How long before Mac OS is Apple Silicon only?

[theorist9]

But you're not writing new firmware - you're maintaining existing firmware that was written years ago.

Maybe this is an issue of semantics. I do know that, when you update to a new OS, both Intel and AS Macs get firmware updates. That does mean the firmware had to be changed. So perhaps the question is whether it's more reasonable to call this "new firmware", or "firmware maintenance". I think it comes down to just how much the firmware needs to be changed. From https://eclecticlight.co/2022/10/06/firmware-and-macos-upgrades/ :

"Firmware and macOS upgrades
One of the side-effects of upgrading your Mac to a new version of macOS such as Ventura is that it will almost inevitably have its firmware updated."

In addition, many security patches require firmware updates. Certainly that was the case when companies addressed Spectre and Meltdown. I suspect addressing those, and other speculative execution exploits, required entirely new firmware, that these firmware updates were challenging to write (the challenge was in part closing these holes without giving up too much of the performance that speculative execution provides), and that this firmware is highly architecture-dependent.

Now you may want to argue that the amount of rewriting that needs to be done to create these updates is not burdensome. Perhaps the changes to the firmware are modest. I don't know myself, since I'm not a software engineer. But what I do know is the question of how burdensome the firmware changes are is a different one from the question of whether there are firmware changes at all (which there are).

 

[Wokis]

If they do that, then arguably, 2019 Mac Pro owners will have gotten even more screwed than people who bought the last Power Mac G5s. The Power Mac G5 people got three years between the discontinuation of their machine and the Snow Leopard release - if Sonoma is the last release for Intel Macs, 2019 Mac Pro owners would get... 15 months?!

Also, today, Apple's refurb store still has 2019 Mac Pros, 2020 iMacs, 2019 16" MacBook Pros and 2018 Mac Minis. Mostly stupid/weird configurations - e.g. who wants an i9 16" MacBook Pro with 16GB of RAM and a 2TB SSD for $2739CAD? Out of those, the 2018 Mac mini is the most endangered for a post-Sonoma world. But would they really drop machines that they were selling <18 months before? (I guess they did it to the iPod touch 7th-gen, didn't they?)

Interesting question - what was the last-selling non-Sonoma-friendly machine? The 2017 12" MacBook (discontinued July 2019)? The 2017 iMac (discontinued March 2019)? So... assuming it's the 12" MacBook, it would have gone ~50 months between discontinuation and the first macOS release that drops it.

I feel it's a bit of an apples to oranges comparison since we're looking back at the time where OS's were a sold product.

What would be the most absurd to me would be keeping x86 support for another year just to appease the 2019 and early 2020 models. That's a relatively low number of devices to cater to while fighting for a semblance of feature parity, or just skipping making the OS work the same and having confused customers.

Easier to just focus on the AS Macs getting the new features and keep the Intel Macs on a separate OS with security patching. Remember Apple doesn't really owe anyone new software features for free.

Refurb store pricing on old products is absurd at times. They sold 2013 Mac Pros when it was clear it was going to get dropped with Ventura.

Mac Pro 2019 users can't even run MacOS with AMD 7000-series GPUs at the moment so Apple continuing with disrespecting this userbase also wouldn't surprise me one bit.

 

[Bodhitree]

So, my theory is that macOS 15 will drop the 2018 models, macOS 16 will drop the 2019 models (including, sadly, the 2019 Intel Mac Pro), and macOS 17 will drop the 2020 Intel models, phasing out what is left of Intel support.

That being said... I wonder how big the installed base of 2020 Intels is. It might be hard to resist the temptation to just cut the 2020s along with the 2019s and be done with Intel a year earlier. Especially since there's only a 27" iMac, a MacBook Air, and a 13" MacBook Pro in 2020 - no Mac Pro, no Mac Mini, no 15-16" MacBook Pro, etc.

It would make sense, wouldn’t it. I reckon they will drop the 2020 models with the 2019’s, and macOS 17 will be announced in 2026 without Intel support.

 

[spiderman0616]

It would make sense, wouldn’t it. I reckon they will drop the 2020 models with the 2019’s, and macOS 17 will be announced in 2026 without Intel support.

Probably about right. I'm so curious about what the general landscape of these things will be by then. For example, will Intel's business have been impacted significantly by then or not so much? Will macOS adoption continue accelerating or will it flatten out again before then? If it is still growing at that point, will this shut off kill the growth somehow?

To be clear, I'm not suggesting Apple Silicon is going to kill Intel or Windows or anything like that. I'm just really fascinated by the empirical results of sea-changes like this. Especially when you throw things like iPad and Vision Pro in the pipeline.

 

[VivienM]

Probably about right. I'm so curious about what the general landscape of these things will be by then. For example, will Intel's business have been impacted significantly by then or not so much? Will macOS adoption continue accelerating or will it flatten out again before then? If it is still growing at that point, will this shut off kill the growth somehow?

To be clear, I'm not suggesting Apple Silicon is going to kill Intel or Windows or anything like that. I'm just really fascinated by the empirical results of sea-changes like this. Especially when you throw things like iPad and Vision Pro in the pipeline.

If you want to see what happens to Intel's business, look at Ampere and some of the other efforts to make non-x64 server/data center processors.

One of the underappreciated dangers to Intel's business is that, well, it's relatively easy for Big Cloud to move to a new CPU architecture. I'm sure Microsoft could run Exchange Online or SharePoint Online on ARM-based servers quite easily (they may already be doing it) and if the cost savings, either up-front or in power/space/etc over that server's lifetime, are there... well, multiply those cost savings by the tens/hundreds of thousands of servers involved in providing these kinds of Big Cloud services and let's just say the case for recompiling their software stack becomes very, very, very strong.

I think Big Cloud going away from x64 would hurt Intel infinitely more than Apple going to TSMC-made Apple Silicon chips for the Mac. Have you seen the price of some of those Xeon Platinums or whatever they are calling the Sapphire Rapids lineup, for something that probably doesn't cost much more to manufacture than the i5s Apple put in MacBook Airs and iMacs?

 

[VivienM]

I feel it's a bit of an apples to oranges comparison since we're looking back at the time where OS's were a sold product.

What would be the most absurd to me would be keeping x86 support for another year just to appease the 2019 and early 2020 models. That's a relatively low number of devices to cater to while fighting for a semblance of feature parity, or just skipping making the OS work the same and having confused customers.

Easier to just focus on the AS Macs getting the new features and keep the Intel Macs on a separate OS with security patching. Remember Apple doesn't really owe anyone new software features for free.

Refurb store pricing on old products is absurd at times. They sold 2013 Mac Pros when it was clear it was going to get dropped with Ventura.

Mac Pro 2019 users can't even run MacOS with AMD 7000-series GPUs at the moment so Apple continuing with disrespecting this userbase also wouldn't surprise me one bit.

But... look at what they've done on the mobile side. The first-gen 12.9" iPad Pro is only getting dropped in iPadOS 17. The 4th gen Apple Watch (one of my lucky Apple purchases... I say this as someone who bought the legendarily-short-lived iPad 3 "with retina display") is still supported in watchOS 10, etc.

One of the underappreciated elements of Apple's business strategy, somewhat like, say, a German luxury car maker, is that they try to keep the resale value of their products high. That lets them offer trade-in programs without setting their money on fire, and it lets passionate fans buy new Apple products more frequently than they otherwise would (if you buy something for $2000 and you can sell it for $1000 in 3 years, you're more likely to buy another $2000 product in 3 years than if your $2000 item was now worth $50). This is something that the Android vendors, or for that matter Dell/Lenovo/etc on their consumer systems, absolutely do not do.

The day a given machine is dropped from future major OS upgrades is the day that that machine's resale value tanks..

One thing that surprises me is that, from what I can tell, there isn't a lot of Apple Silicon exclusives in Sonoma. A few little nifty things here and there, but nothing that would make the ordinary gal happy with her 2020 4-Thunderbolt-port Intel MacBook 13" want to throw it out the window. I don't think Apple is really trying to antagonize people who spent big money on Intel machines that are still under their three-year AppleCare just yet. There will come a time to start antagonizing those people and nudging them to the Apple Store to buy an M4 system, but that doesn't seem to be just yet.

And really, why antagonize people? The Mx machines are selling well. Their performance speaks for itself. The M2 lineup is a spectacular upgrade for people whose machines were dropped in Ventura or Sonoma. Software-wise, I would say the transition has been a smashing success - unlike with PPC to Intel where it took years for the big developers to port their code, here, all the big guys have ARM-native software, most of the small guys do, etc - the remaining Intel software in my experience is mostly small cross-platform guys (e.g. the folks who make PingPlotter) and lazy Electron developers (the folks who make Authy, I'm looking at you). So Apple certainly doesn't need to drop the Intel machines stupidly early to nudge developers into ARM-native code - they are almost all there.

 

[ADGrant]

Software-wise, I would say the transition has been a smashing success - unlike with PPC to Intel where it took years for the big developers to port their code, here, all the big guys have ARM-native software, most of the small guys do, etc - the remaining Intel software in my experience is mostly small cross-platform guys (e.g. the folks who make PingPlotter) and lazy Electron developers (the folks who make Authy, I'm looking at you). So Apple certainly doesn't need to drop the Intel machines stupidly early to nudge developers into ARM-native code - they are almost all there.

Is that adjective needed for Electron developers?

 

[VivienM]

Is that adjective needed for Electron developers?

While I agree that anyone who thinks Electron is a good idea probably deserves to go back to writing software in BASIC on the Apple II, some Electron developers at least make some effort to produce a half-passable product out of this monstrosity of a platform. And then there are the especially lazy ones who think that using Rosetta 2 to run an old version of Intel Chromium to run their JavaScript-based app is a swimmingly good idea, because, well, why not?

The thing is, Electron shipped their first AS-friendly version in... November, 2020. You'd think all Electron apps could be AS-native 2.5 years later... and many are, but there are some glaring exceptions. For me Authy is the big one.

 

[ADGrant]

Has anyone attempted to see if there are any similarities between the T2 and TPM on the Intel/AMD side? I know that the T2 functionality was integrated into Apple Silicon, but I can't help but wonder if they're similar on some level.

I suspect they are not similar at all. The T2 is an ARM64 SoC based on the A10 used in the iPhone 7. It runs an operating system called BridgeOS which is probably based on iOS. It does not only handle encryption and security. The SSD controller is build into the T2 SoC (similar to how the ARM64 Macs work).

I suspect that all the T2 Macs are fairly similar from a software support perspective. They all have recent Intel CPUs (8th gen to 10th gen), similar GPUs (either integrated Intel or AMD). I can't see them not releasing new versions of MacOS for the MacPro until that machine no longer has active support contracts. So that is at least two more releases after Sonoma. I don't see them dropping any T2 Macs in the meantime but the 2019 iMac may be dropped before the end since it doesn't have a T2.

 

[VivienM]

I suspect they are not similar at all. The T2 is an ARM64 SoC based on the A10 used in the iPhone 7. It runs an operating system called BridgeOS which is probably based on iOS. It does not only handle encryption and security. The SSD controller is build into the T2 SoC (similar to how the ARM64 Macs work).

I suspect that all the T2 Macs are fairly similar from a software support perspective. They all have recent Intel CPUs (8th gen to 10th gen), similar GPUs (either integrated Intel or AMD). I can't see them not releasing new versions of MacOS for the MacPro until that machine no longer has active support contracts. So that is at least two more releases after Sonoma. I don't see them dropping any T2 Macs in the meantime but the 2019 iMac may be dropped before the end since it doesn't have a T2.

I could probably answer my own question by booting up a non-macOS OS on my 2020 Intel iMac, but isn't the T2 also the sound controller?

 

[ADGrant]

I could probably answer my own question by booting up a non-macOS OS on my 2020 Intel iMac, but isn't the T2 also the sound controller?

If you use Bootcamp to install Windows, it will install Apple developed drivers. I haven't tried installing Linux except through VMs so that would be interesting.

 

[VivienM]

If you use Bootcamp to install Windows, it will install Apple developed drivers. I haven't tried installing Linux except through VMs so that would be interesting.

And what do the Apple-developed drivers call it in Windows? "Apple Something Audio Controller"?

Just booting up Linux and doing an lspci would be very interesting.

 

[ADGrant]

And what do the Apple-developed drivers call it in Windows? "Apple Something Audio Controller"?

Just booting up Linux and doing an lspci would be very interesting.

You might want to take a look at this website. https://wiki.t2linux.org/

They have a lot of interesting details on getting Linux running on a T2 Mac.

You are right about the Audio btw. https://github.com/t2linux/apple-bce-drv

 

[VivienM]

You might want to take a look at this website. https://wiki.t2linux.org/

They have a lot of interesting details on getting Linux running on a T2 Mac.

You are right about the Audio btw. https://github.com/t2linux/apple-bce-drv

It's funny - at least in vintage Apple forums, I have tended to be dismissive of Intel Macs, sometimes even calling them high-quality dressed-up IBM PC convertibles with a *NIX OS. (As opposed, to, say, the beige Macs in the 1990s... or even the PPC machines starting with the G3 iMac and B&W G3) There's something sad about the fact that the Intel Macs can boot PC operating systems (I wonder if someone has tried to boot MS-DOS 6.22 on an early Intel Mac...) yet have zero compatibility with any "legacy" Mac software. Not to mention none of the cool hardware features that defined the Mac in the Mac II to beige G3 era.

But it seems like that is certainly NOT true with the T2 Intel machines. Those actually have custom Apple stuff for all kinds of things and won't play well with your generic OS for x64 machines.

 

[ADGrant]

It's funny - at least in vintage Apple forums, I have tended to be dismissive of Intel Macs, sometimes even calling them high-quality dressed-up IBM PC convertibles with a *NIX OS. (As opposed, to, say, the beige Macs in the 1990s... or even the PPC machines starting with the G3 iMac and B&W G3) There's something sad about the fact that the Intel Macs can boot PC operating systems (I wonder if someone has tried to boot MS-DOS 6.22 on an early Intel Mac...) yet have zero compatibility with any "legacy" Mac software. Not to mention none of the cool hardware features that defined the Mac in the Mac II to beige G3 era.

But it seems like that is certainly NOT true with the T2 Intel machines. Those actually have custom Apple stuff for all kinds of things and won't play well with your generic OS for x64 machines.

Yes, I think the T2 machines were the first phase in the transition to Apple Silicon.

 

[VivienM]

Yes, I think the T2 machines were the first phase in the transition to Apple Silicon.

I think that's probably right... and I would guess something like the T2 audio controller is very, very similar to both an iOS audio controller and the M1 audio controller, so it probably gave them an opportunity to get those iOS-based components playing nicely with full fledged macOS, Mac workloads, etc ahead of their use in Apple Silicon.

 

[satcomer]

To me sense the release of the new Mac Pro Silicone now the Universal software needs to be updated soon! The clock is ticking and less than a year Apple will stop with the Universal software anymore! I expect our silicon versions of the software with new Mac OS comes out!

 

[VivienM]

But... look at what they've done on the mobile side. The first-gen 12.9" iPad Pro is only getting dropped in iPadOS 17. The 4th gen Apple Watch (one of my lucky Apple purchases... I say this as someone who bought the legendarily-short-lived iPad 3 "with retina display") is still supported in watchOS 10, etc.

One of the underappreciated elements of Apple's business strategy, somewhat like, say, a German luxury car maker, is that they try to keep the resale value of their products high. That lets them offer trade-in programs without setting their money on fire, and it lets passionate fans buy new Apple products more frequently than they otherwise would (if you buy something for $2000 and you can sell it for $1000 in 3 years, you're more likely to buy another $2000 product in 3 years than if your $2000 item was now worth $50). This is something that the Android vendors, or for that matter Dell/Lenovo/etc on their consumer systems, absolutely do not do.

The day a given machine is dropped from future major OS upgrades is the day that that machine's resale value tanks..

One thing that surprises me is that, from what I can tell, there isn't a lot of Apple Silicon exclusives in Sonoma. A few little nifty things here and there, but nothing that would make the ordinary gal happy with her 2020 4-Thunderbolt-port Intel MacBook 13" want to throw it out the window. I don't think Apple is really trying to antagonize people who spent big money on Intel machines that are still under their three-year AppleCare just yet. There will come a time to start antagonizing those people and nudging them to the Apple Store to buy an M4 system, but that doesn't seem to be just yet.

And really, why antagonize people? The Mx machines are selling well. Their performance speaks for itself. The M2 lineup is a spectacular upgrade for people whose machines were dropped in Ventura or Sonoma. Software-wise, I would say the transition has been a smashing success - unlike with PPC to Intel where it took years for the big developers to port their code, here, all the big guys have ARM-native software, most of the small guys do, etc - the remaining Intel software in my experience is mostly small cross-platform guys (e.g. the folks who make PingPlotter) and lazy Electron developers (the folks who make Authy, I'm looking at you). So Apple certainly doesn't need to drop the Intel machines stupidly early to nudge developers into ARM-native code - they are almost all there.

One other thought I had. The danger of dropping too many machines, too soon is that third-party developers will stay focused on the last OS version to support those machines for too long. This can feed a deadly spiral - if third-party developers don't adopt your newest OS innovations, there's no reason for consumers to upgrade to the newest OS and there's no reason for consumers to buy newer machines.

This can be seen in Windowsland. One of the great tragedies of the (unfair, IMO, but that's an unrelated issue) rejection of Windows Vista is that for, many, many years, Windows software developers had to keep supporting XP. That meant not implementing any of the newer APIs, etc in the newer OSes. In 2023, you can find software whose newest version runs on XP. This is a 21.5-year-old OS at this point. A significant amount of current other Windows software will run on Windows 7, a now 13.5 year old OS. And one of the effects of this stagnation is that instead of developers adopting Microsoft's new frameworks, most new software came to be written for Chrome rather than Windows. And, of course, Chrome doesn't need Windows, it doesn't need x86, etc. And with no new software being written for new versions of Windows, consumers and businesses both lose interest in Windows, keeping current, etc, and the platform just atrophies. Windows used to be innovative (at least compared to its previous version), people used to be excited about new versions of Windows, and now, it's gotten to the point where an enthusiast like me is happily sticking to Windows 10 on my 'too-old-for-11's-insane-hardware-requirements' high-end desktop. (And I've gone from zero Macs in 2014 to three Macs in 2022)

Compare this with Mac land, where many developers today might only support Monterey, Ventura, and maybe Big Sur. They can rely on 3-year-old APIs safely, while their Windows counterparts are barely moving on from a 13.5-year old platform.

I suspect that that's a BIG reason for iOS devices having the long OS support life that they do. Apple does not want a world where developers are ignoring the platform's innovations for years because they don't want to stop support for half the installed base out there.

 

[theorist9]

Easier to just focus on the AS Macs getting the new features and keep the Intel Macs on a separate OS with security patching. Remember Apple doesn't really owe anyone new software features for free.

That was an idea I initially had as well--if it's burdensome to support Intel Macs beyond, say, Sonoma (MacOS 14), they could make Sonoma the last Intel OS, but extend its security support out an extra two years, so that support would end in 2028 instead of 2026 for the 2019 Mac Pro. But there's a few reasons they wouldn't do that:

1) It means maintaining Sonoma for *everyone* for longer than Apple wants. Apple wants to push everyone onto the current OS, and extending an old OS so people could stay on it longer would run counter to that message. Further, as part of that message, Apple has explicitly said they don't fully security-patch older supported OS's; only the current OS is fully patched.

2) If they did this then, confusingly, Sonoma's suppport would go longer than that for MacOS 15, which would (if it returned to the normal schedule) end in 2027. Though they could avoid that issue by having MacOS 15 be the last Intel OS, and extending its security patch for just a year. Then it would lose support at the same time as MacOS 16.

3) A big part of the work of maintaining an OS may be keeping the security patches updated, so it might not be much of a time savings for Apple.

For this reason, I'm guessing they'll continue with their current OS support schedule, and that support for the 2019 MacPro will end in 2028 with MacOS 16. But we shall see....

 

[VivienM]

That was an idea I initially had as well--if it's burdensome to support Intel Macs beyond, say, Sonoma (MacOS 14), they could make Sonoma the last Intel OS, but extend its security support out an extra two years, so that support would end in 2028 instead of 2026 for the 2019 Mac Pro.

And fundamentally, the idea that it's 'burdensome' to support Intel Macs is... laughable.

This is the biggest company in the world by market cap, a company with tens of thousands of engineering employees. This is a company that probably has a higher ratio of engineers per product than any other technology company, a company with a very narrow product line, and a company that has been extremely judicious for over two decades in minimizing the number of parts, variations, etc across its product line - it's not like they've used 50 different SSD controllers and 10 different wifi chipsets across a few years of Intel Macs. Most of the code required to support Intel Macs (drivers, etc) has already been written and it's not like any new Intel Macs are coming. They have, if we count the NeXT days, 30 years of experience writing portable code in high-level languages and supporting multiple CPU architectures. Lots of development tools they rely on (e.g. LLVM/clang) are open source and used by others who will continue to maintain them on Intel/x64 for the indefinite future.

How big of a team was required to engineer the Vision Pro headset? I'm sure you need 0.1% of those engineering resources to keep supporting macOS on Intel development.

How big of a team was required for the stealthy Project Marklar in the early 2000s? From what I read, it was pretty much one dude in a home office halfway across the country who maintained the secret Intel port... and he wasn't starting from an established code base with functional drivers, either.

This is a political decision - modern Apple doesn't want devices in the field getting too too too old, first because they'd like to sell you a new Mac every ~5 years, but also doesn't want to end up with Windows XP disease where a need to support old devices/hardware/etc holds the platform back until it atrophies a decade and a half later. So they'll drop Intel when they think enough of the installed base is on Apple Silicon and they think the rest will be willing to make the move in the following 18 months.

And it's worth going back to 2009 - PowerPC was dropped late in the Snow Leopard development cycle not because it was too burdensome (if it was, they wouldn't have started developing it in the first place and gotten as far as they did) but because Steve Jobs made a decision that it was time. And he probably was not wrong... it's not like the major third-party developers were still supporting Leopard on PPC in 2014 or 2015, so Microsoft/Quark/Adobe/etc were probably looking at the same data Steve Jobs was. Even VLC, which to this day supports Windows XP and OS X Lion in its newest version, released its last version for PPC in 2014.

 

[theorist9]

And fundamentally, the idea that it's 'burdensome' to support Intel Macs is... laughable.

No need to be gratuitously insulting. It's behavior like that which causes the internet to be more swamp-like. Just saying you disagree, and why, is sufficient.

For instance, when you wrote:

the 2018 15" MacBook Pro has a T2... and is not on the Sonoma list

I could have responded "That's clueless. The 2018 MacPro is on the Sonoma list". But I didn't. I try to keep things collegial, so I wrote this:

Sorry, not following. The 2018 MacBook Pro is on the Sonoma list.

Or when you wrote:

But you're not writing new firmware - you're maintaining existing firmware that was written years ago.

I could have responded "That's ignorant. New OS's do require new firmware." But I didn't. I instead wrote:

Maybe this is an issue of semantics. I do know that, when you update to a new OS, both Intel and AS Macs get firmware updates.

I perfectly understand taking a swing at someone that's taken a swing at you; I do it often. What I don't understand is being rude to someone that's been polite to you. Maybe you've just got really bad social skills (yes, that was an example of the former).

 

[VivienM]

No need to be gratutiously insulting. It's behavior like that which causes the internet to be more swamp-like. Just saying you disagree, and why, is sufficient.

I perfectly understand taking a swing at someone that's taken a swing at you; I do it often. What I don't understand is being rude to someone that's been polite to you. Maybe you've just got really bad social skills (yes, that was an example of the former).

Sorry - I was not trying to be 'rude' or 'gratuitously insulting.' Dismissive, yes, because I do think that the idea that a company with Apple's staffing levels, financials, and streamlined product line built out of a very small number of frequently-used component parts would be unable or overburdened to support two architectures for another year or two or three is... difficult to believe. That they wouldn't want to for various business reasons, absolutely, but if those business reasons went the other way, I'm sure they could easily keep supporting the existing Intel machines for another 5+ years and the staff/money/etc allocated to that would be trivial.

Plenty of people in this industry support a lot more things with a lot fewer resources than Apple - whether it's the PC OEMs who need to provide BIOSes, etc for dozens of different models every year, the open source folks who support multiple architectures (FreeBSD has 2 Tier 1 architectures, 13 Tier 2 architectures, and supports a lot more components on each of those architectures than Apple), etc.

And I don't understand the point about the new firmware, sorry. Maybe we are arguing about the meaning of 'new' - there may be firmware updates in every OS update, but I would not call that 'new'. Mildly updated, absolutely, but whatever those updates may be, I think it's very unlikely that they require a huge development team, especially this many years after the machines first shipped. And I would add that in Windowsland, you can certainly keep running new versions of Windows with BIOSes/UEFI that hasn't been updated in a decade, so one may wonder how much maintenance is actually required beyond throwing in new microcode from Intel and maybe fixing a few security bugs.

This is complete speculation on my part, so I could be widely off, but I would guess they could continue supporting Intel machines, or at least the machines left on the Sonoma list, with a two-digit number of staff assigned to the project. Maybe less. Might they make a business decision that allocating 15 engineers to something that lowers customers' desire to go out and buy new M3 or M4 machines is a bad idea? Sure.

And maybe I do take this a little personally - I have a 2020 iMac, my mom has a 2020 4-Thunderbolt 13" MacBook Pro. My 2020 iMac, well, that was a gamble on my part buying it a year ago right after the release of the Mac studio. I gambled that the iMac would give me a high-RAM retina display desktop for 1/3rd the cost and about 1/2 the life expectancy of a Mac studio + studio display. I suppose if it gets dropped after Sonoma, I just gambled wrong and 1/2 the life expectancy of a Mac studio turned out to be 1/3 the life expectancy. That's on me. My mom's, what was she supposed to do, her late-2013 15", at six and a half years old and still supported by the then-current and the then-next version of macOS, suffered catastrophic SSD failure 4 months before the first M1s came out?
I've been burned by Apple before - I bought one of those iPad 3s with retina display that got obsoleted in 6 months. Excitedly bought an iPod mini 2nd-gen... which also had a life of about 6 months before the nano unexpectedly came along. I saw what Apple did to the people who bought G5s in early 2006. (Sometimes I've also gotten lucky, e.g. with my series 4 Apple Watch) I've been scorched by Microsoft and have a lovely i7-7700 desktop that I thought would last me a decade but that officially doesn't meet their "performance and reliability expectations" for Windows 11 to show for it, so less than mid-way through its life expectancy, I end up stuck on the old OS or gambling with an unsupported OS that could break in any monthly patch. So yes, when someone says that it might be 'burdensome' for the world's largest company, a company with infinite resources that pays almost US$4 billion per quarter in dividends and spends US$20 billion per quarter on stock buybacks, not to send those machines to e-waste for another year, I might react a bit strongly and dismissively, sorry.

 

[theorist9]

Sorry - I was not trying to be 'rude' or 'gratuitously insulting.' Dismissive, yes, because I do think that the idea that a company with Apple's staffing levels, financials, and streamlined product line built out of a very small number of frequently-used component parts would be unable or overburdened to support two architectures for another year or two or three is... difficult to believe. That they wouldn't want to for various business reasons, absolutely, but if those business reasons went the other way, I'm sure they could easily keep supporting the existing Intel machines for another 5+ years and the staff/money/etc allocated to that would be trivial.

Plenty of people in this industry support a lot more things with a lot fewer resources than Apple - whether it's the PC OEMs who need to provide BIOSes, etc for dozens of different models every year, the open source folks who support multiple architectures (FreeBSD has 2 Tier 1 architectures, 13 Tier 2 architectures, and supports a lot more components on each of those architectures than Apple), etc.

Thanks, I appreciate that. I hear where you're coming from, since I see this on the hardware side--for instance, I never understood why they took so long to update the Mini or the Trashcan Mac Pro, when clearly there was a need to do both much sooner than they did. The argument I kept hearing was that the Mac side is only a small percentage of the company, so they can't put too much focus on it—but it's not as if they need Tim Cook to do these updates personally!

When it comes to the resources the Mac divsion has, what matters is not its relative size, but its absolute size. And in absolute size, its revenue matches that of a Fortune 100 company--it's huge! And if a Fortune 100 computer company sold only 9 computer models (in 2015 I believe it was the Mini, Air, MacBook, 13" MBP, 15" MBP, 21.5" iMac, 27" iMac, and Mac Pro), the idea that they'd have a hard time keeping all of them updated because they are too small makes no sense.

Having said that, the fact is that Apple does like keeping its range of offerings--both hardware and software--limited. And it works for them. Do I know how much it would "distract" them to have to maintain both Intel and AS versions of its OS's? As I said before, no. My only point is that I don't think we should be so quick to just dismiss it as trivial. It may not be.

And maybe I do take this a little personally - I have a 2020 iMac, my mom has a 2020 4-Thunderbolt 13" MacBook Pro. My 2020 iMac, well, that was a gamble on my part buying it a year ago right after the release of the Mac studio. I gambled that the iMac would give me a high-RAM retina display desktop for 1/3rd the cost and about 1/2 the life expectancy of a Mac studio + studio display.

Well, it so happens I bought a 2019 i9 iMac (in pristine condition) a year ago for that very reason. After I added 128 GB RAM and a 2 TB SSD, I was out only $1800—and it's still got AppleCare+ on it. Getting that functionality from AS would necessitate a 128 GB/2 TB Ultra Studio with an ASD, which would have been nearly 4x the cost. And that financial layout made no sense, since I benchmarked the M1's performance against the iMac on Mathematica, which is the most time-demanding program I use, and it averaged only 10% faster. So I figured I would need to wait a few more generations (M3 or M4) before the peformance delta will be big enough to justify the cost.

But where you and I differ—and this is entirely personal—is that, even if Sonoma is the last OS my iMac supports (which gives me 4.5 years total), I'm OK with that, since for me it was only intended as a stopgap until AS gets fast enough to justify the cost. Plus I expect when I upgrade to AS it will be a lot cheaper than it is now, because by then (with LPDDR5x), I should be able to get the RAM I need without having to buy an Ultra (which is otherwise wasteful for me, as I don't need the multiple processors; most of my programs are single-threaded). And my issue with the eWaste also differs from yours: It's that you can't use the iMac as an external display after the processor becomes obsolete. [Yeah, there's AirPlay and Luna, but they're laggy and sacrifice video quality.]

 

[pc297]

They do want to the developers to embrace native Apple Silicon compatibility. It's a fairly delicate balancing act... folks are known to target the least common denominator. The case of 32-bit software and Windows illustrates how a software transition can be majorly messed up.

I do think that most of it can be resolved by taking away the ability to generate x86 code in the default built tool starting from a certain SDK. Devs can still use older SDKs and custom toolchains to target x86 code, but that's going to be more annoying than just fixing the outstanding issues in the code itself for most.

Agreed, at the end of the day it's simply a matter of resources and staff that Apple devotes to various projects; some of them are long-term and some of them temporary and Rosetta 2 falls in the second category like Rosetta 1, Classic, and before that the 68k compatibility mode for ppc cpus.

Maintaining Rosetta 2 costs money and manpower as it continuously needs to be integrated in macOS and likely updated! Otherwise Rosetta would still have been around for much longer; like Classic the last versions simply didn't work under the next version of MacOSX due to broken frameworks etc as Apple diverted resources allocated to them to move on to newer things.

So are things at Apple, if history is any indication; that said it's up to the goodwill of some volunteer users to maintain backwards compatibility for as long as possible as was done in the past for so many features, eg macOS9 helper, XpostFacto, all the @dosdude patches, OSX patcher, Pike firmware, OCLP, 10.6 ppc etc! So maybe some good souls could be kind enough to maintain rosetta and link the correct frameworks i the macOS versions following its abandon, however it's source is entirely closed as it was bought from IBM as PowerLX86 who in turn bought it as QuickTransit...

 

[pshufd]

Agreed, at the end of the day it's simply a matter of resources and staff that Apple devotes to various projects; some of them are long-term and some of them temporary and Rosetta 2 falls in the second category like Rosetta 1, Classic, and before that the 68k compatibility mode for ppc cpus.

Maintaining Rosetta 2 costs money and manpower as it continuously needs to be integrated in macOS and likely updated! Otherwise Rosetta would still have been around for much longer; like Classic the last versions simply didn't work under the next version of MacOSX due to broken frameworks etc as Apple diverted resources allocated to them to move on to newer things.

So are things at Apple, if history is any indication; that said it's up to the goodwill of some volunteer users to maintain backwards compatibility for as long as possible as was done in the past for so many features, eg macOS9 helper, XpostFacto, all the @dosdude patches, OSX patcher, Pike firmware, OCLP, 10.6 ppc etc! So maybe some good souls could be kind enough to maintain rosetta and link the correct frameworks i the macOS versions following its abandon, however it's source is entirely closed as it was bought from IBM as PowerLX86 who in turn bought it as QuickTransit...

If we lose Rosetta 2, then I will either have to run Windows in a Virtual Machine or get a Windows laptop to run one of my programs.