Could x86-64 emulation be possible on ARM Macs?

[Kostask]

What is it that you imagine a "virtualization core" is? What would be an example of an existing CPU that has something like this? What you are saying makes absolutely no sense to me at all.

It would be like a supervisory core, possibly used only at boot. I am not the only one talking about this, nor did i come up with it on my own. Rene Ritchie has already done a video involving this. Rather obvious that you never saw the Rene Ritchie video that I posed, and even directed people to the time stamp at.

It doesn't have to make sense to you, unless you are part of the AS design team. There will be a virtualization core, as stated by Apple in the WWDC introduction, and in Rene Ritchie's video. As far as I know, that extra core will be present in the AS SoC on day 1, and is yet another reason that the A14X (the imagined next iPad Pro core) will NOT be used in the upcoming AS Macs.

 

[Nugget]

It would be like a supervisory core, possibly used only at boot . . . Rather obvious that you never saw the Rene Ritchie video that I posed, and even directed people to the time stamp at.

I did watch the video you linked and he never uses the phrase "virtualization core" so I was curious what exactly it is that you are imagining. The guy in the video appears to have only a vague and imprecise understanding of how virtualization works and is -- like I assume you to be -- speculating about a subject he knows very little about. He's coming at the issue from a usability perspective (which is valid) and is definitely not predicting the specific things you say he is.

It doesn't have to make sense to you, unless you are part of the AS design team. There will be a virtualization core, as stated by Apple in the WWDC introduction, and in Rene Ritchie's video.

Apple have also never used the phrase "virtualization core" and they demonstrated virtualization (of Arm environments, of course) working on the developer kit silicon. This is what makes your claim that additional hardware is needed in order to do something which even the dev kit is doing today.

However none of this has anything to do with x86/amd64 emulation which is the subject of this thread. Are you asserting that Apple plan to add hardware support for x86/amd64 instruction set virtualization in future Apple Silicon Macs? Because that's something that neither Apple or Ritchie have claimed. That's something you seem to have invented. There are a number of significant technical, electrical, and licensing barriers to that being possible or feasible.

I'd invite you to re-visit @theluggage's post #99 in this thread. I think you just don't understand this subject well enough to have a meaningful opinion at this point.

 

[Kostask]

They called it a hypervisor core, and it is in both the WWDC presentation, and in the Rene Ritchie video, in fact Rene Ritchie uses it to explain why it makes it so an iPad SoC cannot be used in an AS Mac.

Furthermore, there is ample evidence that Apple will be using a dedicated hardware hypervisor core, the core that the WWDC and Rene Ritchie were talking about, with some sort of lightweight operating system probably with virtualization software as part of the lightweight OS, to host MacOS as a default. And to further get back into the original subject of this thread, that same hypervisor hardware/lightweight operating system could be used to host Windows, probably a Windows 10 ARM version customized for the AS Macs if Apple and Microsoft can work out licensing.

 

[Nugget]

They called it a hypervisor core, and it is in both the WWDC presentation, and in the Rene Ritchie video, in fact Rene Ritchie uses it to explain why it makes it so an iPad SoC cannot be used in an AS Mac.

You have a very fanciful imagination. The phrase “hypervisor core” does not exist anywhere in the Rene Ritchie video or in the WWDC presentations that I have seen.

Furthermore, there is ample evidence that Apple will be using a dedicated hardware hypervisor core, the core that the WWDC and Rene Ritchie were talking about, with some sort of lightweight operating system probably with virtualization software as part of the lightweight OS, to host MacOS as a default. And to further get back into the original subject of this thread, that same hypervisor hardware/lightweight operating system could be used to host Windows, probably a Windows 10 ARM version customized for the AS Macs if Apple and Microsoft can work out licensing.

There is no evidence of this. The scenario you are describing here makes no sense at all from a technical standpoint. And lastly, it still does not in any way address the subject of this thread which continues to be “x86/amd64 emulation on Apple Silicon machines.”. Even if Apple were to build the Frankenstein architecture you’re describing here, it wouldn’t be a path towards performant x86/amd64 emulation in the Apple Silicon Arm environment.

 

[Kostask]

So the part about hypervisor accelerator blocks, and how Apple will be putting them into silicon just flew over your head, did it? Again, it is in the Rene Ritchie video, and start listening at 4:23-4:34. Obviously, you haven't done that yet, but you are far more interested in belittling others that don't fit some imagined scenario you have come up with. Either that, or your comprehension of English could use improvement.

Just because it seems that your comprehension is so limited, from the Rene Ritchie video itself (starting at 4:23), and as a direct quote:

"....the tasks that are required of a Mac are different enough that the silicon itself should still be different. One of the things we talked about is hypervisor acceleration for example, for running virtual machines on a Mac..." (ends at 4:34).

Please feel free to tell me how this is NOT in the Rene Ritchie video. Tell me that you have better insights into Apple Silicon and where it is going than Rene Ritchie does, and prove it. Tell us how you know that there won't be a hypervisor accelerator (core/block/module/chunk of silicon). If you can't, don't cast aspersions on accurate, trusted journalists with a long established track record, and other posters that don't fit your own narrow minded view of what will be happening, and how it will only be according to your concept.

This is not a Frankenstein scenario. Apple will be putting this into the SoCs primarily for security purposes (one of the main reasons for hypervisors in the first place) for MacOS, but there may be a way to use it to allow for Windows VMs as well, if Microsoft and Apple can work out the licensing details. And this is also why it is relevant to this discussion.

 

[Nugget]

So the part about hypervisor accelerator blocks, and how Apple will be putting them into silicon just flew over your head, did it? Again, it is in the Rene Ritchie video, and start listening at 4:23-4:34. Obviously, you haven't done that yet, but you are far more interested in belittling others that don't fit some imagined scenario you have come up with. Either that, or your comprehension of English could use improvement.

Just because it seems that your comprehension is so limited, from the Rene Ritchie video itself (starting at 4:23), and as a direct quote:

"....the tasks that are required of a Mac are different enough that the silicon itself should still be different. One of the things we talked about is hypervisor acceleration for example, for running virtual machines on a Mac..." (ends at 4:34).

Let's review. First of all, can we both agree that when you said "They called it a hypervisor core, and it is in both the WWDC presentation, and in the Rene Ritchie video" you were flat out incorrect? The words "hypervisor core" do not appear anywhere in the video, no matter how many times you refer to it. He does talk about the utility of virtualization on Macs which does not apply in the iOS/tablet world. But your whole scenario is undercut by a few inconvenient facts:

• virtualization doesn't use "a core" like you seem to think it does. It is an instruction set. The phrase "dedicated hypervisor core" is technical gibberish that you have invented. This is not something that either Apple or the YouTuber you mentioned have said.
• The developer kit using the old iPad processor supports virtualization already, no additional hardware is needed to support virtualization.
• Your fanciful notion that Apple will move macOS to no longer be a bare-metal operating system but will instead ship Apple Silicon Macs with a new Hypervisor OS and virtualize macOS on top of that is... interesting... but not in any way based on evidence or even sound technical analysis.

Bottom line...

It would be like a supervisory core, possibly used only at boot. I am not the only one talking about this, nor did i come up with it on my own

You are the only one talking about this and you did apparently come up with it on your own.

Please feel free to tell me how this is NOT in the Rene Ritchie video.

What you have described is definitely not in the Rene Ritchie video.

Tell me that you have better insights into Apple Silicon and where it is going than Rene Ritchie does, and prove it.

I don't disagree with anything he says in that video. At all. He just didn't say all the things that you have imagined.

This is not a Frankenstein scenario. Apple will be putting this into the SoCs primarily for security purposes (one of the main reasons for hypervisors in the first place) for MacOS, but there may be a way to use it to allow for Windows VMs as well, if Microsoft and Apple can work out the licensing details. And this is also why it is relevant to this discussion.

And, once again, none of what you are imagining has anything at all to do with x86/amd64 emulation on Apple Silicon Macs. Please tell me you understand at least this much. None of what you are imagining is relevant to a discussion about x86/amd64 emulation on Apple Silicon Macs.

Apple Silicon Macs can and will be able to virtualize Arm environments. They will do this with the existing virtualization ops in the Arm instruction set (helpfully described in post 99). We know this because this capability already exists in the developer kits that are in the hands of developers today and it is aligned with Apple's public statements on the subject. None of this involves a "hypervisor core" or "virtualization core" which are two phrases you have invented for yourself (and seem unable to describe, define, or even consistently reference). Your theories do not come from Apple or from any well-regarded YouTuber.

 

[The_Interloper]

Apple Silicon (AS) will not virtualise x86/64. It may, however, emulate it (slowly).

AS will virtualise ARM OS’s.

Windows on ARM (WoA) is currently only compatible with a specific Snapdragon processor. AS cannot virtualise it unless Microsoft makes a compatible AS version. Whether they do is anyone’s guess; I doubt it.

WoA can run x86 apps, but only 32-bit and in emulation. Performance is lousy. 64-bit compatibility is promised and likely to perform even worse.

TLDR: Apple Silicon will only run x86/64 apps virtualised if Microsoft makes an AS version of Windows on ARM and adds in 64-bit compatibility to the currently terrible 32-bit x86 support.

And that is that. ?

 

[Kostask]

Let's review. First of all, can we both agree that when you said "They called it a hypervisor core, and it is in both the WWDC presentation, and in the Rene Ritchie video" you were flat out incorrect? The words "hypervisor core" do not appear anywhere in the video, no matter how many times you refer to it. He does talk about the utility of virtualization on Macs which does not apply in the iOS/tablet world. But your whole scenario is undercut by a few inconvenient facts:

• virtualization doesn't use "a core" like you seem to think it does. It is an instruction set. The phrase "dedicated hypervisor core" is technical gibberish that you have invented. This is not something that either Apple or the YouTuber you mentioned have said.
• The developer kit using the old iPad processor supports virtualization already, no additional hardware is needed to support virtualization.
• Your fanciful notion that Apple will move macOS to no longer be a bare-metal operating system but will instead ship Apple Silicon Macs with a new Hypervisor OS and virtualize macOS on top of that is... interesting... but not in any way based on evidence or even sound technical analysis.

Bottom line...



You are the only one talking about this and you did apparently come up with it on your own.



What you have described is definitely not in the Rene Ritchie video.



I don't disagree with anything he says in that video. At all. He just didn't say all the things that you have imagined.



And, once again, none of what you are imagining has anything at all to do with x86/amd64 emulation on Apple Silicon Macs. Please tell me you understand at least this much. None of what you are imagining is relevant to a discussion about x86/amd64 emulation on Apple Silicon Macs.

Apple Silicon Macs can and will be able to virtualize Arm environments. They will do this with the existing virtualization ops in the Arm instruction set (helpfully described in post 99). We know this because this capability already exists in the developer kits that are in the hands of developers today and it is aligned with Apple's public statements on the subject. None of this involves a "hypervisor core" or "virtualization core" which are two phrases you have invented for yourself (and seem unable to describe, define, or even consistently reference). Your theories do not come from Apple or from any well-regarded YouTuber.

What part of hypervisor accelerator is nt immediately obvious? I did use some terms that deviated from what the direct quotes said, but that is a matter of semantics. My hypervisor core is the same as Rene Ritchie's hypervisor accelerator.

And for the record, I never said that the hypervisor core would, by itself alone, allow for virtualizing an X86/x64 VM, only that it could be used to enable an ARM Windows VM IF MICROSOFT AND APPLE CAME TO AN AGREEMENT. This gets into a bunch of stuff, which may or may not be under discussion between Apple and Microsoft, removing dependence on a specific ARM CPU, improving x86 and adding X64 support for ARM, and probably a lot of stuff I haven't considerd. As to whether this is under discussion between Apple and Microsoft, I don't know either way. There are many discussions underway between Apple and Microsoft, constantly, but nobody knows what those discussions are about or their scope.

Everybody here seems to think that emulation and translation are the only way to get Windows on an AS Mac. I am just pointing out that there may be another way. It does not directly address the emulation question asked, but it does present a possibility. I don't get hung up on whether it is true emulation or translation. Boot Camp is not emulation or translation either, but serves the purpose of allowing the operation of a lot of Windows software on the Intel Macs.

 

[Nugget]

What part of hypervisor accelerator is nt immediately obvious?

The part where you explain what on earth you imagine a hypervisor accelerator would be. Because that's not how virtualization works at all. But, really, you win. You've worn me down. I won't be replying again after this.

 

[MalcolmH]

That, nobody can really say. This is only part of a range of issues being discussed, and it is unknown what the exact range of issues are being discussed, and we will probably never know. This is an extension of the direction that Microsoft has been going towards (i.e. platform agnostic). While Microsoft has always had a presence in the Mac software market (mostly Office), they have branched out into some Linux and Android stuff as well. Some see it as a way for Micosoft to start breaking its dependence on the Windows PC world, and branch into mobile (they also make Android apps and casual games) and cloud based offerings..

I’m sure Windows 10 for Arm (not the IOT version) will work on an AS Mac. It can be installed on a Raspberry Pi today. As per the Pi, there may be some driver work to be done ...

How to install Win10 on a Pi

 

[MalcolmH]

The part where you explain what on earth you imagine a hypervisor accelerator would be. Because that's not how virtualization works at all. But, really, you win. You've worn me down. I won't be replying again after this.

You don’t need a new core to run a virtual machine. However the version of the A12Z in the DTK doesn’t support virtualization, why ? The sub-version of the Armv8 architecture used on the A12Z doesn’t include all the necessary instructions to run a hypervisor efficiently. I assume the A14Z (or whatever it’s called) will be of a more modern architecture that includes them and may even include Apple extensions in hardware (eg there’s something mentioned about Rosetta sorting out memory page ordering automatically. Is anyone understands the issue, please comment).

How virtualization works on Arm

Armv8 architecture extensions

 

[Maximara]

Apple Silicon (AS) will not virtualise x86/64. It may, however, emulate it (slowly).

Rossetta 2 doesn't emulate like that; it translates which is way faster. We saw how fast in the WWDC 2020 Special Event Keynote. with regards to Maya and Shadow of the Tomb Raider. We can literally see what the speed is.

 

[The_Interloper]

I was referring to x86/64 Windows. Those demos were for Mac applications and games, translated in Rosetta 2 as you correctly point out.

The barriers to getting an acceptably working x86 Windows installation in a virtual machine are many and complex, i.e. unlikely.

 

[tdar]

Apple Silicon (AS) will not virtualise x86/64. It may, however, emulate it (slowly).

AS will virtualise ARM OS’s.

Windows on ARM (WoA) is currently only compatible with a specific Snapdragon processor. AS cannot virtualise it unless Microsoft makes a compatible AS version. Whether they do is anyone’s guess; I doubt it.

WoA can run x86 apps, but only 32-bit and in emulation. Performance is lousy. 64-bit compatibility is promised and likely to perform even worse.

TLDR: Apple Silicon will only run x86/64 apps virtualised if Microsoft makes an AS version of Windows on ARM and adds in 64-bit compatibility to the currently terrible 32-bit x86 support.

And that is that. ?

Computers in use by users are a system, a combination of hardware and software.
I’ m a systems engineer ( and have been for a long time) that’s my specialty, marrying the right hardware with the right software to provide the needed performance of the system.
We do not know how WoA preforms on AS until we or Microsoft see/tells us how it does.
You mentioned a snapdragon processor, frankly it’s not very impressive. And likely that is the reason for the poor performance of WoA. Also it’s important to note that WoA presently runs all Microsoft store apps, and win 32 apps. Support for Win 64 is planned for 20H2, which interestingly will be coming out just around the same time as Apple Silicon.
TLDR: we don’t yet know how Windows on arm will run apps on Apple silicone

 

[The_Interloper]

TLDR: we don’t yet know how Windows on arm will run apps on Apple silicone

Exactly, that was my point. Unless Microsoft even releases a version of WoA for AS, we’ll never know.

 

[Maximara]

I was referring to x86/64 Windows. Those demos were for Mac applications and games, translated in Rosetta 2 as you correctly point out.

"Microsoft Arm-based PC run 64-bit (Arm64) apps, 32-bit (Arm32) apps, or 32-bit (x86) in emulation mode. Currently, 64-bit (x64) apps won’t work. Early 2021 is the scheduled date for an x86-64bit emulator." - Exiting x86: Why Apple and Microsoft are embracing the Arm-based PC

So right now not even Microsoft can run x64 code on ARM. There are rumors of Microsoft and Apple being in talks which makes sense. Apple has a reasonable x64 translator and Microsoft has a reasonable x86 translator so the partnering up makes sense. According to the rumors the issue is licensing not hardware or performance.

That said there is Wine Developers Release Hangover Alpha To Run Windows x86_64 Programs On 64-Bit ARM (17 February 2019) so there is something on the open source side of things. "Hangover makes use of Wine but also QEMU and other components. Hangover 0.4 Alpha is capable of running some Windows programs so far but is very much a work in progress." And Seven months ago there was Wine 5.0 Released - run some Windows programs on Android

So there are options as limited as they are.

 

[dmccloud]

Apple Silicon (AS) will not virtualise x86/64. It may, however, emulate it (slowly).

AS will virtualise ARM OS’s.

Windows on ARM (WoA) is currently only compatible with a specific Snapdragon processor. AS cannot virtualise it unless Microsoft makes a compatible AS version. Whether they do is anyone’s guess; I doubt it.

WoA can run x86 apps, but only 32-bit and in emulation. Performance is lousy. 64-bit compatibility is promised and likely to perform even worse.

TLDR: Apple Silicon will only run x86/64 apps virtualised if Microsoft makes an AS version of Windows on ARM and adds in 64-bit compatibility to the currently terrible 32-bit x86 support.

And that is that. ?

WoA can run on multiple processors, not just the SQ1 in the Surface Pro X. That's why Samsung and Lenovo both had ARM-based Windows machines (C630, Galaxy Book S) before the SPX was even announced. The trick is that the Microsoft licensing for WoA works differently than the traditional Windows licensing, and that is what has limited adoption of WoA, not processor restrictions.

 

[Maximara]

WoA can run on multiple processors, not just the SQ1 in the Surface Pro X. That's why Samsung and Lenovo both had ARM-based Windows machines (C630, Galaxy Book S) before the SPX was even announced. The trick is that the Microsoft licensing for WoA works differently than the traditional Windows licensing, and that is what has limited adoption of WoA, not processor restrictions.

That is the one thing I don't understand. If Microsoft wants PC makers to use Windows on ARM why do this funky thing with the licensing rather then how they handle Intel Windows? Better question why do they keep doing the Home vs Pro thing rather than going Apple's route and have one OS?

 

[dmccloud]

That is the one thing I don't understand. If Microsoft wants PC makers to use Windows on ARM why do this funky thing with the licensing rather then how they handle Intel Windows? Better question why do they keep doing the Home vs Pro thing rather than going Apple's route and have one OS?

Two big reasons for the licensing differences on Microsoft's end. The first is that they can get more money from businesses, so they want more money. The second reason is that there are numerous features in Windows 10 Pro that do not exist in Windows 10 Home. When I got my MSI laptop last March, the first thing I had to do was upgrade to Windows 10 Pro because I needed the virtual machine capabilities for school and my side job. With regard to WoA, Microsoft has essentially taken Apple's playbook for MacOS. By limiting the number of devices and manufacturers that currently have licenses for WoA, Microsoft only has to support a very limited number of device drivers. One big reason for the bloat of Windows 10 is the need to include so many different device drivers for the thousands and thousands of HDDs, optical drives, keyboards, mice, etc. that have been released over the years. By limiting who can build systems using WoA, that problem is eliminated. Given how many of these ARM-based Windows machines only have 128GB storage, the ability to streamline the OS footprint by only needing to support a handful of devices rather than thousands of them is a major plus.

 

[leman]

What part of hypervisor accelerator is nt immediately obvious? I did use some terms that deviated from what the direct quotes said, but that is a matter of semantics. My hypervisor core is the same as Rene Ritchie's hypervisor accelerator.

As @Nugget has been (very patiently) trying to explain, things like "hypervisor core" or "hypervisor accelerator" do not exist. The sources you are quoting are talking about "hypervisor acceleration". This term simply means that the CPU implements additional functionality necessarily to run VMs efficiently (this usually involves a third level of virtual memory lookup and some other low-level things). It is not a separate core, and it is not some specialized additional hardware unit — it involves an extensive modification of the CPU on a fundamental level.

And for the record, I never said that the hypervisor core would, by itself alone, allow for virtualizing an X86/x64 VM, only that it could be used to enable an ARM Windows VM IF MICROSOFT AND APPLE CAME TO AN AGREEMENT.

If Apple and Microsoft came to an agreement you could run Windows VMs on an iPhone.

Everybody here seems to think that emulation and translation are the only way to get Windows on an AS Mac. I am just pointing out that there may be another way.

Windows on ARM is currently not a full-featured product. If Microsoft ever makes it a full-fledged OS that is available for the users, I will definitely welcome Windows support on an Apple Silicon Mac. Leakers claim that this is coming anyway.

 

[dmccloud]

Windows on ARM is currently not a full-featured product. If Microsoft ever makes it a full-fledged OS that is available for the users, I will definitely welcome Windows support on an Apple Silicon Mac. Leakers claim that this is coming anyway.

Have you even used either the Surface Pro X, Lenovo C630, or Galaxy Book S? The OS is full-featured, and most software runs on it without any issues. There is some software that has not been ported yet, but that's on the individual developers, not on Microsoft.

 

[Maximara]

Two big reasons for the licensing differences on Microsoft's end. The first is that they can get more money from businesses, so they want more money. The second reason is that there are numerous features in Windows 10 Pro that do not exist in Windows 10 Home. When I got my MSI laptop last March, the first thing I had to do was upgrade to Windows 10 Pro because I needed the virtual machine capabilities for school and my side job. With regard to WoA, Microsoft has essentially taken Apple's playbook for MacOS. By limiting the number of devices and manufacturers that currently have licenses for WoA, Microsoft only has to support a very limited number of device drivers. One big reason for the bloat of Windows 10 is the need to include so many different device drivers for the thousands and thousands of HDDs, optical drives, keyboards, mice, etc. that have been released over the years. By limiting who can build systems using WoA, that problem is eliminated. Given how many of these ARM-based Windows machines only have 128GB storage, the ability to streamline the OS footprint by only needing to support a handful of devices rather than thousands of them is a major plus.

Ok the wants to make money part would make if not for key flaw (more on that later) as does the "need to include so many different device drivers for the thousands and thousands of HDDs, optical drives, keyboards, mice, etc. that have been released over the years."

In fact, this was one why Apple's attempt to have a version of MacOS that ran on Intel PCs in 1992 went nowhere. Code named Star Trek with the tag line "To boldly go where no Mac has gone before" (which was quickly mocked in Computerword as "the OS that boldly goes where everyone else has been") it only ran on a very narrow subset of PCs (sort like all the hardware hoops people have to go through to build a hackintosh today)

Another factor was John Sculley being replaced by Michael Spindler who instead of being the savior that everyone was expecting became 'The CEO that nearly killed Apple'. The weird thing here is Sculley lamented in 2003 "should instead have targeted the dominant Intel architecture" but the fact of the matter he had a side project that did exactly that...which Michael Spindler cancelled infavor of the whole clone thing which was one of the reasons the company was in such dire strait by the time somebody realized getting Jobs back was only way Apple would survive.

But unless Windows 10 Pro has a more limited hardware scope then Windows 10 Home Microsoft is effectively having two related builds of the OS which costs money to maintain in terms of programming, customer support, technical trouble shooting, etc. So any extra revenue from Windows 10 Pro could be eaten up on the back end resulting in a lost of money in the long run.

Now, up to a point such diversity makes sense but past that point you run into "The Many buckets issue". - "Picture it this way," Slavicsek says, "it's raining money outside and you want to catch as much of it as you can. You can either make a really big bucket or waste your time and attention by creating a lot really small buckets -- either way, you're never going to make more rain."

This is why the Mac Clone program failed - only so many people were going to buy a Mac so allowing clones just fragmented the market reducing the amount money Apple could make on the hardware side and the OS was so generously priced that they weren't making up that money with licensing the OS.

In fact, soon after Steve Jobs returned to Apple in 1997, he personally tried to renegotiate licensing deals more favorable to Apple five times and in his words each time was "basically told to pound sand". It was this response that caused him to halt negotiations of upcoming licensing deals with OS licensees that Apple executives complained were still financially unfavorable.

 

[dmccloud]

Ok the wants to make money part would make if not for key flaw (more on that later) as does the "need to include so many different device drivers for the thousands and thousands of HDDs, optical drives, keyboards, mice, etc. that have been released over the years."

In fact, this was one why Apple's attempt to have a version of MacOS that ran on Intel PCs in 1992 went nowhere. Code named Star Trek with the tag line "To boldly go where no Mac has gone before" (which was quickly mocked in Computerword as "the OS that boldly goes where everyone else has been") it only ran on a very narrow subset of PCs (sort like all the hardware hoops people have to go through to build a hackintosh today)

Another factor was John Sculley being replaced by Michael Spindler who instead of being the savior that everyone was expecting became 'The CEO that nearly killed Apple'. The weird thing here is Sculley lamented in 2003 "should instead have targeted the dominant Intel architecture" but the fact of the matter he had a side project that did exactly that...which Michael Spindler cancelled infavor of the whole clone thing which was one of the reasons the company was in such dire strait by the time somebody realized getting Jobs back was only way Apple would survive.

But unless Windows 10 Pro has a more limited hardware scope then Windows 10 Home Microsoft is effectively having two related builds of the OS which costs money to maintain in terms of programming, customer support, technical trouble shooting, etc. So any extra revenue from Windows 10 Pro could be eaten up on the back end resulting in a lost of money in the long run.

Now, up to a point such diversity makes sense but past that point you run into "The Many buckets issue". - "Picture it this way," Slavicsek says, "it's raining money outside and you want to catch as much of it as you can. You can either make a really big bucket or waste your time and attention by creating a lot really small buckets -- either way, you're never going to make more rain."

This is why the Mac Clone program failed - only so many people were going to buy a Mac so allowing clones just fragmented the market reducing the amount money Apple could make on the hardware side and the OS was so generously priced that they weren't making up that money with licensing the OS.

In fact, soon after Steve Jobs returned to Apple in 1997, he personally tried to renegotiate licensing deals more favorable to Apple five times and in his words each time was "basically told to pound sand". It was this response that caused him to halt negotiations of upcoming licensing deals with OS licensees that Apple executives complained were still financially unfavorable.

You do realize that Windows 10 Home is just 10 Pro with certain segments of the code removed, right? It's features like full BitLocker encryption, virtualization, and enterprise class features which no home user would need that are missing from Home. There is no significant additional burden on Microsoft by supporting two versions, especially when you consider that this has been Microsoft's model going back to the days of Windows and Windows NT. If Microsoft is making $20 in profit off each copy of Windows 10 Home, they're clearing at least $50 from Pro, most likely more since they just develop for Pro then take out the pro-specific features. I would also point out that the Mac clone program isn't even in the same league as device driver support

 

[tdar]

You do realize that Windows 10 Home is just 10 Pro with certain segments of the code removed, right? It's features like full BitLocker encryption, virtualization, and enterprise class features which no home user would need that are missing from Home. There is no significant additional burden on Microsoft by supporting two versions, especially when you consider that this has been Microsoft's model going back to the days of Windows and Windows NT. If Microsoft is making $20 in profit off each copy of Windows 10 Home, they're clearing at least $50 from Pro, most likely more since they just develop for Pro then take out the pro-specific features. I would also point out that the Mac clone program isn't even in the same league as device driver support

It’s simpler than that. Windows 10 home is just Windows 10 pro with some of the features turned off. They only develop one version of Windows 10 and that’s Pro.That’s why you can take a Home system, pay your upgrade fee and it turns into Pro.
Yes for those wondering I do know about Enterprise but it’s just Pro with different licensing.

 

[theluggage]

In fact, this was one why Apple's attempt to have a version of MacOS that ran on Intel PCs in 1992 went nowhere.

...no, it went nowhere because Windows had a near-monopoly on personal computing and the majority pf people wanted/needed to run Windows software. Windows dominance was almost total in the early 90s.

Because of that monopoly, Microsoft didn't have to bother about writing and maintaining "hundreds of drivers" - if anybody wanted to sell components or peripherals then they had to make their products work with Windows. If, however, you want to sell MacOS, BeOS, Linux, RISC-OS, AmigaOS as an alternative OS, then you have to make those drivers happen - either by writing it yourselves or making it worth the peripheral makers while.

Even where MS did bundle drivers into the OS - a lot of those were minimal and the first thing you'd do after installing generic Windows on a PC would be to install all of the component/peripheral manufacturer's own drivers to get the full performance from your hardware. Your graphics, for instance, would be 640x480 16 colour until you stuck in the CD from ATI/Matrox/Whoever and installed the third-party drivers.

E.g. A Windows PC could use any number of Windows-only printers that were just bare-bones boxes for putting dots on paper, with totally proprietary protocols that could change drastically from model-to-model. Implementing HPGL or Postscript in the printer was expensive - why would manufacturers bother when they could make a dumb printer totally reliant on Windows drivers and satisfy 95% of their customers? Why standardise the interface across models when you can just ship different drivers? While the same approach could work with Mac, the typical situation was that you'd have to pay a premium for a PostScript or HPGL printer.

Then, apart from the drivers issue, there's the problem that MS, then, were an abusive monopoly and used such lovely tricks as charging OEMs licensing fees based on their total turnover rather than the actual number of copies of Windows they shipped: Result: if a customer wanted a PC with BeOS or NextStep or Linux then the cost of the Windows license was still part of the selling price. When 95% of your customers wouldn't buy a PC without Windows you're not going to argue with Microsoft over licensing terms or, potentially, upset them by selling a competitors OS.

Even without such shenanigans - 95% of PCs came with Windows (not to mention Office), so MS could charge nice low prices per copy and still make money hand over fist. Even if maintaining MacOS - for whatever platform - only cost 50% as much as maintaining Windows if it only had 5% of the market there is no way that Apple could have supported that development by selling PCs at competitive prices - where "competitive" at the time meant "if you can persuade the punter to buy an extended warranty to cover the PC for the period when it is least likely to go wrong, charge them $50 for a $3 video cable and cram as much adware as you can onto the hard drive then you'll turn a profit".

Make no mistake - the only way Sculley or Spindler could have "targetted the dominant Intel architecture" at that time would have been to start making Apple-branded PC clones running Windows and drop MacOS. The climate in the early 90s was totally toxic to anything that wasn't a Wintel PC and killed off ever other non-Wintel platform. Heck, even IBM failed with the PS/2 and were almost killed by their own creation.

...and while it's nice to think that it was Jobs that turned things around for Mac, he was riding on the back of two hugely disruptive technologies: mass adoption of internet, and then the rise of mobile technology - which MS were slow to react to and which undermined their monopoly. By 2005 when Apple did switch (...by making Intel hardware, note, not simply releasing MacOS for Intel) things had changed massively and Apple had already turned the corner. Wintel is still dominant today, but mobile and internet technology have made the tech world massively more diverse that it was in the 90s. The only reason Jobs succeeded was that he managed to ride those waves.

So any extra revenue from Windows 10 Pro could be eaten up on the back end resulting in a lost of money in the long run.

No - the business plan is simple:
1. Write/Maintain Windows Professional - you have to do that anyway
2. Turn off key Pro features needed by Pro users - minimal extra work required.
3. Sell vast quantities cheap OEM licenses for Windows Home to keep sticker prices low and make sure every PC comes with Windows
4. Charge $$$ for upgrades for Pro - pure cash-y money.
5. Profit.

Apple can't/doesn't do step 3 - but can use the fact that every Mac comes with the full OS to help justify the premium cost of Macs. In any case, the Pro/Enterprise versions of Windows include some features that you'd have to buy as part of MacOS Server or third-party products to get the equivalent on Mac. E.g. the pro versions of Windows come with a Hypervisor app (MacOS has a Hypervisor framework but no front-end) and the Windows Subsystem for Linux - for which you need third party software on MacOS.

So right now not even Microsoft can run x64 code on ARM.

Nitpick: not even Microsoft can run 64-bit Windows Apps on ARM. (to be fair, that was probably what you meant)
Sounds like Rosetta2 can quite happily cope with x64 code. Maybe if Microsoft ask Apple very, very nicely they'll license the tech to them...

Also, all Apple have actually said is that Rosetta2 won't translate your x86 copy of Parallels/VMWare/VirtualBox... not that the tech can't be used as part of some future solution for running x86 Windows.

WoA can run on multiple processors, not just the SQ1 in the Surface Pro X.

See for example:

I Ran Windows 10 on Raspberry Pi So You Don't Have To

Using a set of hacks, you can run full Windows 10 on a Raspberry Pi board, but the user experience leaves a lot to be desired.

www.tomshardware.com

TLNR: even 18 months ago it was possible to run Windows 10 for ARM64 on a Raspberry Pi 3B by obtaining the image from an unofficial source. It sucked big time - but then it was running on a $30 maker board using an industry surplus set top box SoC (not even the latest Pi 4 which is substantially more powerful) with no graphics acceleration and a SD card as the only mass storage, so what do you expect? If a tech journalist can get it running in a couple of days, with no official support, on a feeble Pi then it's not going to take a Manhattan Project for someone with resources to get it going in a Hypervisor on Apple Silicon. It will need someone to write paravirtualised drivers for graphics etc. (or for the hypervisor to emulate something it can drive) but the same goes for any guest OS. Basically, it will happen unless (a) Microsoft refuses to allow it or ((b) nobody actually wants it.