Apple and the ARM Transition

I didn’t have any particular posts planned for this week, but as is often the case, the news offered a useful topic: Apple’s upcoming ARM transition.


This is a companion discussion topic for the original entry at https://www.sevarg.net/2020/06/21/apple-and-arm-transition/

(Comments from Blogger)

2020-06-20 by gpshead

A colleague at Transmeta had a special project that demoed win32 applications running on a PlayStation 2 (mips). Library level transitions were the key.

Regardless if apple ships an emulation solution again, I expect they won’t concentrate on performance. Ship it, use it for a year or two only fixing bugs for key large application owners, and abandon it. Same as previous transitions. It’s a means to an end, not a core function.


2020-07-22 by Unknown

Interesting post. But to be a stickler, I’d argue that ARM would actually be Apple’s fourth transition (after all, the Apple II used a Mostek 6502). But I am showing my age.


2020-08-01 by Tyler

“Intel is the king” of x86? Really?
The latest AMD processors are killing Intel in performance and efficiency so I wouldn’t say that Intel is still the undisputed “king” of x86 architecture. They may have a much larger share of the marketplace but AMD is starting to make some gains back in that regard now as well.


2020-08-02 by Russell Graves

AMD has only recently caught up to Intel again - while impressive, they’re still not what most people think of in terms of x86. Though with Intel’s failings lately, and their long delays, well… that may not be as true going forward.

Though I really do expect things to just transition to ARM in the data center. Raw single threaded speed matters a whole lot less, and if the ARM chips are cheaper to set up and cheaper to run, well… people will port apps over.


2020-08-16 by Food Shaman

I realise this is a bit off topic but your article sparked something. What I would like to see, and I’m sure these developments will facilitate such a direction, is a whole stack vendor (hopefully another one besides Apple) offering an ecosystem solution as follows: The customer owns a mobile device which is carried in their pocket. This device will be their main device. Said device slides into a laptop or desktop, which in turn functions as a glorified terminal for the device. This loading function should bring a form of nostalgia related to playing NES games in the 80s. The laptop or desktop does enable further functions, such as additional storage and so forth. However, it should also be able to, relatively seamlessly, increase the the performance. The architecture enabling this would be something like a Pi-cluster. This way, if someone wants to increase the performance of their setup, they can simply add more SoCs.


2020-08-16 by Russell Graves

I’m not entirely sure what problem that solution aims to solve.

You end up with two devices bound to one another instead of separate devices, and it’s hard to optimize a processor for both power-first performance (as in a mobile device) and performance-first (as in a desktop or some laptops). Throwing more CPU cores doesn’t solve a problem when the problem is inherently serial/single threaded as many are.

Now, I’m surprised we haven’t seen this, just because it would allow companies to sell both a laptop-ish thing and a new phone every few years instead of just one or the other. But I’m far from convinced it solves any real issue I have.


2020-11-24 by Unknown

Let’s not forget this stymies ‘hackintosh’. Which used to work with some patches on a lot of old Intel hardware which could be used with MacOS and possibly be faster than a real mac for much less money. Also zero money went to Apple.

Assuming you could even get ARM MacOS running, if Apple have the fastest hardware it negates the point of many Hackintosh builds.


2020-11-24 by Russell Graves

I don’t see Apple caring one way or another about hackintosh, and it’s certainly not a consideration at all in their decisions. They’ve gone pretty hard after the few companies that tried to sell them commercially (rightly so, IMO, because it was in clear violation of the OS X licensing terms), and haven’t really bothered with those who mess with it - but neither have they gone out of their way to make it easy. So, any impact on hackintoshes are of no concern to them.