Christopher Mims, Wall Street Journal:
Many manufacturers are already using mobile chips from smartphones in laptops running Google’s Chrome OS, and are starting to put them in laptops running Microsoft Windows. Apple Inc. already designs its own chips, which are
arguably the fastest mobile processors in the world—will it use them in its own MacBooks?
And:
Imagine something that looks like a MacBook and works like a MacBook, but has the guts of an iPhone. In addition to things like facial recognition and AR capabilities, it could have longer battery life, built-in always-on connectivity to fast 5G networks, and more.
And:
Last September, Apple declared that its A10X processor, which powers the iPad Pro tablet, was already faster than 80% of the Windows notebooks sold in the past year. The iPhone X’s A11 Bionic is even faster.
And:
“You see Intel delaying new technologies anywhere from six to eight months, and that hurts Apple’s roadmap,” says Ben Bajarin, an analyst at market- research firm Creative Strategies. “Apple in particular doesn’t want to have to be hamstrung.” By using its own silicon, Apple could potentially offer machines that do things other notebook manufacturers might not match for some time, he says.
This is all speculation, not news. Will Apple build a Mac of some stripe with an ARM processor as the main CPU (as opposed to the Touch ID ARM chip in some MacBook Pros, which are task specific)? That does seem to be the way the wind is blowing.
The benefits are clear. More of the stack for Apple to control (though manufactured by TSMC, Apple controls the design of chips like the A11). An ARM chip would bring longer battery life, and could bring mobile capabilities like Face ID and on-chip AI for blazing fast machine learning and augmented reality processing.
Could this yield even thinner laptops? Before they do that, I’d hope that Apple considers making the keyboards and battery easier to replace. I’d gladly give up thinness for a speedier turnaround to fix a problem like that.
Imagine a keynote slide where Phil Schiller explains how much easier a keyboard or battery swap-out will be. That’d get my vote.
This idea has been floating around for years now. Makes sense for Apple to do this if they can maintain parity with current Intel chip performance and I can see a convergence of sorts, down the road, for the low-end MacBook.
Modular parts replacement as a future feature? After Apple has spent years moving away from that paradigm? Speculation, indeed, Dave.
But do we always need to have all the computing power in a single device? Is it possible to eventually distribute prioritized tasks using some sort of low-latency “load balancer”?
I could see a coprocessor in the Pro/iMac lines, though.
There’s no space and battery premium there the way there is with the laptops, and “can run all the mac software your MacBook can” as well as “can run it all natively for development testing, rather than having to emulate” is kinda compelling.
XGrid has existed since 10.4.
But, er, no, distributed computing really isn’t the answer for Normal People Stuff on Normal People Hardware.
Since Walt left the Fox Business Journal and Propaganda sheet, er, WSJ, their credibility on Apple news has been in the toilet.
Possibly but why? I mean, the reason people buy them is raw power so I don’t see what you get by adding an ARM coprocessor (unless we get to the point where software is being released that only runs on ARM). And I don’t see the primary processors in those boxes being ARM for a very very long time unless Apple is way further along than I think with ARM being able to match even the 8-10 core Intel chips in the iMac Pro.
What Sigi said. There’s no real advantage to doing what you suggest, t here’s a ton of technical hurdles and it adds complexity that doesn’t really benefit most people. If I have ambient compute resources scattered through my house (which I do – the MBP, iPad, iPhone…) and I can tie the together… what happens when I take my ARM MacBook out of the house? It’s far easier to buy a machine that will do what I need it to do reliably on internal compute resources.
I thought of this distributed computing as the way Siri works (off-device), minus speech recognition and NLP. But I definitely agree with you and Sigi, that this is more complicated, and most likely not worth the hassle, especially for “everyday tasks”.
A lot of Mac users use virtual machines on their Macs running Windows, Linux and even macOS. Moreover, the ability to run virtualized PCs removes most of the fear of switching to a Mac. An Intel-less Mac would be a no-no for all those users.
Maybe Apple launches ARM-based Macs in the future, but I think that, in that case, they will coexist side by side with Intel-based ones. Perhaps ARM MacBooks and Intel MacBook Pros…
What percentage of consumers even know what a virtual machine is? How often do Apple store employees recommend that as a solution for switchers?
There will always be a place for heavy lifting types of Macs – development, graphics, research, etc. But most people want a laptop that does mostly everyday tasks like email, web, social, etc. I expect you’ll see dual chip laptops for quite a while where most of the work is done by ARM chips and less space devoted to batteries. The X86 would only kick in when you need it – over time this would diminish as devs target ARM more and more. Marzipan will facilitate that transition. Also I could see setups where you plug in a GPU when you want to edit your photos or music. Maybe even as part of a charging unit.
I don’t think it’s an “either/or” but actually an “and”. Most people don’t care about the silicon but want a great form factor and the freedom to carry a laptop anywhere.
Interesting idea but Apple won’t do this. They rarely do the fence-straddling thing which is what a dual chip solution would be (and it would add a lot of cost and complexity).
Instead, if they don’t feel they can move everything to ARM at once I think they will simply move the MacBook to ARM, discontinue the Air and leave the Pro on Intel. I would imagine, though, that they would prefer to do what they did with the Intel transition and simply move all of the MacBooks over.
Where I don’t see them moving is a Mac Pro/iMac Pro to ARM, ever (or at least not for a VERY long time).
EDIT: Keep in mind, too, that when people say “the Axx is as fast as…” They’re going off benchmarks which are short term things that load up the CPU for a few minutes. One of the more critical issues will be sustained performance over a lot of tasks. I wonder if we know much about how the A series will do in those conditions.
I think the group of people who use VMWare, Parallels, Bootcamp, Docker, etc is large enough and influential enough that Apple needs to carefully think about how to support them for the foreseeable future.
A virtual machine by its nature can replicate the Intel architecture (at some performance cost). The more acute issue is Bootcamp which would become largely useless (even though Microsoft has Windows for ARM).
I’d be interested to see actual numbers on that group as a gauge of how much this directs Apple’s attention.
As you say, it doesn’t make sense for Apple to do a dual chip strategy and transition. Historically they didn’t do it with either PowerPC or X86. However it’s just a hunch that I have.
The complexity and composition of computing platforms is changing. Yesterday I was reading about Toyota running Kubernetes in the “head unit” of every Camry which would control entertainment, Bluetooth, telematics, etc. The whole thing would constitute over 100 microservices which could be updated independently and use different technologies.
The whole thing got me thinking about alternatives when I was reading this article. I think Apple has more flexibility in solving these problems than people think. 😉
john.doofus’ reply is very much to the point: for example, VMs or Bootcamp are frequently used by university students and researchers who need to run certain software packages. Do you know those images of classrooms full of people using Macs? Say goodbye to them if new Mac models are based only on an ARM CPU. And that kind of user is one of the most influential: friends, colleagues, family members…
Performance of VMs simulating a x86 CPU on an ARM one would be abysmal, even if Apple could add some virtualization support to future Ax chips.
There are enough users to fund the continued development of VMWare Fusion, Parallels, VirtualBox, Docker for Mac, and Bootcamp. These are non-trivial products, sold at low per-unit costs (or given away free). Seems likely they have a significant number of users.
Apple has announced its forthcoming Mac Pro will be modular. Making a MacBook Pro modular too isn’t much of a stretch, especially as we are at “peak thin”.
Remember Zoolander and its tiny feature phones, well before today’s supersized phablets?
Everything is a cycle. Where superthin was once cool, making thicker devices with modicums of modularity may well come back.
But as you’d no doubt say, speculation, indeed, Mo.
Except for the fact a modular Mac Pro is on the way, according to Apple itself.
Simple, buy a Mac with an Intel processor. If you don’t need no stinkin’ virtual machine, you won’t give a damn about Intel.
Simple! An A-xx Mac does NOT mean Intel-powered Macs will suddenly disappear.
What is wrong with you people jumping to end-games that aren’t going to be played for years and years yet? 🙂 🙂 🙂
There are VMs that run x86 Linux or Windows applications on Rasberry Pis (which are about 1/10th the performance of an iPhone X in benchmarks). Certainly not as good as running natively, but it could definitely be viable for many needs.