There would probably be no reason to do an ISA from scratch unless there has been some fundamental academic advance in how to make ISAs that is not implemented in ARM. Or maybe all the hardware security flaws lately could be another reason?
But more likely, you just get different versions of the ARM ISA (we are up to version 8 already) and through the App Store, Apple can force developers to compile to the newest instruction sets (you often see them setting these App Store developer deadlines for things), and then once the old instructions are no longer being used that part of the silicon can be removed or repurposed. Controlling the App Store giving their hardware team an advantage... there sure are strong arguments for vertical integration aren't there?
There would probably be no reason to do an ISA from scratch unless there has been some fundamental academic advance in how to make ISAs that is not implemented in ARM. Or maybe all the hardware security flaws lately could be another reason?
But more likely, you just get different versions of the ARM ISA (we are up to version 8 already) and through the App Store, Apple can force developers to compile to the newest instruction sets (you often see them setting these App Store developer deadlines for things), and then once the old instructions are no longer being used that part of the silicon can be removed or repurposed. Controlling the App Store giving their hardware team an advantage... there sure are strong arguments for vertical integration aren't there?
1) I don't see a need for Apple to create their own ISA, either, but I also can't rule it out. Designing their own ARM chips that eschews features that Apple doesn't need seems like more than enough for Apple to idealize their own chips.
2) Currently, it looks like Apple is using ARMv8.2-A for the Apple A11 with the Monsoon core. No idea what their other chips are using or even if that data is accurate since other sources simply refer to ARMv8-A with the revision marker.
Though they could also just choose to keep calling this new version macOS and simply state that in version 10._ they're breaking compatibility and dropping a ton of APIs and requiring developers to use the new ones.
I would imagine that's the point that you start talking about macOS 11, not 10.x
Though I agree with you, and think there would be good name recognition from a common OS name, especially if this supposed new product line comes under a new brand.
Mac, with OS iPhone, with OS iPad, with OS Tv, with OS "Star", with OS
That's a good consumer sell, even if it loses a bit of clarity for developers. I think Apple cares more about the consumer.
Don't quit your day jobs, guys, really... Nope, Apple is not going to switch everything over to "Apple OS", nor is that a good consumer sell.
I think that’s what the developer should see, the consumer should see device-specific UIs.
Though they could also just choose to keep calling this new version macOS and simply state that in version 10._ they're breaking compatibility and dropping a ton of APIs and requiring developers to use the new ones.
I would imagine that's the point that you start talking about macOS 11, not 10.x
Though I agree with you, and think there would be good name recognition from a common OS name, especially if this supposed new product line comes under a new brand.
Mac, with OS iPhone, with OS iPad, with OS Tv, with OS "Star", with OS
That's a good consumer sell, even if it loses a bit of clarity for developers. I think Apple cares more about the consumer.
Don't quit your day jobs, guys, really... Nope, Apple is not going to switch everything over to "Apple OS", nor is that a good consumer sell.
I think that’s what the developer should see, the consumer should see device-specific UIs.
Developers aren't going to see less relevant data than customers.
There would probably be no reason to do an ISA from scratch unless there has been some fundamental academic advance in how to make ISAs that is not implemented in ARM. Or maybe all the hardware security flaws lately could be another reason?
But more likely, you just get different versions of the ARM ISA (we are up to version 8 already) and through the App Store, Apple can force developers to compile to the newest instruction sets (you often see them setting these App Store developer deadlines for things), and then once the old instructions are no longer being used that part of the silicon can be removed or repurposed. Controlling the App Store giving their hardware team an advantage... there sure are strong arguments for vertical integration aren't there?
Could they build core that consume LLVM's IR directly? Isn't that what they store already as the AppStore Universal Binary? What instruction set do the GPU cores, image processor and "Bionic" parts of the A11 use?
I'm wondering would it make sense that the complier team would want to build a native processor instead of getting to an ideal state then converting to someone elses processor.
Apple doesn't need a new device family to make an iPad work as a laptop. Nor does it need a new device family to make a MacBook work as an iPad....
... To one you add a cursor and you're there. To the other you add a touchscreen that flips and you're there... Neither is a big deal.
Apple fans make too much of the split of between iOS devices and Mac devices.
But, Apple has never been hobbled by respecting artificial boundaries. Instead, they focus on producing great devices that make a people's lives better -- whatever that may be...
But, one thought: "A SIM card tray"? Isn't that kind of old fashioned? Hopefully we can soon move on from such silliness...
There would probably be no reason to do an ISA from scratch unless there has been some fundamental academic advance in how to make ISAs that is not implemented in ARM. Or maybe all the hardware security flaws lately could be another reason?
But more likely, you just get different versions of the ARM ISA (we are up to version 8 already) and through the App Store, Apple can force developers to compile to the newest instruction sets (you often see them setting these App Store developer deadlines for things), and then once the old instructions are no longer being used that part of the silicon can be removed or repurposed. Controlling the App Store giving their hardware team an advantage... there sure are strong arguments for vertical integration aren't there?
Could they build core that consume LLVM's IR directly? Isn't that what they store already as the AppStore Universal Binary? What instruction set do the GPU cores, image processor and "Bionic" parts of the A11 use?
I'm wondering would it make sense that the complier team would want to build a native processor instead of getting to an ideal state then converting to someone elses processor.
Maybe. But I don't really think of LLVM IR as being an idealised form, and compiling to ARM to be a compromise. It's more just transitioning your program from one model to another: from a model designed for portability to one designed for execution. These are different requirements so its not necessarily true that trying to minimise the gap between the two models is optimal.
But its the kind of thing where you'd just have to do a lot of experiments and see what happens to know for sure. If you want to design an LLVM-near CPU I'm sure you could do it. All the necessary documentation for such things is on the web these days, you don't have to be a big corporation, you just have to like reading and be persistent.
Though they could also just choose to keep calling this new version macOS and simply state that in version 10._ they're breaking compatibility and dropping a ton of APIs and requiring developers to use the new ones.
I would imagine that's the point that you start talking about macOS 11, not 10.x
Though I agree with you, and think there would be good name recognition from a common OS name, especially if this supposed new product line comes under a new brand.
Mac, with OS iPhone, with OS iPad, with OS Tv, with OS "Star", with OS
That's a good consumer sell, even if it loses a bit of clarity for developers. I think Apple cares more about the consumer.
Don't quit your day jobs, guys, really... Nope, Apple is not going to switch everything over to "Apple OS", nor is that a good consumer sell.
I didn't say they would, or should, though I think they could. You're going to have to explain to me how having four different brandings for their operating systems leads to better name recognition for consumers than having one.
Though they could also just choose to keep calling this new version macOS and simply state that in version 10._ they're breaking compatibility and dropping a ton of APIs and requiring developers to use the new ones.
I would imagine that's the point that you start talking about macOS 11, not 10.x
Though I agree with you, and think there would be good name recognition from a common OS name, especially if this supposed new product line comes under a new brand.
Mac, with OS iPhone, with OS iPad, with OS Tv, with OS "Star", with OS
That's a good consumer sell, even if it loses a bit of clarity for developers. I think Apple cares more about the consumer.
Don't quit your day jobs, guys, really... Nope, Apple is not going to switch everything over to "Apple OS", nor is that a good consumer sell.
I didn't say they would, or should, though I think they could. You're going to have to explain to me how having four different brandings for their operating systems leads to better name recognition for consumers than having one.
Because consumers don't give a shit about the name of the OS. That's it. As a techie enthusiast you're waaay over thinking this. Normals buy an iPhone, an iPad, an iMac, an Apple Watch... they're not going to spend any clock cycles on what the name of its OS is. The OS is relevant to devs. But ask my dad what OS his Apple TV is running and why he should care and he won't have any idea what you're talking about.
This is exciting news because if Apple wants to retire the Mac I would rather move to an iOS notebook than to Windows.
Also, when Apple retired the Airport a few weeks ago, I and others said it might be because they were moving away from WiFi and focussing on 5G cellular (where the puck is going). And now we have a rumor of a notebook with cellular (joining the phone, ipad and watch).
I wouldn't bet on Apple getting rid of Mac or getting rid of WiFi.
While im not thrilled about another monthly expense Apple should have offered cellular support in its laptops years ago. Cell support in the iPad gives it huge advantages for travel.
The other issue with WiFi, when you actually find an open connection, is the terrible performance. I actually have turned Wifi off on (at times) my iPhone due to some locations offering terrible WiFi performance.
The real problem here is the rumors about an IOS derived system. As long as we have the critical issues with file handling i dont see iOS replacing Mac OS. IOS can drive one batty with its hanstrung file management.
Yeah I can see people getting their hopes up over this...especially those that think Apple should work on an iOS based laptop with a keyboard and mouse.
There's a lot to unpack with the following paragraph..
Prototypes in circulation boot using an Extensible Firmware Interface (EFI) like Apple's current Intel-based Macs, which the report suggests points to an ARM-powered MacBook convertible. However, the machines are said to run a "derivative" of iOS, not macOS.
I can see a lot of speculation with little to no actual factual data going into creating that conclusion.
But let's say Apple does have a great hybrid in mind, would it use iOS or macOS as its core? Hopefully people are past the simplistic notion that ARM = iOS and Intel = macOS, but I doubt I could be so lucky. If Apple were to make a hybrid device that can function as both a tablet and as a laptop or docked desktop, how would they do it? Would it be better to start with iOS and then build up the HW/drivers and frameworks needed to then support the new I/O needs for also making it like a Mac? Can macOS be ported over to iOS? Would they even want to, or would they want to use this product as a way to ushering in the future of macOS by creating a very efficient, Swift-based desktop UI for this rumoured "Star."
It wouldn't too off base with how they've brought more efficient code back to the Mac after finding ways to make it more efficient for the iPhone. It's also not like Apple doesn't have a history of multiple UIs being stored on a device, but this would be a completely different undertaking than including CarPlay on an iPhone. At this point, I'm expecting to see an ARM-based Mac before I see an Apple tablet running the iOS UI that will convert to a desktop OS when used in a certain way or plugged into certain HW.
"But let's say Apple does have a great hybrid in mind, would it use iOS or macOS as its core?"
People seem to forget that iOS and MacOS are in fact the samething once you get below the GUI!!! You still have the Unix core and many libraries are the same exact thing.
GUI's being what they are can eassily motph into something else. In my mind Apple will have to move farther away from iOS that it will Mac OS for this new solution. Such a platform needs robust muktitasking file management and other things missing from iOS.
I still belieeve Apple has the goal of a Knowledge Navigator type device so what may bemost interesting here is AI features built in. That is a "local" Siri if you will. Local in the sense that things that can be done on the device are while the cloud is engaged automatically
The thing that’s baking my noodle is; with so much obsolescence, should Apple bother with ARM ISA for macOS at all or just cut to their own ISA with Swift runtime?
That would depend on whether Apple is using ARM's instruction set or not. As far as I know they are. What have you heard about Apple developing their own ISA that is more efficient? Can swift even be used to write such low-level code and have it be better than what ARM offers?
Nothing but total speculation on my part. Their outperformance with ARM-compatible processors hints at something under the hood beyond the likes of Qualcomm & Samsung and my guess is the PA/Intrinsity acquisitions were about building micro-architecture IP. They also have a lot of silicon beyond that covered by ARM ISA (ISP, GPU, Neural). The ISA is the only layer of the stack they don’t own (A-series) and flattening that stack is the key to power/performance efficiency. If they wanted to, they could add web rendering, image/video/audio signal silicon accessible via iOS frameworks in Swift only and do what they want in between. Basically, ditch what we know as a CPU, move to a real APU (not AMDs version) and ‘emulate’ legacy ISAs as needed; appleOS on appleISA.
You are close here! From my perspective there is little advantage to develoing an entirely new CPU ISA, extensions though are a different matter. So the CPU will stay ARM for a very long time. Where Apple will see value is in more and more specialized hardware. Today we see that in image processing, the GPU and most importantly hardware to accelerate some AI calculations. Todays AI hardware is actually trivial and is only the beginning whan it comes to adding AI capabilities to our machines.
As for Swift i can see Apple rewriting most of the GUI code and a good portion of lower level system code in Swift. There would be a huge payoff in the long run with more secure and robust software. Basically they could kill off a massive number of software issues related to C/C++/ObjectiveC plus they woukd streamline a ton of legacy apps with new Swift code based.
So yeah one has to believe Swift is a major part of the project. If there is any hardware suppport it will be minor additions to the ARM 64 bit hardware ISA. I would expect however to see many new Swift libraries that leverage AI hardware that Apple has started to debut. We might see a step that is as big as the step from DOS to the first GUIs, this due to the way AI will have been worked into software.
There would probably be no reason to do an ISA from scratch unless there has been some fundamental academic advance in how to make ISAs that is not implemented in ARM. Or maybe all the hardware security flaws lately could be another reason?
But more likely, you just get different versions of the ARM ISA (we are up to version 8 already) and through the App Store, Apple can force developers to compile to the newest instruction sets (you often see them setting these App Store developer deadlines for things), and then once the old instructions are no longer being used that part of the silicon can be removed or repurposed. Controlling the App Store giving their hardware team an advantage... there sure are strong arguments for vertical integration aren't there?
1) I don't see a need for Apple to create their own ISA, either, but I also can't rule it out. Designing their own ARM chips that eschews features that Apple doesn't need seems like more than enough for Apple to idealize their own chips.
2) Currently, it looks like Apple is using ARMv8.2-A for the Apple A11 with the Monsoon core. No idea what their other chips are using or even if that data is accurate since other sources simply refer to ARMv8-A with the revision marker.
As for #1 above i can see Apple removing all support for ARM 32, bit ISA's from their chips. In fact the writing seem to be on the wall already. There are significant advantages for Apple in due space and power usage. Plus you reduce chip complexity making it faster.
As for #2 i see Apple keeping compatibilty with ARMs ISA definitions simply because the advantages significantly out weigh the disadvantages. In fact i woukd expect Apple to work with ARM if they needed a special instruction. Apple has leveraged special hardware outside of the CPU for a long time now i dont expect massive changes there.
This is exciting news because if Apple wants to retire the Mac I would rather move to an iOS notebook than to Windows.
Also, when Apple retired the Airport a few weeks ago, I and others said it might be because they were moving away from WiFi and focussing on 5G cellular (where the puck is going). And now we have a rumor of a notebook with cellular (joining the phone, ipad and watch).
I wouldn't bet on Apple getting rid of Mac or getting rid of WiFi.
While im not thrilled about another monthly expense Apple should have offered cellular support in its laptops years ago. Cell support in the iPad gives it huge advantages for travel.
The other issue with WiFi, when you actually find an open connection, is the terrible performance. I actually have turned Wifi off on (at times) my iPhone due to some locations offering terrible WiFi performance.
I would've loved a MacBook Pro/PowerBook with built-in cellular years ago, but I don’t think Apple should’ve done it because I wanted it. I hated the horribly designed 3G dongle I purchased from Cinuglar (I assume) that stuck out of the USB-A port, but it got the job done.
Because it took up so much space on the side I added a short USB cable so I could have use of the other ports when it was plugged in and ended up velcroing the cellular dongle to the lid, which also worked fine, but was unattractive.
Since Apple made it simple to choose iPhone tethering from the WiFi drop down (I think it can even be chosen automatically over other WiFi connections) I’ve had no use for that. Even if it was just $10/month for the carrier to to add a device, on top of the extra device cost, I doubt I’d pony up th extra money whole I still had an iPhone.
The Apple Watch would have to be so vastly improved that I could forego having an iPhone that a cellular MBP would make sense again, but at this point they don’t even let you use an iPad to setup and manage your Watch.
Yeah I can see people getting their hopes up over this...especially those that think Apple should work on an iOS based laptop with a keyboard and mouse.
There's a lot to unpack with the following paragraph..
Prototypes in circulation boot using an Extensible Firmware Interface (EFI) like Apple's current Intel-based Macs, which the report suggests points to an ARM-powered MacBook convertible. However, the machines are said to run a "derivative" of iOS, not macOS.
I can see a lot of speculation with little to no actual factual data going into creating that conclusion.
But let's say Apple does have a great hybrid in mind, would it use iOS or macOS as its core? Hopefully people are past the simplistic notion that ARM = iOS and Intel = macOS, but I doubt I could be so lucky. If Apple were to make a hybrid device that can function as both a tablet and as a laptop or docked desktop, how would they do it? Would it be better to start with iOS and then build up the HW/drivers and frameworks needed to then support the new I/O needs for also making it like a Mac? Can macOS be ported over to iOS? Would they even want to, or would they want to use this product as a way to ushering in the future of macOS by creating a very efficient, Swift-based desktop UI for this rumoured "Star."
It wouldn't too off base with how they've brought more efficient code back to the Mac after finding ways to make it more efficient for the iPhone. It's also not like Apple doesn't have a history of multiple UIs being stored on a device, but this would be a completely different undertaking than including CarPlay on an iPhone. At this point, I'm expecting to see an ARM-based Mac before I see an Apple tablet running the iOS UI that will convert to a desktop OS when used in a certain way or plugged into certain HW.
"But let's say Apple does have a great hybrid in mind, would it use iOS or macOS as its core?"
People seem to forget that iOS and MacOS are in fact the samething once you get below the GUI!!! You still have the Unix core and many libraries are the same exact thing.
GUI's being what they are can eassily motph into something else. In my mind Apple will have to move farther away from iOS that it will Mac OS for this new solution. Such a platform needs robust muktitasking file management and other things missing from iOS.
I still belieeve Apple has the goal of a Knowledge Navigator type device so what may bemost interesting here is AI features built in. That is a "local" Siri if you will. Local in the sense that things that can be done on the device are while the cloud is engaged automatically
"In my mind Apple will have to move farther away from iOS that it will Mac OS for this new solution. Such a platform needs robust muktitasking file management and other things missing from iOS."
Sounds a lot like macOS but without Carbon, Appkit replaced by UIKit, and support for only App Store apps.
There would probably be no reason to do an ISA from scratch unless there has been some fundamental academic advance in how to make ISAs that is not implemented in ARM. Or maybe all the hardware security flaws lately could be another reason?
But more likely, you just get different versions of the ARM ISA (we are up to version 8 already) and through the App Store, Apple can force developers to compile to the newest instruction sets (you often see them setting these App Store developer deadlines for things), and then once the old instructions are no longer being used that part of the silicon can be removed or repurposed. Controlling the App Store giving their hardware team an advantage... there sure are strong arguments for vertical integration aren't there?
1) I don't see a need for Apple to create their own ISA, either, but I also can't rule it out. Designing their own ARM chips that eschews features that Apple doesn't need seems like more than enough for Apple to idealize their own chips.
2) Currently, it looks like Apple is using ARMv8.2-A for the Apple A11 with the Monsoon core. No idea what their other chips are using or even if that data is accurate since other sources simply refer to ARMv8-A with the revision marker.
As for #1 above i can see Apple removing all support for ARM 32, bit ISA's from their chips. In fact the writing seem to be on the wall already. There are significant advantages for Apple in due space and power usage. Plus you reduce chip complexity making it faster.
When they stopped supporting 32-bit I would’ve assumed they’d ax ot for all future Apple chips.
Though they could also just choose to keep calling this new version macOS and simply state that in version 10._ they're breaking compatibility and dropping a ton of APIs and requiring developers to use the new ones.
I would imagine that's the point that you start talking about macOS 11, not 10.x
Though I agree with you, and think there would be good name recognition from a common OS name, especially if this supposed new product line comes under a new brand.
Mac, with OS iPhone, with OS iPad, with OS Tv, with OS "Star", with OS
That's a good consumer sell, even if it loses a bit of clarity for developers. I think Apple cares more about the consumer.
Don't quit your day jobs, guys, really... Nope, Apple is not going to switch everything over to "Apple OS", nor is that a good consumer sell.
I think that’s what the developer should see, the consumer should see device-specific UIs.
Developers aren't going to see less relevant data than customers.
I meant consumers should see device-specific, branded OSs. Developers should see one OS with multiple UI frameworks/target devices, that way the App Store can do the device-specific deployment.
Though they could also just choose to keep calling this new version macOS and simply state that in version 10._ they're breaking compatibility and dropping a ton of APIs and requiring developers to use the new ones.
I would imagine that's the point that you start talking about macOS 11, not 10.x
Though I agree with you, and think there would be good name recognition from a common OS name, especially if this supposed new product line comes under a new brand.
Mac, with OS iPhone, with OS iPad, with OS Tv, with OS "Star", with OS
That's a good consumer sell, even if it loses a bit of clarity for developers. I think Apple cares more about the consumer.
Don't quit your day jobs, guys, really... Nope, Apple is not going to switch everything over to "Apple OS", nor is that a good consumer sell.
I think that’s what the developer should see, the consumer should see device-specific UIs.
Developers aren't going to see less relevant data than customers.
I meant consumers should see device-specific, branded OSs. Developers should see one OS with multiple UI frameworks/target devices, that way the App Store can do the device-specific deployment.
Unless they go into About, I don't know that they do. Apple offers macOS for Macs, iOS for iDevices, watchOS for Watch, and tvOS for the Apple TV. I think it makes it nicer to market when talking about the benefits of their OSes for different device categories. Despite years of wishing Apple had a cohesive OS landscape they finally do.
PS: At least it's better than a half-dozen version of Windows for the same device category or calling everything Windows and trying to mimic the same UI that worked on the desktop onto a PDA that you need a stylus on a pressure-sensitive touchscreen to work the thing.
Comments
2) Currently, it looks like Apple is using ARMv8.2-A for the Apple A11 with the Monsoon core. No idea what their other chips are using or even if that data is accurate since other sources simply refer to ARMv8-A with the revision marker.
Isn't that what they store already as the AppStore Universal Binary?
What instruction set do the GPU cores, image processor and "Bionic" parts of the A11 use?
I'm wondering would it make sense that the complier team would want to build a native processor instead of getting to an ideal state then converting to someone elses processor.
The other issue with WiFi, when you actually find an open connection, is the terrible performance. I actually have turned Wifi off on (at times) my iPhone due to some locations offering terrible WiFi performance.
The real problem here is the rumors about an IOS derived system. As long as we have the critical issues with file handling i dont see iOS replacing Mac OS. IOS can drive one batty with its hanstrung file management.
GUI's being what they are can eassily motph into something else. In my mind Apple will have to move farther away from iOS that it will Mac OS for this new solution. Such a platform needs robust muktitasking file management and other things missing from iOS.
I still belieeve Apple has the goal of a Knowledge Navigator type device so what may bemost interesting here is AI features built in. That is a "local" Siri if you will. Local in the sense that things that can be done on the device are while the cloud is engaged automatically
As for Swift i can see Apple rewriting most of the GUI code and a good portion of lower level system code in Swift. There would be a huge payoff in the long run with more secure and robust software. Basically they could kill off a massive number of software issues related to C/C++/ObjectiveC plus they woukd streamline a ton of legacy apps with new Swift code based.
So yeah one has to believe Swift is a major part of the project. If there is any hardware suppport it will be minor additions to the ARM 64 bit hardware ISA. I would expect however to see many new Swift libraries that leverage AI hardware that Apple has started to debut. We might see a step that is as big as the step from DOS to the first GUIs, this due to the way AI will have been worked into software.
As for #2 i see Apple keeping compatibilty with ARMs ISA definitions simply because the advantages significantly out weigh the disadvantages. In fact i woukd expect Apple to work with ARM if they needed a special instruction. Apple has leveraged special hardware outside of the CPU for a long time now i dont expect massive changes there.
should’ve done it because I wanted it. I hated the horribly designed 3G dongle I purchased from Cinuglar (I assume) that stuck out of the USB-A port, but it got the job done.
Because it took up so much space on the side I added a short USB cable so I could have use of the other ports when it was plugged in and ended up velcroing the cellular dongle to the lid, which also worked fine, but was unattractive.
Since Apple made it simple to choose iPhone tethering from the WiFi drop down (I think it can even be chosen automatically over other WiFi connections) I’ve had no use for that. Even if it was just $10/month for the carrier to to add a device, on top of the extra device cost, I doubt I’d pony up th extra money whole I still had an iPhone.
The Apple Watch would have to be so vastly improved that I could forego having an iPhone that a cellular MBP would make sense again, but at this point they don’t even let you use an iPad to setup and manage your Watch.
Sounds a lot like macOS but without Carbon, Appkit replaced by UIKit, and support for only App Store apps.
PS: At least it's better than a half-dozen version of Windows for the same device category or calling everything Windows and trying to mimic the same UI that worked on the desktop onto a PDA that you need a stylus on a pressure-sensitive touchscreen to work the thing.
Er no. That’s what MS did with Windows and what Samsung did with Android. Neither solution works well.