonna happen. Apple is good at designing what they do but they don't know how to fabricate certain components or have the capacity to do so. I can't see them manufacturing their own architecture for the mac, at must maybe they could design a custom featured x86/x64 architecture but they would likely still have to rely on Intel for the manufacturing.

i would be upset if they switched to arm at this point. i wouldn't care if it was faster or more power efficient. it would break a lot of the software i use. it would break nearly every game. it would stop many game ports from happening. it would make my current intel mac the last mac i would buy. and i would probably get a windows surface tablet.

Why don't they try out some of AMDs processors, failing that make custom silicon that runs on the same architecture, it's not like OS X can't run on AMD Processors, there's many Hackintoshes out there that've managed it. I've always failed to see Apple's exclusivity with Intel with processors, especially when it comes to HDD/SSD you end up with a random manufacture one.

Regardless, I'd hardly call it a crisis at the moment with Mac processors and performance, a couple of months is inconvenient, but not end of the world.

"Windows compatibility — and the Mac's ability to operate as a host to many other x86-friendly operating systems — is a selling point both with consumers and with enterprise that shouldn't be underestimated, and something that likely would be sacrificed if Apple transitioned away from Intel."

This was one of the "key" reasons I transitioned to the Mac. I had the best of both worlds on ONE computer. I hope they don't move away from X86/X64.

Another CPU transition ? Software developers love that, the upgrade cycle it forces, makes them allot of money , the amount of software and plugins we lost access to due to 68k ~ ppc ~ intel ~ 64 bit only transition , is a long list, when are those instruction set reconfigurable CPUs coming I heard talk off 15 years ago.

Sent from the iMore App

Also how about apples based off amd's jaguar platform as used in Xbox one and ps4 ? Cheaper x86 silicon . . . than an Intel CPU with sufficient performance?

Sent from the iMore App