MP All Models My AS Mac Pro Prediction

[Adult80HD]

OK, it's really just a SWAG--super wild-a$$ guess--but they will also announce a GPU with ray-tracing capabilities at the same time. It will perform better than skeptics believe, but as always not quite as good as Apple's graphs will show. But it will be *quite* good for a first release.

 

[Boil]

Take a look at this...

ASi Mac Pro may debut with M2 Ultra (Extreme) SoCs, with AMD MI200-series derived PX-format accelerator options; ASi hardware ray-tracing (most likely derived from this licensing) will come along with the M3-family of SoCs...

 

[th0masp]

Why would they care about raytracing specifically? Traditionally the Mac has largely been a no-show in 3D, has it not? And the applications you get are the afterthought-ports of Win/Linux software, nothing that's targeted at the platform unless I'm missing something here.
Only times I ever saw them in the office was sometimes one on the owner's desk and usually a bunch in the audio department. Tiny numbers next to the PCs.

 

[Adult80HD]

Why would they care about raytracing specifically? Traditionally the Mac has largely been a no-show in 3D, has it not? And the applications you get are the afterthought-ports of Win/Linux software, nothing that's targeted at the platform unless I'm missing something here.
Only times I ever saw them in the office was sometimes one on the owner's desk and usually a bunch in the audio department. Tiny numbers next to the PCs.

The MacPro space has always had 3D in mind, via 3rd-party GPUs. It's pretty well-known that ray-tracing is coming on the M3, and Apple is a very big player in the mobile gaming space. Something in the Mac Pro for the shops that use and for developers building games on the mobile platform makes sense to me.

I also think that they will skip right to an M3 version the Mac Pro and it will be available later this summer or fall. An M2 now would be underwhelming and I don't see them doing underwhelming. All speculation of course!

 

[rpmurray]

It was six years between the 2013 and the 2019, so I'm guessing we won't see a new Mac Pro until at least 2025. They released the Mac Studio which is a stopgap, much like the iMac Pro was, until they can get their sh*t in one bag.

 

[th0masp]

The MacPro space has always had 3D in mind, via 3rd-party GPUs. It's pretty well-known that ray-tracing is coming on the M3, and Apple is a very big player in the mobile gaming space. Something in the Mac Pro for the shops that use and for developers building games on the mobile platform makes sense to me.

Sounds like you still see Apple as a computer manufacturer like it was before the iPhone, competing with the PC world and all. I think they moved on from focusing on that quite some time ago. Still happy to sell you a laptop, though.

The GPU's in the last Mac Pro seemed intended for GPU-compute primarily to accelerate video editing. I think it's likely they will try to come out with a new piece of equipment to continue in this fashion. That's also where they do have software support.

But raytracing, specifically - and Apple's own flavour at that? Who's going to support this software-side - and keep supporting it through the years?

And nah, I don't believe developers of mobile games will be putting 5-digits worth of workstation under their desks. I imagine they will develop on whatever and if they need to test anything gadget-specific that is not available on desktop they'll be firing up a simulator in the cloud.
Last time I was in this racket, the test station for iOS was a Mac Mini. For building the code, not for actual development. That was all PC-based.

 

[deconstruct60]

Why would they care about raytracing specifically? Traditionally the Mac has largely been a no-show in 3D, has it not? And the applications you get are the afterthought-ports of Win/Linux software, nothing that's targeted at the platform unless I'm missing something here.

Same sermon they preach at every M-series introduction so far.... Performance/Watt. Hardware raytracing would run the RT code that the modern Apple libraries already provision for at lower wattage consumption. Hence, going to get better Perf/Watt.

Folks get lost in the sauce when HW RT is all about racing to be 'King Kong' of ultimate 8K 3D rendering. That isn't necessarily the whole scope. HT RT would help with their AR headsets. It isn't about rending the most complex RT screen possible. It would be about doing some RT to help better match reality ( which is inherently RT for 'free') at lower battery consumption. (the battery energy being the scarce resource )

Some of the patents that have surfaced from Apple about RT hardware has been about aborting unnecessary rays early in the process. That is more about skipping long term useless computation far, far , far more than brute forcing them to be completed quicker ( to no impact that ends up on screen ). If it is pointless , faster (and/or higher resolutions) isn't necessarily better.


Very similar with how the ProRes hardware accelerator doesn't bring new functionality. ( if there is no Afterburner or Mn ProRes accelerator hardware present the CPU core library version is auto selected); speed is one side effect but it is also far more better energy consumption metrics. ( ProRes support went from a wide spread beta-test in Afterburner to entire platform wide in 3 years. Same attention to scope likely here. )

Apple will probably limit the HW RT to something that gets the best bang-for-area consumed, TSMC N3 probably gives them a bigger transistor budget to allocate a relatively modest transistor budget for it. Highly doubtful this will be the biggest ( in area consumption) , most aggressive RT hardware possible.

Only times I ever saw them in the office was sometimes one on the owner's desk and usually a bunch in the audio department. Tiny numbers next to the PCs.

Apple isn't going to be about delivering HW RT to the most exclusive audience possible. Their HW RT is likely going to get distributed to 100's of millions of users which far, far , far outnumber the folks you are pointing at. It will be trotted out to both the A and M series SoCs. The Mac Pro will get some 'trickle down' from that; not vice versa. Apple probably is not building some 4090 (or 3090 or 5090) 'killer' to suck back the folks who left solely just to get those.

Won't be surprising if Apple has augments for better upscaling also if that gets them a better Pref/Watt bump. Same foundation on that sermon. ( skip computations that are actually necessary to get good results ).

 

[Adult80HD]

Same sermon they preach at every M-series introduction so far.... Performance/Watt. Hardware raytracing would run the RT code that the modern Apple libraries already provision for at lower wattage consumption. Hence, going to get better Perf/Watt.

Folks get lost in the sauce when HW RT is all about racing to be 'King Kong' of ultimate 8K 3D rendering. That isn't necessarily the whole scope. HT RT would help with their AR headsets. It isn't about rending the most complex RT screen possible. It would be about doing some RT to help better match reality ( which is inherently RT for 'free') at lower battery consumption. (the battery energy being the scarce resource )

Some of the patents that have surfaced from Apple about RT hardware has been about aborting unnecessary rays early in the process. That is more about skipping long term useless computation far, far , far more than brute forcing them to be completed quicker ( to no impact that ends up on screen ). If it is pointless , faster (and/or higher resolutions) isn't necessarily better.


Very similar with how the ProRes hardware accelerator doesn't bring new functionality. ( if there is no Afterburner or Mn ProRes accelerator hardware present the CPU core library version is auto selected); speed is one side effect but it is also far more better energy consumption metrics. ( ProRes support went from a wide spread beta-test in Afterburner to entire platform wide in 3 years. Same attention to scope likely here. )

Apple will probably limit the HW RT to something that gets the best bang-for-area consumed, TSMC N3 probably gives them a bigger transistor budget to allocate a relatively modest transistor budget for it. Highly doubtful this will be the biggest ( in area consumption) , most aggressive RT hardware possible.




Apple isn't going to be about delivering HW RT to the most exclusive audience possible. Their HW RT is likely going to get distributed to 100's of millions of users which far, far , far outnumber the folks you are pointing at. It will be trotted out to both the A and M series SoCs. The Mac Pro will get some 'trickle down' from that; not vice versa. Apple probably is not building some 4090 (or 3090 or 5090) 'killer' to suck back the folks who left solely just to get those.

Won't be surprising if Apple has augments for better upscaling also if that gets them a better Pref/Watt bump. Same foundation on that sermon. ( skip computations that are actually necessary to get good results ).

Thanks for taking the time to write this, it saved me a similar effort. FWIW, in my mind I see them creating a GPU for an MPX slot, similar to what they did with the Afterburner card. That's assuming the Pro will support add-on cards, but I feel like that's almost a requirement at this point to call anything a replacement for the existing Mac Pro.

 

[deconstruct60]

It was six years between the 2013 and the 2019, so I'm guessing we won't see a new Mac Pro until at least 2025.

Probably not. People are impatient but likely will have something before 2024 arrives. Apple has already blown past their 'about 2 year deadline' . Pretty doubtful they will drift longer than 2023. If only to start pushing the Intel MP toward the vintage/obsolete countdown clock. ( The MP is likely on some every even or even odd path. So there is either an M2 or M3 sequence to jump onto this year. )

They released the Mac Studio which is a stopgap,

Probably not. Apple explicitly said they were replacing the iMac 5K with the Mac Studio. And that Mac Pro was later. ( i.e., do not want to talk about it , but Studio was not a Mac Pro replacement. )

much like the iMac Pro was,

And yet Apple continued to sell the Mac Pro 2013 for another two years. Not much of replacement in the Apple line up if they keep selling it.

There was some heavy overlap between the MP 2013 and iMac Pro , but the iMP was also explicitly called out as not being the MP replacement over the long term.

If the iMP comes back it will get squeezed by the Mac Studio (probably not going away) and the Mac Pro. If Apple brings out another Display docking station that is placed between Studio Display and XDR then the iMP probably is going to have a bucketload of problems coming back.

The iMP got seriously messed up by Intel's stumbling with non delivering a viable upgrade for W-2100 series ( the W-2200 was the same die just priced and binned different) and AMD's slightly less bump, but delayed, GPU roll outs. (and 'Ice Lake Xeon' being a bit of a too hot and no single thread bump , very late disappointment). Throw on top growing 'right to repair' regulations and inflection point on display panel resolutions/lighting and the iMac form factor has problems the Mac Studio/Display avoids. The 'cherry on top' was the iMac design drifting over to an 'iPad thin on a stick' focus.

For a long time, Apple constructed fratricide barriers around the iMac 5K to keep the panel sales higher and more steady. Apple seems to have much lower fratricidal protections around the Mac products now. The iMac 24" , Mini , MBA , and MBP 13" all share the exact same SoC. Rumors are that a MBA 15" is coming to make splitting that 'shared SoC' pie even more competitive.

Until there is a next-gen panel that Apple wants to artificially herd people into, the big screen iMac has major problems. That is not a 'stop gap' thing, but more 'the market changed' thing. I don't think Apple's microLED plans turned out like they though it would. The double-layer OLEDs may not be a high volume , controlled cost option for Apple either at the 27-32" level.

And the competition with 3rd party monitors is up also, which makes the iMac a harder sale.

until they can get their sh*t in one bag.

Probably, not 'one bag'. If the RAM DIMMs disappear Apple can point at screen modularity. [ If Apple got lots of grumpbles about the iMac Pro lack of easy RAM upgrades , then taking away the screen modularity would be a problem , not a feature to run to. ] Vert similar reason why it would be bozo move to remove all the PCI-e slots from the next Mac Pro. Apple is only going to loose hypermodularity focused folks on the soldered RAM. Toss the slots and it is just a bigger group walking away.

 

[deconstruct60]

Thanks for taking the time to write this, it saved me a similar effort. FWIW, in my mind I see them creating a GPU for an MPX slot, similar to what they did with the Afterburner card.

I don't. The Afterburner card is likely a dead end as far is 'discrete card' goes. It was only done discretely to make sure there were no bugs before getting weaved into the integrated SoC dies. HW RT really can't be put on a separate die from the rest of the GPU cores and L1/L2 caches. That is why it is taking 'so long' to arrive.

Now that PreRes acceleration is on the main die ... extremely likely there are not going to be any Afterburner 2.0 or major firmware updates to Afterburner. It was a wide-distribution protype and it did its job well. That job is over.

Similar on Metal modifications for Infinity Fabric for MPX modules. If Apple needs that for a very large SOC with internal NUMA issues they might use it, but probably heading to be used on single package systems in the future ( either Apple or AMD ones).

The MPX connector mainly solves Thunderbolt issues ( the CPU lacking a GPU). That 'problem' doesn't really exist on any Apple SoC.

That's assuming the Pro will support add-on cards, but I feel like that's almost a requirement at this point to call anything a replacement for the existing Mac Pro.

Add-on compute cards? Perhaps. Add on display GPU cards. I'm pretty doubtful. Apple's asking folks to substantively disrupt their apps for uniformly , unified memory semantics. I really doubt they are going to now 2 years in going to turn around and say 'oh never mind all the work you did the last 2 year... go change it again. ". Not for the relatively super small demographics of Mac Pro edge case user base.

 

[Boil]

I still want the first generation ASi Mac Pro to debut with M3 Ultra & M3 Extreme SoC options, on a daughtercard would be best for (limited) future upgradability...?

I also would like to see ASi (GP)GPU solutions rather than any third-party options, it seems like Apple would be reversing course after telling devs to optimize for pure ASi GPU cores for the past two+ years...

I expect we will all find out more come WWDC at the latest...?

 

[deconstruct60]

I still want the first generation ASi Mac Pro to debut with M3 Ultra & M3 Extreme SoC options, on a daughtercard would be best for (limited) future upgradability...?

Daughtercards are not likely to get much future generation upgradablity. May be able to buy used parts in the future that are different daughtercard, but not likely to be able to get a M(n+1 or n+2 ) tray out of it.

Trying to subsume both what would have been "Northbridge" and "southbridge" chips onto the daughtercard isn't going to get deeper entanglements into the present that inhibit the future connectivity.



. But yeah a M2 Ultra/Extreme really don't make much economic sense at all, unless going to skip M3 generation and would have to squat on M2 until M4. The big chips are likely either on the 'odds' or 'evens' update sequence. M3,M5,M7 or M2,M4,M6 . A short cadence probably wouldn't work. The Mac Pro's don't upgrade that fast (for the Extreme) model. There is no product to 'hand me down" the Ultra/Extreme too once replacement them ( so even if churned the Mac Pro faster (i.e., every 12-16 months) they would have bigger problems recouping the investment).
The last three update cycles have been 3 years , 6 years , and at least 3 years. Every 1.5 (or less) years probably isn't happening. R&D costs too high and the run-rate of units far, far, far too low to just throw these big SoCs out the window at a relatively rapid pace. To throw them away that fast, Apple would have to greatly increase the price and that would put the product just even deeper into a pricing death spiral.

Pretty good chance Apple's plan was either start the Mac Pro out on M1 (if everything went perfectly ... which by time pandemic started that was at deep threat ) or just wait until M3 to get started. M2 makes making an Ultra a bigger pain. The M2 chips are bigger and the M1 Max was about at the reticle limit for making a Ultra on InFo-LSI. It may be that they were going to switch to CoWos-LSI anyway for the Ultra. The Extreme would have to , since "two chips" was already at the limit. Even with N3 couldn't shrink that back down into InFo-LSI limits. And splitting the package tech R&D testing doesn't add much. N3 would help though in making both packages more manageable.

I also would like to see ASi (GP)GPU solutions rather than any third-party options, it seems like Apple would be reversing course after telling devs to optimize for pure ASi GPU cores for the past two+ years...

A ASi module with a huge NUMA overhead and/or non Unified memory would still be reversing course software wise. Wouldn't make a different if it has Apple sprayed on it or AMD , it is still a app software hiccup. And relatively tiny niche too.

A "mac on a card" would be far less of a break because there are some apps that distribute workload out to a cluster now. A "mac on a card" would be faster, simpler networking ( no wires , no switches , just plug it in and add a virtual Ethernet over PCI-e driver on both sides). No reversals in GPU driver semantics. Portable 'distribute to cluster" code that works on any Mac ( so relatively vastly bigger market for the software vendors doing the work. ) .

I expect we will all find out more come WWDC at the latest...?

The last three WWDCs did absolutely nothing for plug-in GPUs. I wouldn't pins lots of hopes to WWDC. If Apple was going to release Mac Pro in June-October time frame then a limited , hands-off-just-look sneak peak could come this Spring before WWDC comes. There is about zero upside in waiting until WWDC to do the reveal (presuming have the parts firmly contracted for and no huge showstopper bugs in macOS extensions for it ). That just reinforces that they are about over a year late delivering the Mac Pro. Squatting on it for an extra 2-3 months buys nothing. And if isn't ready to go for another > 6 months after WWDC there isn't much to say. There is a magical Goldilocks spot were the parts/software all perfectly align with WWDC, but that isn't very likely.

If Apple waits until WWDC to say anything , then there is probably something seriously broken and/or way off in the weed wrong with the Mac Pro. Mostly likely Apple was trying to target something in 2022 (perhaps late 2022) for the Mac Pro. It is a so discombobulated that the 'sneak peak' had to slide far more than 6 months ... it is seriously 'broke' somehow. And the let it get that deeply into a screwed up state is a bad sign for the long term.

 

[Boil]

I still want the first generation ASi Mac Pro to debut with M3 Ultra & M3 Extreme SoC options, on a daughtercard would be best for (limited) future upgradability...?

Daughtercards are not likely to get much future generation upgradablity. May be able to buy used parts in the future that are different daughtercard, but not likely to be able to get a M(n+1 or n+2 ) tray out of it.

Trying to subsume both what would have been "Northbridge" and "southbridge" chips onto the daughtercard isn't going to get deeper entanglements into the present that inhibit the future connectivity.

But yeah a M2 Ultra/Extreme really don't make much economic sense at all, unless going to skip M3 generation and would have to squat on M2 until M4. The big chips are likely either on the 'odds' or 'evens' update sequence. M3,M5,M7 or M2,M4,M6 . A short cadence probably wouldn't work. The Mac Pro's don't upgrade that fast (for the Extreme) model. There is no product to 'hand me down" the Ultra/Extreme too once replacement them ( so even if churned the Mac Pro faster (i.e., every 12-16 months) they would have bigger problems recouping the investment).
The last three update cycles have been 3 years , 6 years , and at least 3 years. Every 1.5 (or less) years probably isn't happening. R&D costs too high and the run-rate of units far, far, far too low to just throw these big SoCs out the window at a relatively rapid pace. To throw them away that fast, Apple would have to greatly increase the price and that would put the product just even deeper into a pricing death spiral.

Pretty good chance Apple's plan was either start the Mac Pro out on M1 (if everything went perfectly ... which by time pandemic started that was at deep threat ) or just wait until M3 to get started. M2 makes making an Ultra a bigger pain. The M2 chips are bigger and the M1 Max was about at the reticle limit for making a Ultra on InFo-LSI. It may be that they were going to switch to CoWos-LSI anyway for the Ultra. The Extreme would have to , since "two chips" was already at the limit. Even with N3 couldn't shrink that back down into InFo-LSI limits. And splitting the package tech R&D testing doesn't add much. N3 would help though in making both packages more manageable.

Maybe SoC on daughtercard is there for incremental upgrades then; need more CPU, GPU, or RAM, purchase new daughtercard from Apple (with a horrible "trade-in" rate from Apple for the old daughtercard) and swap it in...?

I also would like to see ASi (GP)GPU solutions rather than any third-party options, it seems like Apple would be reversing course after telling devs to optimize for pure ASi GPU cores for the past two+ years...

A ASi module with a huge NUMA overhead and/or non Unified memory would still be reversing course software wise. Wouldn't make a different if it has Apple sprayed on it or AMD , it is still a app software hiccup. And relatively tiny niche too.

A "mac on a card" would be far less of a break because there are some apps that distribute workload out to a cluster now. A "mac on a card" would be faster, simpler networking ( no wires , no switches , just plug it in and add a virtual Ethernet over PCI-e driver on both sides). No reversals in GPU driver semantics. Portable 'distribute to cluster" code that works on any Mac ( so relatively vastly bigger market for the software vendors doing the work. ) .

If the ASi (GP)GPU is used for offloading compute/render tasks from the system SoC/iGPU, then there shouldn't be any NUMA issues...?

SoC iGPU used for display output, ASi (GP)GPU used for compute/render jobs in the background...?

I expect we will all find out more come WWDC at the latest...?

The last three WWDCs did absolutely nothing for plug-in GPUs. I wouldn't pins lots of hopes to WWDC. If Apple was going to release Mac Pro in June-October time frame then a limited , hands-off-just-look sneak peak could come this Spring before WWDC comes. There is about zero upside in waiting until WWDC to do the reveal (presuming have the parts firmly contracted for and no huge showstopper bugs in macOS extensions for it ). That just reinforces that they are about over a year late delivering the Mac Pro. Squatting on it for an extra 2-3 months buys nothing. And if isn't ready to go for another > 6 months after WWDC there isn't much to say. There is a magical Goldilocks spot were the parts/software all perfectly align with WWDC, but that isn't very likely.

If Apple waits until WWDC to say anything , then there is probably something seriously broken and/or way off in the weed wrong with the Mac Pro. Mostly likely Apple was trying to target something in 2022 (perhaps late 2022) for the Mac Pro. It is a so discombobulated that the 'sneak peak' had to slide far more than 6 months ... it is seriously 'broke' somehow. And the let it get that deeply into a screwed up state is a bad sign for the long term.

By "WWDC at the latest" I was referring to the ASi Mac Pro as a whole; I would think Apple wants to get the ASi Mac Pro out ASAP, to complete the transition to Apple silicon and all that...

 

[enc0re]

The only way* waiting until WWDC makes sense is if the Mac Pro somehow ties into the XR story. Who knows, maybe the Mac Pro delays were really about the XR delays.

*unless Apple is having actual engineering problems getting Mac Pro done. But that makes no sense, unless there will be something truly novel to the box. Which, well, let's hope there isn't.

 

[Thessman]

Remembering past MacPro showings there was always a special chip, be it from Motorola or intel or whatnot, so I guess the ASMac Pro will come with its own very special chip, probably based on M3.
Let's see!

 

[majus]

Remembering past MacPro showings there was always a special chip, be it from Motorola or intel or whatnot, so I guess the ASMac Pro will come with its own very special chip, probably based on M3.
Let's see!

M3 Ultra for entry level; M3 Extreme for the high-end?

 

[Thessman]

M3 Ultra for entry level; M3 Extreme for the high-end?

M3 you name it with daughter card addons?

 

[Mac Hammer Fan]

An AS Mac Pro with M3 would be significantly more attractive than an M2. But I was told the new Mac Pro would be released in march, so in this case it will be M2.

 

[ian87w]

I think one has to see who's buying the Mac Pros to begin with. Prior to Apple Silicon, seems like the typical Mac Pro buyers were graphic/video editors, relying on GPU for performance acceleration due to the pathetic IGP of intel. Those things have been solved with Apple Silicon, with the Pro/Max/Ultra versions have specific accelerators exactly for that purpose. Thus I feel majority of people that were used to the Mac Pro would actually be served well with the Mac Studio. Performance without the bulk. These are workstation class already.

The remainders seem to be those who require a huge amount of RAM as the Mac Studio is limited to max of 128GB. The current intel Mac Pro can go up to 768GB of RAM. That's pretty much the only thing the Mac Studios couldn't offer. I don't know how Apple is going to do it on Apple Silicon, but then again, I have a feeling Apple is debating internally if they should even bother with those ultra niche markets.

I don't think plain GPU horsepower is the focus anymore with the availability of ML and accelerators. I could be wrong though. Only time will tell what Apple will do.

 

[avro707]

Correction: 1.5TB maximum ram. This forum reckons 2.0TB is possible with the W3275M, which I’ll eventually put in mine to replace the W3245.

Also can do native Windows operation.

One computer does both operating systems, negating the need to have a separate PC.

I’ll be off now, I’m one of those Intel Mac Pro traitors.

 

[deconstruct60]

An AS Mac Pro with M3 would be significantly more attractive than an M2. But I was told the new Mac Pro would be released in march, so in this case it will be M2.

Release with no "sneak peak" ? that would be odd. The more this is a 'check the last box on the transition and rapidly run off to other products" ... the more disappointing it seems likely to be.

If on M2 and have to wait until M4 generation to get an update ... even more likely.

 

[ZombiePhysicist]

An AS Mac Pro with M3 would be significantly more attractive than an M2. But I was told the new Mac Pro would be released in march, so in this case it will be M2.

If apple does this it will be a testament to how incapable and how much they suck. The m3 is just a couple of months out beyond yet they want to saddle the Mac Pro with this pathetic gimp chip. Hope I’m wrong but this is smelling more and more like a doa fiasco by the day (If the rumors are to be believed).

 

[Boil]

Really hoping Apple will do the right thing and debut the ASi Mac Pro with 3nm...

Would make sense, nobody was really wanting the N3B allotment, Intel cancelled their order & Apple picked it up; so why not use those wafers, which have been in production for over two months now, and use them for the ASi Mac Pro...?

 

[th0masp]

Correction: 1.5TB maximum ram. This forum reckons 2.0TB is possible with the W3275M, which I’ll eventually put in mine to replace the W3245.

Also can do native Windows operation.

One computer does both operating systems, negating the need to have a separate PC.

I’ll be off now, I’m one of those Intel Mac Pro traitors.

Perhaps because its not all that big a selling point? Rebooting to switch between different OSes feels so ... 1990s to me. As would be having to carefully select hardware so its usable by or at least tolerated and trouble-free within both systems.
I'd rather have a proper sidekick machine with the other operating system on the desk. Could be a laptop or a mini-PC if performance is not the main priority and the expectation is that it won't get used all that often.

Spoiler

That being said I dislike the whole Apple Silicon move, very much dislike proprietary, soldered-on SSDs and non-upgradeable RAM as well as Apple's move away from standards of the PC world while pushing their own stuff like Metal instead.
They had something really cool there for a while: almost as open as a PC, comparable enough in the performance department, far more pleasant OS, big boost in the availability of software compared to the old days. Bliss. Seems like the good times lasted about from 2006-2013 with another blip on the radar in 2019. Hindsight and all.

 

[ian87w]

Perhaps because its not all that big a selling point? Rebooting to switch between different OSes feels so ... 1990s to me. As would be having to carefully select hardware so its usable by or at least tolerated and trouble-free within both systems.
I'd rather have a proper sidekick machine with the other operating system on the desk. Could be a laptop or a mini-PC if performance is not the main priority and the expectation is that it won't get used all that often.

Spoiler

That being said I dislike the whole Apple Silicon move, very much dislike proprietary, soldered-on SSDs and non-upgradeable RAM as well as Apple's move away from standards of the PC world while pushing their own stuff like Metal instead.
They had something really cool there for a while: almost as open as a PC, comparable enough in the performance department, far more pleasant OS, big boost in the availability of software compared to the old days. Bliss. Seems like the good times lasted about from 2006-2013 with another blip on the radar in 2019. Hindsight and all.

The "future" would be using cloud computing for those instances when you need a separate platform temporarily. Microsoft is already offering Windows 365, streaming a "rented" Windows session.