Instead of Rosetta 2, why Apple doesn't make a Windows x86 emulator?

[Gerdi]

Apple Silicon does not use UEFI or ACPI. And neither of these will tell the system how to handle basic low level interrupts or program the memory controller.

So what? Raspberry PI does not have UEFI and ACPI either. Instead the community provided UEFI and ACPI and Windows boots happily without any kernel changes.
Besides, programming the memory controller is not done from Windows directly - it can request certain power and performance states via ACPI.
Interrupt controllers are described in the MADT within ACPI including the system interrupt model.

I also put additional information into my last post. This should make it more clear, that the Asahi team did not go with UEFI and ACPI - they instead need to opt for kernel changes.

 

[aidler]

Because it would be something truly useful for customers…

 

[leman]

So what? Raspberry PI does not have UEFI and ACPI either. Instead the community provided UEFI and ACPI and Windows boots happily without any kernel changes.
Besides, programming the memory controller is not done from Windows directly - it can request certain power and performance states via ACPI. Interrupt controllers are described in the MADT within ACPI including the system interrupt model.

Raspberry PI uses a regular ARM CPU that follows the spec. Windows on ARM kernel is developed for the spec. But M1 does not follow the spec. That's the very point. Even if you provide a boot loader that emulates EFI and somehow get Windows kernel to boot it won't be able to do the basic stuff like handle hardware interrupts or send messages between the CPU cores. ACPI here or there, but Apple does not use a standard-compliant interrupt controller.

What you are essentially describing is providing some kind of a low-level hypervisor that will boot up and emulate a standard Windows-compliant ARM system which would allow Windows to load. And I suppose that you suggest that Apple provides and maintains that hypervisor. But at this point... what is a difference from just running Windows in a VM to start with?

 

[Lounge vibes 05]

Obviously, but that's a different question altogether.


You mean like not going with Intel processors and creating bootcamp?

You anthropomorphize way too much.

Apple created boot camp because they already knew people were going to force X86 Windows on an X86 Macintosh.
It was really a no brainer.
And they have never spoken out against, in the future, allowing Windows on ARM to run on M1 computers.
From everything they’ve said, it’s Microsoft who’s not letting that happen.
As for running Windows applications directly in macOS, yeah that’s just ridiculous and never going to happen.
As people have already said, that would reduce the incentive for developers to produce Mac applications down to zero.

 

[Lounge vibes 05]

If you're talking to me I couldn't tell, might want to include a quote. If not, never mind. (and that's riddance)

As for the 100% converted, well, they are unconverting some now. We'll see if it becomes a trend. I'm not holding my breath either way.

I don’t think it’s a trend, the Mac has been on an upward swing since the first M1 Machines launched, while the rest of the PC industry has been on a major decline

 

[Colstan]

It seems that every month we get a variation of "I want Windows on my Mac" posts, this is just a new variation on it. @leman has already covered the reasons why this technically doesn't make sense because of the effort involved. Apple also isn't going to do it because they know that the expense isn't worth it, since it doesn't make business sense. Five years after the Intel switch, according to a provider that services Macs, 15% of Intel Macs had Boot Camp installed in 2010. That number reduced to 2% by 2020, based upon Apple Insider's source.

Apple Insider did an informal poll among their visitors in 2020 and found that 35% of their readership had Boot Camp installed. The folks who visit Apple-focused websites and engage in debates on forums, such as that here at MacRumors, don't represent the average Mac customer. While I'm sure there are better metrics available, according to the latest Steam survey, nearly 38% of Mac gamers are on Apple Silicon. (I'm not using Steam to speak for the overall market, just the trend.)

How many users (outside of this forum) are still using Boot Camp or are desperate for Windows compatibility beyond what Parallels and CrossOver provide? Between the technical reasons, manpower involved, and lack of financial or business incentive, neither Apple nor Microsoft are going to put any significant effort into making Windows apps run on the Mac in the fashion that some desire.

I realize that five years from now people will still be asking for Boot Camp, eGPU support, Windows emulation, and all manner of things that they want to magic into existence because it fits their personal wish list, but it's best to dispel these notions whenever possible. Apple showed Parallels during WWDC 2020 and have helped CodeWeavers with Rosetta 2 compatibility using CrossOver, which they probably think is plenty enough.

I don’t think it’s a trend, the Mac has been on an upward swing since the first M1 Machines launched, while the rest of the PC industry has been on a major decline

The Mac just gained marketshare last quarter, up 4.3% while the general PC market declined by -5.1% year-over-year. Since the switch to Apple Silicon, Apple has repeatedly had record revenue for the Mac for multiple quarters now. What they are doing is working, so I see no logical reason that Apple should deviate from their current strategy.

 

[kevcube]

It would probably be a hefty fee paid to Microsoft for licensing, or some legal battles. Emulators are in a legal grey area for consumers, how businesses would handle it is tricky.

Yes emulators can be a gray area, if you’re emulating Nintendo games. Rosetta 2 is translation, not emulation. And then there’s virtualization…

 

[Ethosik]

Yes emulators can be a gray area, if you’re emulating Nintendo games. Rosetta 2 is translation, not emulation. And then there’s virtualization…

That is why I was speaking to the original post. Asking Apple to develop a full Windows x86 emulator

 

[Digitalguy]

It seems that every month we get a variation of "I want Windows on my Mac" posts, this is just a new variation on it. @leman has already covered the reasons why this technically doesn't make sense because of the effort involved. Apple also isn't going to do it because they know that the expense isn't worth it, since it doesn't make business sense. Five years after the Intel switch, according to a provider that services Macs, 15% of Intel Macs had Boot Camp installed in 2010. That number reduced to 2% by 2020, based upon Apple Insider's source.

Apple Insider did an informal poll among their visitors in 2020 and found that 35% of their readership had Boot Camp installed. The folks who visit Apple-focused websites and engage in debates on forums, such as that here at MacRumors, don't represent the average Mac customer. While I'm sure there are better metrics available, according to the latest Steam survey, nearly 38% of Mac gamers are on Apple Silicon. (I'm not using Steam to speak for the overall market, just the trend.)

How many users (outside of this forum) are still using Boot Camp or are desperate for Windows compatibility beyond what Parallels and CrossOver provide? Between the technical reasons, manpower involved, and lack of financial or business incentive, neither Apple nor Microsoft are going to put any significant effort into making Windows apps run on the Mac in the fashion that some desire.

I realize that five years from now people will still be asking for Boot Camp, eGPU support, Windows emulation, and all manner of things that they want to magic into existence because it fits their personal wish list, but it's best to dispel these notions whenever possible. Apple showed Parallels during WWDC 2020 and have helped CodeWeavers with Rosetta 2 compatibility using CrossOver, which they probably think is plenty enough.


The Mac just gained marketshare last quarter, up 4.3% while the general PC market declined by -5.1% year-over-year. Since the switch to Apple Silicon, Apple has repeatedly had record revenue for the Mac for multiple quarters now. What they are doing is working, so I see no logical reason that Apple should deviate from their current strategy.

I think your post is mixing/confusing bootcamp with the emulator. Apple said clearly they are ready to support Bootcamp of Windows on Arm. And to be honest I don't think that would require more resources than Intel Bootcamp. The bulk of the work is on Microsoft to adapt Windows on Arm to Apple Silicon, not Apple.
What Apple is definitely never going to do is the Windows emulator suggested by the OP. This would require far more resources and might even fail to provide acceptable results.
As for Apple Silicon market share, they only moved from 8.1% to 8.9%, much less that what I was expecting.... I had predicted Apple Silicon to bring Macs solidly in the double digit market share in a couple of years, possibly doubling the market share to over 15%, or even get to 20%, and that doesn't seem to be happening at all. I guess most M1 buyers were previous Mac owners, not so much Windows users....

 

[Tagbert]

I don’t think it will happen and I don’t understand why/how it could replace Rosetta.

True. the question conflates two separate things.

• Rosetta allows Mac apps written for the Intel processors run on Arm processors under Mac OS. It depends on those apps being written in modern Mac libraries.
• A Windows/x86 emulator would emulate an Intel PC to allow Windows and Windows apps to run.

Those are two completely different features and scope of work. Rosetta exists and fits well with Apple's priorities. The emulator is a big, hard problem and would take lots more resources to build and the results would probably not be as good as we would like. At the least it would be slower.

 

[Tagbert]

Licensing would be the biggest thing. The Windows APIs are copyrighted, and others have gotten in a lot of trouble doing it.

We just need to wait for the Microsoft/Qualcomm exclusivity agreement to expire, so Microsoft can sell Windows for ARM, then we can just use Bootcamp, for that software that is Windows only.

No, Windows APIs are not copyrighted, as much as Oracle has tried to make API copyrights a thing. So far it has failed in the courts. There may be some other issues that would make this non feasible, but copyright is not one.

 

[Nermal]

They could get the advantages of letting people run Windows for little cost.

I actually wonder whether it might have been a net saving. Boot Camp didn't come along until ~3 months after the first Intel Macs. Prior to that, people were trying to install Windows anyway and there were plenty of forum posts from people who had managed to kill the EFI partition. "I tried to install Windows and now I just get a flashing cursor in the corner of the screen". Boot Camp was technically very simple and it may have actually been cheaper to develop than the support costs of fixing up the 'bricked' Macs.

Or it may not have been

 

[mr_roboto]

So what? Raspberry PI does not have UEFI and ACPI either. Instead the community provided UEFI and ACPI and Windows boots happily without any kernel changes.
Besides, programming the memory controller is not done from Windows directly - it can request certain power and performance states via ACPI.
Interrupt controllers are described in the MADT within ACPI including the system interrupt model.

I also put additional information into my last post. This should make it more clear, that the Asahi team did not go with UEFI and ACPI - they instead need to opt for kernel changes.

You really don't understand this as well as you think you do. The Asahi Linux bootloader chain includes U-Boot, which provides a UEFI environment. Yet Asahi also requires kernel level changes.

There are simply too many fundamental differences between Apple's platform and what the rest of the Arm world has standardized on. Several are things which UEFI does not provide adequate abstractions to paper over.

In the absence of kernel changes, Windows cannot boot on Apple Silicon unless it is running inside a hypervisor.

 

[Colstan]

As an addendum to my previous post which featured IDC's numbers, the front page is showing numbers from Gartner that places the Mac's marketshare gains at +8.6% while the overall market declined by -7.3%. In the U.S. market, those figures are +18.6% and -16.5%.

Some here have predicted that Apple will suffer without Windows support. Marketshare data from analysts and quarterly revenue directly from Apple show the opposite. If Apple Silicon isn't a success in a mature market, then I don't know what is, and they did it without Boot Camp or Windows emulation. As I said, the folks on this forum don't represent the average consumer, which is a tough pill to swallow for those who desire greater Windows application support.

I think your post is mixing/confusing bootcamp with the emulator.

No, I know exactly what the difference is. From my perspective, Boot Camp and an emulator are attempts to achieve the same goal, and my point was that neither is worth the effort. Apple has provided support for Parallels and CrossOver and shown no indication that they are going to go beyond that.

Apple said clearly they are ready to support Bootcamp of Windows on Arm.

No, Apple has said that support is up Microsoft for Windows on ARM Macs. Please show me where Apple has stated that they are ready to support Boot Camp on Apple Silicon. I've seen people take that as an inference, but Craig said no such thing, he simply stated that the underlying technology is there, much like it is for Asahi Linux. He made no mention of Boot Camp and gave no indication that they would work with Microsoft on such a feature. Microsoft's only comment has been that it isn't a supported scenario.

 

[Digitalguy]

As an addendum to my previous post which featured IDC's numbers, the front page is showing numbers from Gartner that places the Mac's marketshare gains at +8.6% while the overall market declined by -7.3%. In the U.S. market, those figures are +18.6% and -16.5%.

Some here have predicted that Apple will suffer without Windows support. Marketshare data from analysts and quarterly revenue directly from Apple show the opposite. If Apple Silicon isn't a success in a mature market, then I don't know what is, and they did it without Boot Camp or Windows emulation. As I said, the folks on this forum don't represent the average consumer, which is a tough pill to swallow for those who desire greater Windows application support.

No, I know exactly what the difference is. From my perspective, Boot Camp and an emulator are attempts to achieve the same goal, and my point was that neither is worth the effort. Apple has provided support for Parallels and CrossOver and shown no indication that they are going to go beyond that.

No, Apple has said that support is up Microsoft for Windows on ARM Macs. Please show me where Apple has stated that they are ready to support Boot Camp on Apple Silicon. I've seen people take that as an inference, but Craig said no such thing, he simply stated that the underlying technology is there, much like it is for Asahi Linux. He made no mention of Boot Camp and gave no indication that they would work with Microsoft on such a feature. Microsoft's only comment has been that it isn't a supported scenario.

I think that's obvious that they were referring to bootcamp and not to parallels or crossover. They don't even need to mention the word bootcamp.
Again, we can agree that an emulator is not worth the effort, but we can agree to disagree that bootcamp is too much effort for Apple.
That doesn't mean it's going to make much difference in Mac's market share, and on this we can agree. I don't see many people buying Macs to only or mainly use them on bootcamp with Windows on Arm.
I would say Apple would see it as something positive if Microsoft decided to support it, but not a big deal at all, given that Parallels already exist for those who need to run some Windows only software.

 

[HenrikWivel]

Back in the days, IBM made that mistake with OS/2. IBM added a Windows 3.1 emulator to the operating system, thinking it would lure Windows users to OS/2 and that they would then “see the light” and switch, but it didn’t happen. What they did was remove the incentive to write native OS/2 apps, which helped the demise of OS/2.

As an operating system OS/2 was far superior to windows 3.1, and the funny part was that IBM’s version of Windows was more stable that native windows running in top of DOS.

 

[PianoPro]

As mentioned earlier, the Windows 11 VM possibility exists and works well and can be be activated with a Windows licence (retail or OEM). Some people are afraid that this would not be totally ok if Microsoft audits their computer, since it does not match the EULA for Windows on Arm.

An ARM version of a Windows 11 VM would only let me run proprietary x86 engineering software if Microsoft provided x86 emulation in the OS that worked very well with speed. I don't know how much incentive they have to do that since unlike Apple with Rosetta 2 they aren't committing to ARM as the future of Windows and don't really need to build that bridge.

I think the days of effectively running x86 Windows apps and MacOS on the same AS Mac are probably over for good. It's not the end of the world and we move on.

 

[4743913]

So you are saying Microsoft who purposely cripples its products on non-windows platforms, refuses to license Windows on ARM, would not freely license windows on Mac? Not going to ever happen bucko - you know, Microsoft owns those rights, right?

if this were the case, all those mac users would not be using Visual Studio Code..

 

[4743913]

I think the days of effectively running x86 Windows apps and MacOS on the same AS Mac are probably over for good. It's not the end of the world and we move on.

not for me or my apple care'd forever i9 macbook pro and iMac

 

[KPOM]

Microsoft doesn't need Apple to ship a high volume of Windows on Arm systems. Both Apple and Microsoft are ducking part of the work here. It is convenient 'finger pointing' for each to point at the other. Apple has a bunch of higher priority issues to work out to make the transition to M-series evolve well for Macs. (tons of unoptimized software , new SoCs in the pipeline that will also need optimizations , etc. ) and Microsoft has big issues too ( getting Windows 11 off the ground , keeping Intel , AMD , and the potential Arm CPU vendor(s) happy , etc. )

Also, Qualcomm doesn't yet have an ARM chip that is remotely competitive with the M1, much less the M1 Pro/Max/Ultra. Microsoft puts its own name on the Surface Pro X. It would be somewhat embarrassing for them for the fastest supported PCs running Windows on ARM to be Macs.

 

[bobcomer]

I don’t think it’s a trend, the Mac has been on an upward swing since the first M1 Machines launched, while the rest of the PC industry has been on a major decline

Apple still has a very long way to go to catch up...

 

[bobcomer]

Back in the days, IBM made that mistake with OS/2. IBM added a Windows 3.1 emulator to the operating system, thinking it would lure Windows users to OS/2 and that they would then “see the light” and switch, but it didn’t happen. What they did was remove the incentive to write native OS/2 apps, which helped the demise of OS/2.

As an operating system OS/2 was far superior to windows 3.1, and the funny part was that IBM’s version of Windows was more stable that native windows running in top of DOS.

Just to be a bit technical, Windows 3.1 didn't run in an emulator, it was much closer to virtualized. Not HW virtualized of course, but since Windows 3.1 and OS/2 ran on the same processor. What it really was was a real Windows 3.1 running on top of a DOS shell, on top of OS/2. Them were the days...

I really liked PS/2 hardware, so much more advanced than what passed for a computer in DOS days.

 

[genovelle]

But that disproves what you said. (That they want nothing to do with running Windows on a Mac) There's also the work they did when Apple was still PPC.

Anyway, to answer the original question about Apple, they don't believe there's enough people like me to spend the money to do it, and with that I agree. Most people like me and have the same needs, wouldn't even look at a Mac after the change to the M1. I only do because, as a hobby, I'm a total OS geek, I run them all. I certainly don't buy a Mac because of work. The best way to use a PC/Windows on a Mac is via remote desktop, which really doesn't have anything to do with the Mac at all. Sure, I'd like to consolidate the number of machines I have, but it is what it is.

Interestingly sales seem to be accelerating since they ditched Intel and Windows compatibility. Their strategy is working once again.

 

[KPOM]

Interestingly sales seem to be accelerating since they ditched Intel and Windows compatibility. Their strategy is working once again.

Most Mac users don’t need Windows compatibility. And the M1 series chips are impressive. The MacBook Air and iMac effectively got price cuts since there are no processor variants. $999 is the price of entry and it’s just RAM and SSD options. Everyone gets the same speedy processor.

It’s a different story for the 14”/16” MacBook Pro and Mac Studio but even there the entry level models are very compelling propositions for many buyers.

 

[TiggrToo]

Because it would be something truly useful for customers…

For a very small number of users. At which point, why bother.