Could x86-64 emulation be possible on ARM Macs?

[JPNMac]

Rosetta 2 will not support virtualization using x86-64. And people are talking about it. But could x86-64 emulation be possible?

 

[Amethyst1]

Highly likely - a port of QEMU would be sufficient for that to happen. The big question is - how fast will it be?

 

[throAU]

definitely possible.

whether it will be fast enough to be useful is another question entirely.

I would argue however that outside of a few niche users, its not really required for the ARM Macs to succeed. Some will whine about it, but most stuff is either running in docker, the web, a cloud server or cross platform these days.

Games aren't but they come and go and more games will come to the Mac with its better ARM hardware and common iPad/iPhone/tv/Mac hardware compatibility in the future.

 

[leman]

A lot of things are possible. Emulation is relatively simple, but it’s unlikely to be useable. Virtual machines could theoretically use their own transpiler from x86 to ARM - similar to what Rosetta does (there are open source libraries that do this). What also might be quite feasible is to leverage Rosetta to run x86 applications in a virtualized environment.

 

[Voyageur]

Games aren't but they come and go and more games will come to the Mac with its better ARM hardware and common iPad/iPhone/tv/Mac hardware compatibility in the future.

I think that it is futile to mention about games on the Mac until popular hits appear in large numbers.

 

[maflynn]

ut could x86-64 emulation be possible?

Its possible, but is it likely? I'm speaking from ignorance, but I believe whomever attempts this will need to have intimate knowledge of the X86 architecture, and that's something I don't believe Parallels has. They are the ones who are losing the most with the move to ARM. Vmware is a huge company and its Mac product accounts for a tiny percentage of its business. Parallels is a one trick pony that sprung to life thanks to Apple's move to Intel, now with Apple's departure, it will be bad news for them. Apple has zero interest in producing an emulator as they want you to run macOS, not windows.

 

[Amethyst1]

Its possible, but is it likely? I'm speaking from ignorance, but I believe whomever attempts this will need to have intimate knowledge of the X86 architecture

I've used QEMU to emulate x86 and x86-64 on ARM and PowerPC so it's definitely possible but slow. Very slow.

 

[maflynn]

I've used QEMU t

I know of qemu, which is why I said its possible.

 

[Amethyst1]

Parallels is a one trick pony

Not to nitpick, but Parallels was around before the Apple-Intel switch, if virtually unknown. Parallels Workstation was released for Linux and Windows in 2005 and subsequently ported to OS X in 2006. It was that port that kicked them into high gear so to speak though.

 

[Nugget]

I would argue however that outside of a few niche users, its not really required for the ARM Macs to succeed. Some will whine about it, but most stuff is either running in docker, the web, a cloud server or cross platform these days.

This is needlessly inflammatory. Some will be impacted by this and their complaining or frustration is understandable and legitimate. It's not "whining" even if you don't like to hear it.

Also, are you aware that the bulk of the Docker ecosystem revolves around x86 docker containers? That compatibility will also evaporate on ARM-based Macs. For anyone who relies on building, running, and testing Docker containers for an x86 production architecture (Kubernetes, etc) the switch to ARM will pose huge challenges to their workflow.

At my company we have about 30 developers who will probably not be able to stick with macOS because they will need to be able to produce and test docker images that can be deployed to our production systems. A handful of us have already switched to Linux or WSL and it seems inevitable that the rest of us will have to follow in that path.

 

[maflynn]

the switch to ARM will pose huge challenges to their workflow.

Unless they decide to stick with the X86 platform, either using an intel Mac or going the Pc route.

I would argue however that outside of a few niche users,

Agreed, though I think the x86 platform sort of had a halo effect if you will, offered opportunities for some people to get a mac (And use windows), or use macos as a hackintosh. Now that's coming to end

Some will whine about it,

That's what we do best as apple fans

 

[jerwin]

some news:

Apple's ARM switch will be the end of Boot Camp

with a quote from an Apple source

We're not direct booting an alternate operating system. Purely virtualization is the route. These hypervisors can be very efficient, so the need to direct boot shouldn't really be the concern."

--Craig Federighi

 

[maflynn]

with a quote from an Apple source

Looks like anyone buying a Mac will be forced to run macOS

 

[DWomack]

QEMU

Take a look at this...

GitHub - utmapp/UTM: Virtual machines for iOS and macOS

Virtual machines for iOS and macOS. Contribute to utmapp/UTM development by creating an account on GitHub.

github.com

 

[throAU]

I think that it is futile to mention about games on the Mac until popular hits appear in large numbers.

I think you'll see a surge of games on the Mac when it shares hardware with the iPad/iPhone/appleTV.

And I don't just mean crappy mobile games. Controller support is now a thing and the market will grow now the Mac is onboard.

I mean as a developer you can target hundreds of millions of iOS devices AND the Mac market now. Thats a far more massive user base than any console or even the PC market for what is now mostly shared effort for all of the above platforms.

edit: also - the base model iPad is already MUCH more powerful than say, the Nintendo Switch....

[automerge]1593126874[/automerge]

Its possible, but is it likely? I'm speaking from ignorance, but I believe whomever attempts this will need to have intimate knowledge of the X86 architecture,

Bochs already exists and is open source (and portable).

Bochs - Wikipedia

en.wikipedia.org

 

[KPOM]

A lot of things are possible. Emulation is relatively simple, but it’s unlikely to be useable. Virtual machines could theoretically use their own transpiler from x86 to ARM - similar to what Rosetta does (there are open source libraries that do this). What also might be quite feasible is to leverage Rosetta to run x86 applications in a virtualized environment.

As long as Rosetta 2 is around (which might not be very long), CodeWeavers thinks their WINE app CrossOver will continue to work. It doesn’t use emulation or virtualization, but operates something like Rosetta in that it translates Win32 and Win64 code into MacOS instructions.
[automerge]1593132048[/automerge]

Looks like anyone buying a Mac will be forced to run macOS

It makes sense. Apple didn’t allow direct booting of alternate OSes in the 68K or PowerPC era. Boot Camp was an accommodation when Apple was specifically targeting Windows users as a growth market. They aren’t doing that anymore.

 

[MichaelDT]

...It makes sense. Apple didn’t allow direct booting of alternate OSes in the 68K or PowerPC era. Boot Camp was an accommodation when Apple was specifically targeting Windows users as a growth market. They aren’t doing that anymore.

Not at first, but by the end you could boot Linux, BSD, even Amiga clone. If I remember originally for Linux is was MkLinux by Apple but eventually direct boot was cracked?

 

[KPOM]

Not at first, but by the end you could boot Linux, BSD, even Amiga clone. If I remember originally for Linux is was MkLinux by Apple but eventually direct boot was cracked?

If it was cracked, then it wasn’t officially supported.

 

[Emanuel Rodriguez]

I know of qemu, which is why I said its possible.

That's not what you said. You'd said it's unlikely because someone would have to have intimate knowledge of x86. That is not required, because qemu exists, and already runs on ARM, and can emulate x86_64 on it. Is it slow? On the raspberry pi I tried it on, yes it is, very much, but it definitely exists.

I would NOT try to run Windows on it, though.

 

[Erehy Dobon]

My guess is that it is possible.

The most pressing question is whether or not the effort to do so will be fiscally beneficial or somehow adequately gratifying.

I don't expect large numbers of programmers to be at my beck-and-call to fulfill all of my wishes free of charge.

 

[MichaelDT]

If it was cracked, then it wasn’t officially supported.

not supported but possible

GRUB on Open Firmware (PowerPC) - Gentoo wiki

wiki.gentoo.org

probably on our way to impossible on ”Apple Silicon” A real shame, my biggest complaint about my Mac in the 90s was how crippled it was due to Apple‘s custom hardware. It seems we are headed back that way sadly. 2006-2011 intel Macs really were a golden age for the platform, modular standard PC like designs, but rock solid apple construction (An engineers dream).

 

[Erehy Dobon]

Sorry, but 2020 Intel is not the same as 2010 Intel.

That's part of the point.

But a lot of technologists can't see the forest for the trees... Apple's senior management does.

In the past 5-7 years, Intel has handed scads of reasons to its partners that it was woefully inept at executing their roadmap on schedule.

And the INTC stock price reflects this. Intel misses targets and the company's stock underperforms every single relevant metric in the past five years: S&P500, Nasdaq Composite, Nasdaq-100, SOX semiconductor index, you name it.

Intel didn't just fail Apple. They failed pretty much EVERYONE.

And Intel isn't just a lone case. How many people do you know who are touting the excellence of Cisco Systems in the past five years? AMD can't capitalize on Intel's stumblings as much as Nvidia has.

 

[leman]

As long as Rosetta 2 is around (which might not be very long), CodeWeavers thinks their WINE app CrossOver will continue to work. It doesn’t use emulation or virtualization, but operates something like Rosetta in that it translates Win32 and Win64 code into MacOS instructions.

And I believe Parallels etc. works similarly. They insert traps or otherwise catch some system calls, but the virtualized app itself runs as “native” code. Given that ARM has excellent virtualization support, I think that utilizing Rosetta here will be possible.

 

[cknlol]

Sorry, but 2020 Intel is not the same as 2010 Intel.

That's part of the point.

But a lot of technologists can't see the forest for the trees... Apple's senior management does.

In the past 5-7 years, Intel has handed scads of reasons to its partners that it was woefully inept at executing their roadmap on schedule.

And the INTC stock price reflects this. Intel misses targets and the company's stock underperforms every single relevant metric in the past five years: S&P500, Nasdaq Composite, Nasdaq-100, SOX semiconductor index, you name it.

Intel didn't just fail Apple. They failed pretty much EVERYONE.

And Intel isn't just a lone case. How many people do you know who are touting the excellence of Cisco Systems in the past five years? AMD can't capitalize on Intel's stumblings as much as Nvidia has.

It's the trend of the industry if you've been a sole market leader without competition for so long as Intel has.

 

[donawalt]

Apple has zero interest in producing an emulator as they want you to run macOS, not windows.

I don't agree with this. The days of '100% Apple' have been long over at Apple, which is why their success has been phenomenal in the last 10 years. Why so aggressive in getting a quality MS Office implementation on Mac? Why allow Windows users to run iCloud and Music apps? Why all the investment to make sure Windows, Linux, etc. apps run well on Macs?

While the percentage of mac users who are serious in needing to run Windows virtually may be low, I wonder if the percentage of users who buy pro/high end mac computers and run Windows isn't much higher.

Apple wants its mac to be the hub of all its pro users do, it wants Apple Silicon to be a home run in all respects - otherwise this announcement is perceived as a decision like 'Apple wanted to get away from Intel, even though their product will be better in some ways, worse in others'. Virtualization/emulation, however they solve it *well*, could be one of the last things the ARM transition completes, but high end pro machines/users has been an area Apple has shown increasing interest in over the last couple of years.

High end pro users pay more, and their computers are higher profit margin for Apple too. It's a lucrative market. And Apple could lose that market in this transition, much less not see it grow. Consider this - while many are saying "Buy a Windows machine to supplement your mac for those legacy apps", there is in reality a much better chance that ALL the apps, including the Mac ones, run on Windows already. Buy a Mac and Windows computer, or just buy one Windows computer? I think more people would buy a Windows computer for that next purchase, and dump the mac - instead of now having to support/maintain/use two uniquely different computers - that's a total pain. In that scenario, Apple is then back to macs being considered a toy/light consumer computer only - back to 1985.

Apple just bought a company called Fleetsmith, which offers device management products for the enterprise - a purchase for Apple that only make sense if Apple wants to move deeper into the enterprise. Over half of all businesses now support Macs. If Apple is interested in thriving in the enterprise, they surely are interested in integrating the world outside their Apple ecosystem, imho.

I think we don't know the answer because it won't be revealed until they know how well it will work, that it will be successful, in what timeframe, on what computers, with what partners helping, etc. There is reason to be optimistic I think, but it requires patience. ?

Edit: @Joelist just said the same thing about the enterprise risk in another thread:

This is a big gamble for Apple no doubt. They are putting their entire move into the Enterprise at risk because those large businesses tend to have their own software (x86 almost invariably). So far Apple being on x86 has made that issue moot and they have been gaining Enterprise traction but if they don't really deliver better performance and (more importantly) a painless path for those corporate apps the Mac move in the Enterprise will die. You even saw this in the announcement where they spent a small period of time on iWork but a lot of time showing off Microsoft Office on Apple Silicon - that is aimed at the Enterprise.