It's a royal pain for engineers too. If you actively support, say four versions, then when you fix a bug, you have to check it into the live line for the next release, and three prior versions. So you have to backport it to three code lines, and also run regression tests for your change and for integration testing. You also have beta testing for the old releases. Carrying old software versions is expensive and I assume the same is true for hardware.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
macOS Tahoe Might Support One Fewer Mac Than Previously Rumored
- Thread starter MacRumors
- Start date
- Sort by reaction score
It's a royal pain for engineers too. If you actively support, say four versions, then when you fix a bug, you have to check it into the live line for the next release, and three prior versions. So you have to backport it to three code lines, and also run regression tests for your change and for integration testing. You also have beta testing for the old releases. Carrying old software versions is expensive and I assume the same is true for hardware.
I think they should do 7 years of free updates and then paid updates after that for machines to get updates after that. Plenty of people I know would happily pay a small fee to keep their older machine usable.
I was disappointed with Microsoft with Windows 11, though at least Windows has a clearer support lifecycle and policy for its software.
Ironically, that's what Microsoft effectively did - you can get security updates for three years for Windows 10 if you pay.
I would have preferred to pay for 11. We know 11 runs unsupportedly just fine on those machines...
We saw Intel update microcode for 9+ year old products to deal with Spectre.
You had some people also believe Apple would lower prices and not eat the profits left on the table with the switch to M1.
We're already at 5 years, I would expect M1 gets updates for a very very long time. The foreseeable future. Truth is there really isn't anything dramatically different about M1 vs. M4 that would actually cut off M1 sooner.
There is literally nothing about M1 that would cause it lose support before the M4. Support is 90% architectural and 10% performance.
The CPU in M1-M3 uses the ARMv8 architecture while the M4 switched to the ARMv9 architecture. If support is 90% architecture how can there be "literally nothing about M1 that would cause it lose support before the M4"? This doesn't even consider changes in the architecture of the GPU and NPU which aren't as well documented.
You really think they are going to keep supporting M1 macs for 10 years? I think we will be lucky if they support them for 8 plus 2 extra years of security updates.
Maybe for the M4 or M6, which have been or are expected to be big jumps in performance, we’ll see one extra year of support.
But yeah, ideally I’d love to see Apple Silicon Macs support extended to 10 years, eventually.
The MacBook Air 2020 is only about 6 months older than the MacBook Air M1 and apparently it is being dropped. It also has a 10th generation Intel processor same as the MacBook Pros that are being retained (and architecturally newer than the Intel Xeon W-32xxM used in the Mac Pro 2019 that was retained).
Not that I disagree with the decision to retain those but rather it should be clear that across product line Apple drops support for models based on things other than architecture and specs.
I bet Apple will keep the M1 for macOS 28 but I bet that will be its last year.
literally all of the really expensive ones with the Iris GPU +128MB cache. while everyone else was busy selling CPUs with the HD 4600 to their $3k paying customers.
Ironically enough, Apple was the only one not saving one cent on them.
then came 2016, Intel made an even bigger iGPU for Apple with the new xeons for laptops but they decided to follow Windows manufacturers and cheap out on the 15" inchers.
Last edited:
Agree this is why I would say Apple should release fewer OS versions with more time between them. It would streamline things for everyone if major releases only came every 2-3 years but they were really solid. A model supported for 3 releases @ 3 years each would be 9 years. Maybe even only support/patch N and N-1 and but that would span 6 years rather than 3 years of (N, N-1, and N-2 where N is typically not feature complete not to mention production ready for the first 6-12 months).
Yes I've noticed the same pattern except if the MacBook Air 2020 is dropped before macOS 26, that would actually be 1 + 5 upgrades. A little light.
Agree support based on introduction date rather than discontinuation date (or whatever) hurts anyone who doesn't buy slowly updated models at introduction (i.e. anything not on the annual refresh cycle). Apple shouldn't expect its buyers to run a regression on historical patterns to determine how long their purchase will be supported.
Isn't that (1+5) what the 2017, 2018, 2019 Airs got? So... that makes me think the 2020 won't get more...
Rosetta2 is not just helpful for running old Mac apps but is used to support Intel binaries running in VM/container. The ability to run Intel executables for Windows in Parallel and/or Docker containers is valuable for a lot of people.
How much do you think the Intel legacy code is holding back macOS and why?
I have a feeling that is a use case that Apple couldn't care less about...
Also, isn't the 'right' way of running Intel executables on Windows to use the ARM version of Windows and Microsoft's x86 emulator?
Maybe but if Apple really didn't care about Rosetta2 users, why did they extend Rosetta2 to VMs and the like and then add AVX2 emulation in Sequoia?
True that should work though albeit a little less optimally as I understand MS's x86 emulator is not as good as Rosetta2. Similarly each other OS that needs could bring their own emulator. Most likely not as good as/slower than Rosetta2 but would allow Apple to push the functionality off. Seems odd though after investing to make it work as well as they did.
That's an interesting question.
I guess I'm just expecting Rosetta 2 to go away an OS version or two or three after Intel support is dropped. Treat 64-bit Intel apps the same way as, oh, I don't know, 32-bit Intel apps? PPC apps?
I guess one difference is that Apple is believed to own Rosetta 2 while Rosetta 1 was licenced. But... I just can't imagine they want to carry that Intel baggage any longer than they have to.
I’d bet at least another 3 years, next year they’ll drop intel, the year after will likely be an optimization update, I’d bet *maybe* the year after (that would be 2028) that we lose M1s
If that’s the logic I’d expect them to drop ARMv8 all at the same time, which would imply a mostly regular support cycle for the M3, a longer than normal for M2, and extremely long legs for M1.
It’s worth remembering though that they’re still introducing ARMv8 M3 chips alongside the M4 ones, the M3 Ultra is the new studio's higher end chip
Also worth noting that while Apple is exposing the ARM ISA the internals arent reference designs
Portable code usually has conditional compilation for code that is platform-specific so it doesn't have to incur a performance penalty on any of the ports.
For games and other programs that are Intel Windows only. Fidelity Active Trader Pro is one example. Intuit Turbotax is another.
I'm considering going back to just Windows on my desktop. It currently has an iMac Pro, custom Windows build and an M1 Max Studio on it and one of my revenue-generating programs runs really well on it but there's a 66% performance penalty running on the M1 Max Studio. My calculations indicate that even an M4 Max Studio would still be inferior to my i7-10700 Intel CPU. If Apple drops Rosetta 2, then I really have no reason to run Apple Silicon for the next couple of years on the desktop.
PPC ended a little soon but a little more understandable in the context. I think a lot of people are still a little bitter about dropping 32-bit Intel apps.
One difference from the PPC situation is that it broke compatibility with their old platform but that's about it. One of the selling points of switching to Intel chips was the broader compatibility. Bootcamp in the short-term but the on-going capability to run programs/applications from the most popular desktop platform is a valuable capability.
I'd also guess they don't but then I'd guess they also want to drop lots of things including pre-ARMv9 or 8GB models (and I am sure lots of older frameworks). However, the question should be around what creates value for the end-user rather than what costs Apple less. Apple could save money simply by not having any OS but then not much value.
If two or three years after Intel support is dropped, Apple opensourced Rosetta2 that would be a good alternative as far as I am concerned. Guessing they wouldn't want to give up the IP but always sad to see IP like that decay in a vault somewhere.