Though we did include this review in yesterday’s list of the new MacBook Air reviews, I wanted to link to John Gruber’s review separately, for his take on Intel.
The elephant in the room at last week’s Apple event was Intel.
Apple introduced two products based on Intel chips — the new MacBook Air and new Mac Mini — but barely mentioned the company’s name. The word “Intel” appeared on a single slide during VP of hardware engineering Laura Legros’s presentation of the new MacBook Air. She also spoke the word once, saying the new Airs have “the latest Intel integrated graphics”. In the presentation of the new Mac Mini, “Intel” never appeared in a slide and wasn’t mentioned.
And:
Apple is not going to throw Intel under the bus — they’re taking an “If you don’t have anything nice to say, don’t say anything at all” approach, as they should.
Push is slowly coming to shove here. As John points out, the new iPad Pro simply crushes the Intel-powered MacBook Air in GeekBench score, especially that MultiCore score (18,051 to 7,847).
As products, the Mac and the iPad are still on different tracks, not plug replaceable. But it does seem highly likely (a “when”, not an “if”) that Apple will eventually ship a Mac powered by their own bionic chip.
It will happen.
When they pull it off, it should be a significant boost to their bottom line.
I think they should do this just to gain control over the update cycle. Even if the initial releases aren’t substantially faster than whatever is current from Intel and even if the battery life doesn’t benefit a ton in the first gen. With other companies I’d worry about the transition but Apple’s done this, and well, before. In fact, I think I’m going to hold off upgrading my 2016 MBP until 2020 or so just to see if this happens (I know, it’s 4 years at that point… but resale value dives after that point and I like to keep reasonably current)
I’m fairly sure Apple could make a hybrid x86-64 (with some missing instructions)/ARM64, which might also be interesting. I have no idea if they _will_, though.
What would the point be?
Missing instructions means incompatibilities, after all.
And hybridizing at the CPU level sounds, to this programmer, like just a nightmare for operability.
The effort and expensive of making a competitively useful x86 CPU [which is what “make a hybrid on the A12 or whatever” really means, I think?] would be immense, and they’d just be throwing it away in a few years.
Probably better off with a straight coprocessor (like an Atom), or making people suck it up and deal with emulation during the transition, like with Rosetta.
And when Apple decides to ditch the main stream processor family the Mac goes screaming right back to the bad old days of ‘weird incompatible niche product that nobody uses or produces software for’.
I think that time may be dimming your memory of just how incredibly bad things were for the Mac before the switch to Intel…the fact that Macs are more than a fraction of a percent of the computer market is because they switched to a supplier who was in the business of making chips for personal computers; and despite the naysayers, Intel is very very much still in that business.
If they do.
You’re assuming the world is the same as it was then. It’s not. There’s no real need for Macs to run Windows now.* At this point Intel is simply holding them back without any concomitant advantage elsewhere.
*before someone pedantically points out that some people do this… yes. Some. Very few, though, I’d bet.
Those geekbench scores are for a mobile device with a tiny battery and maximum 5 watt TDP (? or even less?). Rest assured that apple has been testing prototypes with active cooling and AC power, and that the scores on their first ARM laptop will be far, far higher.
OTOH, you’re going to need every bit of those higher benchmarks to emulate an x86 for running legacy apps. Count on it taking Adobe and MS two or three years, minimum, to port their apps to the new platform.
Because the instructions they’re missing would be the ones Apple already has workarounds for because Apple has shipped CPUs without them.
Then again, maybe they’d just license that part of the design as well.
The A12X already had four high power and four high efficiency cores. Adding another couple cores for X86 compatibility would be a pretty good product.
Don’t forget Quark!
No, I am not serious.
Funny, judging by the number of Macs I’ve set up Virtualbox on to run various windows-only apps I’d say far more than ‘very few’.
The stumbling block for that plan is IP. Intel owns the X86 instruction set and will not license it. Apple could of course buy AMD, but that strikes me as taking an elephant gun to kill a gnat. Far easier and cheaper to just make a decent emulator and bake it into the OS.
Microsoft has made an ARM version of windows, which runs x86 apps under emulation.
So bootcamp on an ARM based Mac will still work, you just need to use the ARM version of WIndows.
Sigh. This again. You != the market. Anecdotes aren’t data.
Seriously, US education needs to hammer this home. I run into this fallacy (“my experience is generalizable to a broad population”) EVERY DAMN DAY and it’s never true.
I know a fair number of Mac users and none run Windows on their Macs. Guess what… that isn’t an indicator of the market EITHER nor does it invalidate your experience. Both are merely data points.
If you work in or for large corporations, you still often need Windows for one thing or another – custom line of business applications, ancient webapps that only run in Internet Explorer, weird Office add-ons, etc. And I think quite a few people still do this – the market is big enough to support continued development of Fusion, Parallels, and VirtualBox, after all.
That said, there are some interesting alternatives this time around – Windows for ARM, x86 emulation on ARM, Cloud-based VMs, etc.
I owned a Powerbook G4. It was actually just fine, and the universe you describe was not in fact the case.
And this is even more true now, as Rick points out. And, unlike then, Intel can’t execute their way out of a paper bag right now.
I can see this in large corps. However, how many people there have Macs? I guess my thinking is that that would be the predominant use case – primarily Windows corporate installs where Macs are allowed. Outside of that, I don’t see this has more than a very edge thing and that’s my point – the world is not like 1995. To assert that everything will revert to that state is naive and wrong-headed.
That’s why I said it might be missing some instructions. But Apple has handled that problem before.
I’m not very positive about ARM’s ability to emulate x86-64 in software quickly enough to be useful, but I haven’t done enough research to be sure. It seems like just the sort of problem that should be pushed to hardware… especially if Apple is making the hardware.
Another possibility, possibly better, is for Apple to just include a lower-performance Intel CPU on the motherboard somewhere.
You might be surprised… the big corp I work for has a BYOD policy, and healthy internal Mac discussion boards that include topics like figuring out how to run the corporate windows build inside a VM. I don’t think it is unique in this respect.
One of the projects I’m working on involves collaboration with researchers. They use Macs, but have virtualization software installed.
For years, I did consulting work at big corps and govt. Having the ability to run Windows via VMWare Fusion was very important. I didn’t need it every day or even on every project, but when I needed it, it was indispensable. I had several colleagues who did the same thing.
So, I agree that it is a minority of people who run Windows on their Macs, but I think it is too dismissive to call it a “very edge thing”. Also, a lot of the customers in this minority are Apple’s “A” customers – people who buy high-end devices frequently, and influence others’ purchasing decisions.