Tuesday, June 9, 2020 [Tweets] [Favorites]

ARM Macs to Be Announced at WWDC 2020

Mark Gurman (tweet, Hacker News, MacRumors, AppleInsider):

The company is holding WWDC the week of June 22. Unveiling the initiative, codenamed Kalamata, at the event would give outside developers time to adjust before new Macs roll out in 2021, the people said.

[…]

Inside Apple, tests of new Macs with the Arm-based chips have shown sizable improvements over Intel-powered versions, specifically in graphics performance and apps using artificial intelligence, the people said. Apple’s processors are also more power-efficient than Intel’s, which may mean thinner and lighter Mac laptops in the future.

Previously:

Update (2020-06-11): John Gruber (tweet):

I don’t think the transition from x86 to ARM will be nearly as rocky as the 32-bit to 64-bit transition, but for some apps it will take time.

[…]

Will ARM Macs run older x86 software via emulation? Apple shipped a rather amazing emulator in the transition from Motorola 680x0 chips to PowerPC in the 1990s, and again in the 2000s with the PowerPC to Intel transition (Rosetta). There are seemingly no rumors one way or the other regarding emulation for the Intel-to-ARM transition. If I had to bet right now, I’d say no, there will be no x86 emulation on ARM Macs — and that factors into why Apple is pre-announcing this transition months ahead of releasing hardware.

[…]

What happens with virtualization software like Parallels and VMware Fusion? The obvious answer is they emulate x86 or they go away. How will that perform? My understanding is that however good the performance of ARM chips is, the instruction set differences make it slow for ARM to emulate x86. This is no little thing — there are a lot of developers whose workflows depend on virtualization software.

Months is not a long time considering how long the Intel transition took and that these days developers are still cleaning up after new versions of macOS six months after WWDC.

See also:

Update (2020-06-22): See also:

11 Comments

"Inside Apple, tests of new Macs with the Arm-based chips have shown sizable improvements over Intel-powered versions, specifically in graphics performance and apps using artificial intelligence, the people said. Apple’s processors are also more power-efficient than Intel’s, which may mean thinner and lighter Mac laptops in the future."

ARM chips and thinner laptops… I'm sure this is exactly the message that owners of a Mac Pro Late 2019 or a butterfly-keyboard MacBook Pro were waiting for.

Time to throw all Tunderbolt 3 accessories out I guess...

Lack if Intel x86 microprocessors in Mac is a deal breaker for us. We love the Mac and hate Windows interface. But much more important than that is our workflow. And for that we need full Intel x86 compatibility.

For instance, when we use Microsoft Office for Mac (including track changes in documentes when collaborating for manuscripts, PhD dissertations, PowerPoint presentations with animations, video, special protein fonts, transitions, etc), Clarivate Analytics EndNote for bibliographic management or other applications like DNAStar Lasergene or Molecular Biology Insights Oligo, among many others.

I am not talking here only about Boot Camp or VMware Fusion to run Windows (which is also a must for us to electronically sign some documents for research project grant application, etc), but mainly for working on Mac with Mac native applications that are fully native with 90% of the world that use Windows in x86. We also need the power of Mac desktops, including Mac Pro on x86 for bioinformatics. If Apple switches Mac to ARM, we will be forced to switch to PC with Windows. A shame for all!

There is an opportunity for Apple to use this change as an opportunity to introduce lower-priced Macs while still earning the margins they want. If Apple can sell a $299 ARM-powered iPad for education, they should be able to make a $349 ARM-powered MacBook, no? It would be nice if Cocoa/AppKit were accessible to people who can't otherwise afford the entry level MacBook Air.

@TeX
If the Windows interface is going to be a rough adjustment, it may be worth checking out Pop!_OS, a Linux distribution from a laptop/desktop manufacturer. The interface tries to be inviting for workstation and data analysis work (Nvidia/CUDA drivers are pre-installed for instance), and Linux can run Windows pretty easily via its KVM virtualization software. I haven't used this distribution myself yet, but it is on the shortlist of things I will try as a macOS alternative.

Presumably Microsoft will recompile Office for ARM Macs, just as they do for iOS and Android, and I would expect they have a slow but usable Intel emulation on it, just as they had Rosetta for PPC apps on Intel.

Do people really not remember previous processor transitions?

@Tex

Given that Microsoft have maintained a version of Office for ARM on Windows for several years now; and given that the Mac Office product has already negotiated PowerPC, to x86 to x86-64; I expect Office for ARM to be released the same day ARM Macs are relased.

Further, I fully expect the reason Apple somewhat brutally killed off 32-bit support in Catalina is precisely because they have built an ARM-to-x86 translator (a la Rosetta) and only wanted to support x86-64. Doing it early in Catalina means that they took the reputational hit in advance, so there will be no downside in the press when ARM launches.

The issues for professionals are not Microsoft or Adobe tools. It's tooling that required VMWare, or Bootcamp. It's odd apps that were Windows/Intel only; or the esoteric engineering/scientific-computing apps that were available for Mac, but semi-maintained and may not be ported to ARM. Moreover for machine-learning and scientific computing, there is no Metal backend for Tensorflow, or PyTorch, or Caffe; so if they go off AMD GPUs, data-scientists will have to make do with Swift and CreateML, which is nonsense, so they'll buy Windows or Linux laptops instead. Then again, AMD GPUs aren't especially easy to work with, so maybe most have already left.

This would be a good opportunity to introduce Macs with touch screens. They'd reach feature-parity with Surface and other Windows devices.

I hope ARM Macs run macOS and not iPadOS.

I hope I'm not limited to installing software from an App Store.

As far as ARM developer hardware is concerned, I wonder how possible it would be for the 10.16 beta to enable loading a dev environment onto the A10-based T2 chip. It would be an Apple-esque flourish to end the announcement by saying, "And every Mac we ship today can already be used to develop for ARM Macs of the future."

They could have just switched to a competent supplier, and use Renoir.

They could have just switched to a competent supplier, and use Renoir.

Renoir is manufactured in part by TSMC, the same supplier Apple uses.

If you mean in terms of CPU design: Renoir is great, but Zen 2 is just one microarchitecture. Intel is not doing so hot right now, but Ice Lake is looking promising, and Tiger Lake is as well, so they’ll probably recover. Apple is undisputably doing amazing, at least on single-threaded performance.

I don’t know if these rumors are true, but regarding the question of why Apple has never offered an AMD CPU, my guess is they want a CPU partner who has a promising long-term roadmap, not a select few hit products, and they’ve decided either Intel or Apple themselves is a better choice.

This will be the end of hackintoshes. macOS will get more locked down. Probably require a T2 chip or newer. They'll probably use the opportunity to make it more iOS-like, ex App Store will be the only way to install anything. No more kexts. etc.

Stay up-to-date by subscribing to the Comments RSS Feed for this post.

Leave a Comment