Apple Plans to Use Its Own Chips in Macs From 2020, Replacing Intel
Mark Gurman (tweet, Hacker News, MacRumors, 9to5Mac, iMore):
The initiative, code named Kalamata, is still in the early developmental stages, but comes as part of a larger strategy to make all of Apple’s devices -- including Macs, iPhones, and iPads -- work more similarly and seamlessly together, said the people, who asked not to be identified discussing private information. The project, which executives have approved, will likely result in a multi-step transition.
[…]
The shift would also allow Cupertino, California-based Apple to more quickly bring new features to all of its products and stand out from the competition. Using its own main chips would make Apple the only major PC maker to use its own processors.
[…]
While the transition to Apple chips in hardware is planned to begin as early as 2020, the changes to the software side will begin even before that.
[…]
Apple’s current chip designs made their name in thin and light mobile devices. That would indicate Apple will start the transition with laptops before moving the designs into more demanding desktop models. Apple has to walk the fine line of moving away from Intel chips without sacrificing the speed and capabilities of its Macs.
But when you start thinking about the details, this transition would (will?) be very difficult. First, while Apple’s existing A-series chips are better for energy-efficient mobile device use (iPhone, iPad, just-plain MacBook), Apple’s internal team has never made anything to compete with Intel at the high-performance end (MacBook Pros, and especially iMacs and Mac Pros). I’m not saying they can’t. I’m just saying they haven’t shown us anything yet.
The problem with phone CPU benchmark tests is that they’re only measuring peak performance. Try running a processor-intensive task on your phone for an hour.
Good luck with that.
Porting macOS, unchanged, stagnant, to ARM as-is would be a massive waste. What’s happening is decidedly not that: it’s very clear that a major transition is happening re the software stack, which is what I’ve argued for for a very long time. iOS and macOS are merging in some way
It’s increasingly looking like the future of the Mac doesn’t look like the Mac as we know it, with a rumored app stack replacement/transplant & a move away from x86.
It’s not the Mac at all. It’s some hellish combination of the worst attributes of iOS and the worst attributes of Tim Cook’s Apple.
People really want to believe their world isn’t about to end I think Apple has been singularly about ARM for the past decade and it would be very wishful thinking to hope they’d change that trajectory now
If you took the current macOS and put it on a desktop class ARM processor, I think it would be a pretty simple transition compared to PPC->Intel. I doubt many would notice, and most apps would just compile out the box. I don’t think it’s the end of anything.
If this is true, it’s another step towards the next great Apple machine: a consumer laptop running iOS. Call it the MacPad, or revive the name iBook. Use the trackpad the way 3D Touch is used on iOS devices to easily move the cursor. (And build more tricks into it.) I’ll buy it.
If Apple builds an ARM-based Mac, what are the hassles involved in porting code from Intel->ARM, beyond recompile?
Anyone who says “you just have to click a check box” or “it’s trivial” without actually having done the transition for a shipping product is engaging in wish fulfillment or marketing.
What you said...
The last PPC Mac shipped in 2006. Rosetta was not removed from Mac OS X until 2011.
The last Intel Core Duo 32-bit Mac shipped in 2007. i386 support has still not been removed.
I have yet to hear a single person give a plausible transition plan.
Is there a Rosetta-like tech? IDK, but remember that the significant performance hit was semi-acceptable only because Intel chips were much faster than PPC. Don’t let anyone tell you otherwise. I had a last-gen iMac G5 and a first-gen iMac Core Duo. Core Duo crushed the G5.
WWDC could be really awkward.
“Here’s our brand new $5k+ Mac Pro”
“Next year we’re deprecating Intel”
Ok, reading this report about Apple contemplating dumping Intel chips for its own put me in a nostalgic mood. Here’s a little story about how Steve Jobs operated with the press.
Previously: Apple Rumored to Combine iPhone, iPad, and Mac Apps to Create One User Experience, Microsoft Launches Windows 10 on ARM.
Update (2018-04-03): Alberto Sendra:
I interpret it differently. If you need a high performance laptop/computer with macOS you need to buy one before 2020 No assurances that pro apps are gonna make the transition anytime soon.
I don’t see why this should come as a surprise. The real question is what their developer experience is going to be like, and how accomodating it will turn out to be for those of us who use Macs as primary devices for cross-platform development.
Those of us who write about Apple have long opined about the iOSification of macOS, and the ability to allow iPhone and iPad apps to run on the Mac will be a big deal. It might not work; or it might only work for very simple apps. But it will be a game-changer. I don’t expect Apple to fully iOSify the Mac platform, but allowing iOS apps to run on Macs in a special environment makes sense.
Update (2018-04-04): Marco Arment:
A bit concerned over the rumors of big changes to macOS.
Apple hasn’t prioritized macOS quality in years, and it seems that they can barely touch it these days without leaving a trail of sloppy bugs.
I’d love 2005-Apple to revamp macOS. I’m not sure I trust 2018-Apple to do it.
Less of a transition, more of a bug multiplier. Maintaining software quality while dramatically expanding the scope of said software is difficult to impossible, especially given the circumstances.
I don’t think Apple would drop Intel completely. It’s easier for me to imagine them using custom CPUs for their consumer-grade Macs and sticking with Intel for the high-horsepower Pro desktops and notebooks. At least for starters.
[…]
ARM is such a huge move — and presents such a big opportunity for change — that I would expect it to accompany a whole new historical age for the Mac. Either Apple would do radical (and long-overdue) modern rethink, akin to what Microsoft did with Windows 10…or they would effectively transform MacOS into an enhanced version of iOS, in function if not in name.
Update (2018-04-06): See also: Accidental Tech Podcast.
Update (2018-04-22): See also: Andy Ihnatko.
Update (2018-09-05): Cherlynn Low (via John Gruber, tweet):
Lenovo is showing off the industry’s first device to use the made-for-PCs mobile chipset: the Yoga C630 WOS (which stands for Windows on Snapdragon).
[…]
The Yoga C630 is supposed to last about 25 hours of continuous local video playback, which should be enough to last through an entire day of running around attending business meetings. That endurance is thanks in large part to the Snapdragon 850, which promises not only 25 percent longer battery life than the 835 but also 30 percent faster performance. When I opened a slew of apps like Excel, PowerPoint, Maps and Edge on the Yoga C630’s desktop environment, I barely encountered any delay. Any interruptions I saw were related to WiFi troubles rather than actual performance.
Previously: ARM Mac Notebook Rumors.
16 Comments RSS · Twitter
I can't imagine how this change wouldn't kill the use of Docker. How could users run x86 linux containers anywhere near as performant on ARM? Docker is why you see most people in sysadmin/devops using Apple computers.
There was a time when I'd have been quite excited by such news. Frankly, at this point, I'm just wondering how the people who can apparently no longer bring us a functional keyboard will implement this, then tell me it's 'better'.
Apple made comments on the Touchbar MacBook Pro’s ARM components helping them get closer to where they want to be. I don’t think it’s far fetched that we could see a MacBook Pro with both ARM and x86 CPUs, similar to the way notebooks currently use integrated graphics and switch to a dedicated GPU for more intensive tasks. ARM for low power tasks and longer battery life, x86 for max power and CPU-intensive tasks. Less expensive hardware could go ARM only.
"ARM for low power tasks and longer battery life, x86 for max power and CPU-intensive tasks." It's hard to imagine something like this considering all the issues that would need to be addressed (e.g. supporting at the same time apps that are ARM only and apps that are x86_64 only for instance). A solution like Rosetta seems easier to implement.
A couple things:
1. Don't imagine that the Ax chips we have seen in phones are the best that Apple can do. If this thing is indeed only two years out, they have been secretly designing and testing laptop and desktop style chips with *much* higher thermal ceilings for years now, and those chips probably blow their phone siblings out of the water both in benchmarks and in the ability to do heavy work for long periods.
2. They will have to nail emulation of x86 code for this to work. Think back, way back, to the transition from 68000 macs to PowerPC macs. Apple made it mostly seamless by implementing hardware emulation for old applications on the new macs. Considering Apple's SOC prowess and how well they have been doing at integrating hardware acceleration for various tasks into their Ax chips, I'd bet that their new chips' x86 emulation will be at least partly done in hardware and will make every other software-only emulator out there look like a slug.
Also, at this point, Apple is big enough to just flat out buy a license to the x86 instruction set if that's what they decide to do.
I'm sure Apple has experimented with an ARM-based Mac. However, does it make business sense: is a custom processor justified for the Mac, considering its sales volume? Furthermore, switching from x64 to ARM would hurt business sales, where developers and other technical users rely on virtualization and Docker. Maybe Apple will split the Mac products into consumer and professional lines, where the consumer moves to ARM and the professional remains on x64. Developers would ship fat binaries that run on both architectures.
3 issues that people fail to overcome when discussing Apple moving Macs away from Intel...
1. People assume Apple is going to switch to ARM. This makes sense because Apple already designs ARM CPU's. However there are two other options; Apple licenses the x64 ISA and designs their own x64 based CPU's, or Apple develops their own architecture and moves ALL their products to it, Mac, iPhone iPad, Apple Watch, etc. Over the past decade if there's one thing Apple has demonstrated is that they do not want to rely on other's technologies and have been more than willing to bring as much as they can in-house.
2. The people that assume Apple will be moving to ARM seem to think that means Apple will simply shoehorn their current SoC designs into a Mac. Why on earth do people think this? Apple is perfectly capable of producing new designs meant for heavy workloads rather than high efficiency. It's the "highly efficient" part that is the most difficult - Intel has struggled with it for years. And we already know what they can do when they decide to open the "speed" gates just a little... the AxX series of chips are beasts. Imagine what would happen if they threw out all concerns of power efficiency?
3. And then there are those that think if Apple moves to ARM it MUST mean they're binging iOS with it. Apple's history has demonstrated that Macs are perfectly capable of moving from one architecture to another without changing the "Mac" part.
Has anyone bothered to question that fact that single source, Bloomberg, is behind this? Or that their Apple predictions are rather dismal?
I am worried about quality assurance every step of the way. The Mac is still the bedrock for all software on any Apple product. Devs and nerds all around the world are already antsy about the Mac – we desperately need some reassuring news come WWDC.
1. The switch to ARM is all about size. That's been Apple's obsession for the last few years. End of story. Thinner laptops and iMacs.
2. There will be an x86 translation layer, maybe in silicon. Remember that x86 chips are RISC micro-architecture at their core, and have been for years (decades?).
3. Has anybody checked when things like intellectual licenses for x86 expire? Could it be maybe around 2020?
Michael:
3. And then there are those that think if Apple moves to ARM it MUST mean they're binging iOS with it. Apple's history has demonstrated that Macs are perfectly capable of moving from one architecture to another without changing the "Mac" part.
I think the concern is that what makes a Mac, well, a Mac, has been a sticking point for a few years now. Better trackpads are great, worse keyboards not so much, dearth of ports, style over function to a detrimental extreme, push for developers to use the Mac App Store, only to abandon any real work on the Mac App Store, not updating hardware for years at a time, providing hardware that doesn't meet anyone's needs (cough, Mac Pro , cough)....I could go on, but....
@Keir Thomas,
x86 (or should we say x64 now?) is a moving target. The patents to the 80286 are probably expired, but that's not going to run anything modern. To run apps made for Intel-based Macs transition, you need the 2006 and later instruction set, which still has a long way to go on its patents.
The merger of iPad and Mac is inevitable because:
- the biggest problem with iPad is the missing Mac features such as Photoshop and Ableton Live and notebook format and 27-inch screens
- the biggest problem with Macs is the missing iPad features such as touch, Pencil, iPad/iPhone apps, low-power video playback/Web browsing/cloud sync
If iOS and Mac were 2 companies, the best thing iOS could do is buy Mac to get the technology for one complete line of computers from watch to workstation all running one hardware/software platform.
I hope the reason that this merger is taking so long is that they are going to do it right. Ideally, out-of-the-box an iMac should respond like an iPad, with no accessories required. The user should then be able to choose any or all of the Apple accessories (keyboard, Pencil, AirPods, mouse, trackpad) without being required to choose any of them because there is built-in touch and software keyboard to fall back on. And you should just be able to run any app written in Xcode plus Web apps and ISO MPEG media. How that works underneath is Apple’s job. Users expect all screens to have touch now and they expect tablets to run PC apps.
This doesn’t need to be disruptive to users. An iMac should come in the box by itself with no accessories and work like an iPad, and the user should be able to choose Apple Pencil, Apple Mouse, Apple TrackPad, or Apple Keyboard accessories (or zero to 4 of those) and they just work. AirPods too, of course. It should all just work basically the same as now except your Macs have additional iPad features and vice versa.
Windows 10 (not W10 mobile) has been shipping on ARM since December on notebooks built around Qualcomm's Snapdragon 835 PC platform. The OS & libraries are ARM-native and its using a funky x86-to-ARM translation layer to run x86 applications at "near native" speeds. There have been demos with Photoshop & Acrobat.
Qualcomm has been working on x86 emulation in-chip but Intel's lawyers put a shot across their bows which is perhaps why W10-on-ARM does the x86 translation in code.
The ultimate aim though is "always on" mobile devices not workstations, pro laptops & the like.
If Apple is planning a major transition, similar in scale to OS 9 -> X, I wonder if major 3rd party apps like Photoshop would go along for the ride this time around... not because it’ll be impossible, but because Adobe (and others) will see the development work as too much to support something that pros may already be balking at (much like the iPad).
Major app makers went along with the OS X transition because their customers were switching, and customers were switching because the benefits were copious and obvious (especially from 10.3 onwards). Any new macOS transition that is announced in this atmosphere of concern about software quality could keep a critical mass of customers from buying new hardware ... and then developers can’t justify spending resources developing for the platform (again, much like the iPad).
I worry that Apple may push for a big transition, thinking (as they do at least in public) that they’re playing from a position of strength, when in reality they have less trust now than in 2001.
I mean, as mentioned, if Microsoft can work with Intel binaries on ARM, then Apple should be able to do better (full 64-bit support for one thing) given they are designing the chips and the software. Could work, but I will miss alternative OS support, assuming Apple makes the Mac more of a closed widget like iOS. Very curious here. This could be kind of neat technologically, but a disaster from a end user controlling their "personal computer". Guess we will see.
