There's also a long history of benevolent behavior by Apple...
In other words, there's proof for whatever it is you are looking for, which makes all of our narratives speculation.
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.
Apple's Craig Federighi Further Explains Why Stage Manager is Only for M1 iPads
- Thread starter MacRumors
- Start date
- Sort by reaction score
In other words, there's proof for whatever it is you are looking for, which makes all of our narratives speculation.
Craig is that you? Basic iPad 9 owner here (so I have no agenda...I'm out of the equation anyway) but you'd have to be incredibly gullible to truly believe there are insurmountable technical reasons preventing all of OS16's features from running at the very least on the last A-series iPad Pros, even if some things are slightly pared down (though I don't believe they would need to be).
If I were to guess, there are two possibilities. One is that it's a bug and they accidentally left it on, which is not unprecedented. I've seen them accidentally leave a feature on when it's not supposed to be in a past beta. If it's a bug, it'll be shut off in beta 2. Two is that they're testing a lower virtual memory requirement to see how it performs. We'll see if it's left on. If so, then the requirements may change.iPad Air 5 base model lacks memory swap despite being a requirement for Stage Manager
64GB iPad Air 5 lacks memory swap despite Stage Manager 9to5Mac
With iPadOS 16, Apple announced multiple new features that are exclusive to iPads based on the M1 chip, such as...9to5mac.com
…
"But here’s the thing. As noted by developer Steve Troughton-Smith on Twitter, the base model of the iPad Air 5 is not compatible with virtual memory swap. That’s probably because the 64GB of internal storage is not enough for memory swapping. As Apple quietly suggests on its website, memory swapping on the iPad requires at least 128GB of storage in addition to M1.
This is totally understandable, but then comes the question: Why does Apple keep saying that virtual memory swap is a requirement for Stage Manager when the 64GB iPad Air 5, which supports Stage Manager, clearly has no virtual memory swap?"
Keep in mind betas are just that... betas. Things change over time. Bugs get fixed, features get deprecated or added, and requirements could change.
My inclination is that it's a bug and they left it on when it wasn't supposed to be. Whoever was doing the installer likely set the wrong flag thinking all iPad Air 5's got the feature.
There are facts such as no OS in existence run on A-series SoC's have virtual memory. There is no disputing this because we all know which OS'es are run by A-series and which ones are run by M1's. Only the M1's are running virtual memory.
We also know Apple creates their own custom SoC's and that they tailor the silicon on them for their desired feature sets with things planned years in advance. These are not general purpose SoC's but very customized. I think we can all agree on that. Apple will not put in things that won't ever get used. Would you agree with that?
If you don't agree, here's support for that. Every silicon wafer TSMC makes has multiple A-series dies on them. The bigger they are, the fewer they can fit on the wafer. Also the bigger the size and fewer the number of chips, the higher the likelihood of a lower yield. This is all quite accepted in the field of manufacturing chips. No company is in the habit of wasting transistors on features they will never use. Would you agree with that? Apple would be taking unnecessary risks adding features to the A-series chips, costing them more money than necessary and being able to have fewer SoC's on each wafer. If Apple were planning to run virtual memory on the A-series chips, they would have supported the 2018 and 2020 iPad Pros with virtual memory. But since they didn't, the only conclusion is that the capabilities do not exist.
Therefore, if no OS running on A-series supports virtual memory, would you agree that the controllers capable of managing virtual memory do not exist on the A-series SoC's?
The obvious conclusion since Stage Manager relies on virtual memory to meet the requirements of running 8 apps simultaneously where each app can allocate 16GB, that means virtual memory must be present for those requirements to be met since 8 x 16 = 128GB. Since no iPad has more than 16GB, real RAM is insufficient. The support for this is that the stated requirements for virtual memory is 256GB M1 iPads. The math shows 128GB is required for a swap file since with virtual memory, swap files must be the same size as the max memory supported. Therefore to use the entire flash drive of an iPad for virtual memory, Apple would have to use 128GB of that flash. Since that leaves no space for apps or the OS, the requirements are bumped up to the next higher capacity, which would be 256GB since there are no iPads with storage between 128GB and 256GB. And would you look at that? The stated requirements for virtual memory happens to be... 256GB of flash storage.
The final conclusion is that because virtual memory is required and only the M-series of chips supports virtual memory, Stage Manager will only work on M1 processors.
Some of the information above are indeed facts. The rest are logical deductions based on those facts.
That was mind blowing at the time. They used every trick in the book to get that one working.
Like what? You do know virtual memory (swap) is like pagefile on Windows & that's what a 30 year old feature? The chances that Axx can't support it due to something missing in hardware is "virtually" zero!
Everyone, you don't have to take Craig or Apple's word for it.
The specs for the M1 chip, A12Z, A12X are known. Just run the numbers with the maximum scenario. 8 of the most performance heavy apps running on the 12.9" display at 120hz and on a connected 5k display at 60hz, etc...
The specs for the M1 chip, A12Z, A12X are known. Just run the numbers with the maximum scenario. 8 of the most performance heavy apps running on the 12.9" display at 120hz and on a connected 5k display at 60hz, etc...
It was "accomplished" on many other computers, but it had never been accomplished on the iPhone or iPad -- until we got iPads with the M1 chip. None of the A-series chips were ever designed with the capability of selectively swapping pages of an app in and out. Would you like a link to a video where someone explains the details of this new capability that the M1 chip added to the iPad line? Would that help you understand?
Imagine someone has a 9th Generation iPad with an A13 chip. Imagine that Apple had enabled Stage Manager for this configuration -- even though they knew that the A13 chip cannot do the lightning-fast paging and swapping operations required for snappy performance. Further, Apple realized that having enabling Stage Manager has unacceptably-long delays when an external monitor is connected to the iPad. In this sceanario, Apple disables Stage Manager from running when an external display is connected.
Having the SM functionality disappear when an external monitor is attached is a complication. But Apple would never allow an OS functionality to disappear when an external monitor was added to the system; they would not tolerate a complication like that.
This A-series iPad limitation has no relationship to connecting a laptop to external monitors. Those laptop machines have processors that can selectively swap out pages from memory. They've been doing it for decades. And it's a safe bet that all new iPads will have chips with this capability -- in the next few years.
Do you need link to a video that explains this paging limitation of A-series processors?
Would you like a link describing the memory swapping system used by previous Apple OS's that didn't require hardware support in the processor to implement?
I was speaking to the recent iPad Pros, not a standard iPad consumer product. I can't imagine a processor that's faster than Intel's second most recent generation not being fast enough to handle basic GUI overlapping window tasks that processors 1/10 as fast can easily handle.
It wouldn't "disappear". The user would be told his configuration doesn't support an external monitor.
No thanks, please refer to the technical information above that you apparently weren't aware of at the time you composed your reply to me.
The fact is known that 2020 A12z iPad Pros can't output more than 4K/30. It will never connect to a 5K display which requires at least 32Gbps bandwidth.
Known fact: https://support.apple.com/kb/SP815?locale=en_US
In short, no 2020 ipad pro can connect to any shipping Apple monitor. Studio display or XDR.
Definitely. Don't even bother asking -- just provide it. I'm fascinated to know how an A-series chip could possibly provide paging of an app with no hardware support. It would seem like every single memory access would be prohibitively expensive, since the system can't generate a hardware fault when a page is not in memory. Please educate us -- please tell us how one could implement a paging system with no support in A-series processors with no hardware support for pages and page faults. What technical blogs covered this discussion? Provide links to those technical articles.
Remember the famous quote. You are entitled to your own opinion, but not your own facts. Unless you have facts to support your conjecture that A-series page faults are feasible, your opinion is worth very little.
Facts, please! Which exact non-M1 iPad Pro models are you talking about? The 5th generation uses an M1, so you're not talking about that. The 4th generation iPad Pro contains an A12Z chip. The 12Z is a bit more powerful than the A13 I noted in the A13 in the 9th Generation iPad I mentioned above.
Why exactly do you think an A12Z SOC would have sufficient processing power to simulate page faults in software? How exactly would the software determine that a particular page was not in RAM? How would one implement that without crippling the speed of the system?
What blogs in any Apple technical blog discussed how one would implement software-only paging on an A12Z (or any A-series) chip? Has anyone anywhere ever talked about this topic? You're the first person I've ever witnessed claim that this was possible. Please provide facts to support your opinion.
You've got it backwards. If you think an A12Z can implement software-only page faults, you must imagine how the software would handle those page faults. How exactly would you detect a page fault, and how would you do those checks without crippling the performance?
If this were possible, then technical articles somewhere would have talked about it. Professors would have lecture about it in classes: "How to handle paging on hardware providing no support for page faults (and without crippling system performance)". Can you point to a single academic (or non-academic) who has ever addressed that issue anywhere?
You're just quibbling with semantics. Having a software function that stopped working because someone added an external monitor would be a complication. That's what I mean by "disappear".
Users would definitely be confused by functionality that disappeared when they added hardware. The obvious solution is to provide the functionality with SOCs that can provide Stage Manager on ALL hardware configurations.
What technical information? Where? All you've done is render an opinion that hardware support for paging is no big deal -- that it's somehow easy to provide such functionality in A-series chips. You failed to provide us any facts -- technical information at all how this would work.
Don't just repeat your opinion, please. Provide us with the facts. How would one do this?
Where has anyone anywhere provided a writeup of how one would handle page faults with NO hardware support. Nobody has ever covered that topic; it can't be done. If you think it can be done, explain how. Dazzle us with your explanation of the facts.
Provide the technical details, or admit that you have no idea how it can be done. One or the other, please.
I didn't say A-series page faults - I said a memory swapping system. Perhaps there's a famous quote about reading carefully somewhere.
Memory Management in Mac OS (2002)
Page 10, Virtual memory, implemented via double-indirection memory block pointers.
I was referring to all non-M1 iPad pro models. You're the one that came up with non-pro models.
See above. The original Mac OS implemented overlapping windows on a 68K using the method described in the link.
That's the problem when you don't know your computing history.
I didn't say page faults.
You've dedicated a wall of text based on something you didn't read carefully.
It wouldn't stop working - it just wouldn't extend to the external monitor.
Wouldn't disappear - it's just not available on the external monitor.
Done
More wall of text dedicated to a misinformed assumption, combined with a deficit of knowledge.
Then you should read what you have said in your postings here. The point of the discussion is your claim that A-series SOCs should be able to support Stage Manager. If so, you need to explain how those A-series chips would support virtual memory -- including page faults. How do you get decent performance from an A-series chip when it can't even detect page faults?
Insufficient. Show us where you think that's what is done on the iOS operating systems.
FloatingBones said:
What's the #!$$ difference? How exactly is an A13 chip running on an iPad doing something different than an A12Z chip running on an iPad Pro?
You failed to answer my question: what exact non-M1 iPad Pro models are you talking about? Please answer now.
What does this history have to do with what's happening today? If you're claiming that Stage Manager doesn't count on any special hardware in the M1 SOC to run at reasonable speeds, you're going to have to quote something a bit more recent than Memory Management in Mac OS (2002).
What recent document shows that none of the hardware in the M1 chip is necessary to run Stage Manager app-switching at reasonable speeds? You have failed to reference a single pertinent document.
FrloatingBones said:
That's just dumb. If someone buys an external monitor, they expect all functions of iPadOS to work on that monitor. Claiming it would be acceptable to have Stage Manager "not extend" to the monitor is just dumb.
Why would any normal user find that acceptable?
You might find the "wouldn't extend" failure acceptable, but a normal user would never find that acceptable.
Nope. A 2002 (i.e. pre-iPhone by 5 years; pre-M1 by 18 years) memo is wholly insufficient to describe anything at all about M1 functionality on iOS. This is about as silly as claiming that users would be happy that Stage Manager simply "wouldn't extend" to function on an external monitor. That's wholly nonsensical.
Extrapolating a 2002 memo to functionality of an operating system that didn't even exist until 2007 is a grossly misinformed assumption. It shows a gargantuan deficit of knowledge. Please do better.
Find some memos that are actually talking about the A-series chips that make your point -- some paper showing that A12Z functionality is sufficient to make Stage Manager run at acceptable speeds on an iPad (and M1 functionality is somehow superfluous). If it exists, such a memo could not possibly been published before 2022 WWDC.
Here's a hint: no such memo exists.
I provided a detailed API link showing how virtual memory swapping can be implemented without hardware support in the MMU, in a prior Apple OS no less, directly contradicting every assertion in your nonsensical walls-of-text. I wasn't expecting you to admit you were wrong and I wasn't disappointed.
You categorically failed to explain how a 2002 memo has anything to do with an OS that didn't even exist until 2007, and hardware that didn't implement that OS until 2020. Extrapolating a 2002 memo to functionality of an operating system that didn't even exist until 2007 is a grossly misinformed assumption. It shows a gargantuan deficit of knowledge.
It's almost as silly as claiming that users wouldn't mind if Stage Manager "is just not available on the external monitor" that they buy to run with their iPad. If the function doesn't work on all hardware configs, then it is fundamentally deficient. Apple would never do something that silly. This is something you can't explain, either.
You again failed to answer my question: what exact non-M1 iPad Pro models are you talking about?
If you can't address those three concerns, then you're not arguing in good faith.
Don't know if these have been posted but....
wccftech.com
www.gsmarena.com
iPadOS 16 Code Confirms Internal Mode That Allows Apple to Run Stage Manager on Older iPads
It has been recently discovered in iPadOS 16 internal code that Apple could enable Stage Manager on older iPad models without an M1 chip.
wccftech.com
iPad Air 5 64GB model will not include iPadOS 16 virtual memory swap feature
Will still have the Stage Manager feature. Apple recently announced two new features coming to the M1 iPads with the upcoming iPadOS 16 update. This...
Do you know if there are any VR Headset games with the fictional stories of Ironman or Pacific Rim?
No, you have a severe misunderstanding of what a ‘right’. You don’t have a right to anything that’s not own/possessed by you. It’s a service.
This was in response to somebody claiming everybody owning an iPad were ever happy with it. Not the case for me.
Of course Stage Manager doesn't change **** all about that. It's a **** device that only makes sense for people that draw on it or cannot touch type.
After we saw documents in the Epic vs Apple trial, showing that apple voluntarily "cripples" software for its older devices, I won't believe to any word from Federighi.
You're just greedy and trying to make us buy a newer Mac with these subtle features.
<3 I will proudly keep my 27" iMac Late 2013 with OCLP, with all features working.
And I'm pretty sure the same OCLP will let us use Stage Manager flawlessly.
You should just enjoy your money, Craig, and let us have what we deserve!
You're just greedy and trying to make us buy a newer Mac with these subtle features.
<3 I will proudly keep my 27" iMac Late 2013 with OCLP, with all features working.
And I'm pretty sure the same OCLP will let us use Stage Manager flawlessly.
You should just enjoy your money, Craig, and let us have what we deserve!