If Apple does make this jump, would it jump all its macOS hardware platforms at the same time, or would it run a trial balloon first on a single platform to gauge public interest? Which platform has the least number of users running Windows? Would that be the Mac Mini or maybe the MacBook Air? Or could the new ARM platform be a brand new Mac device, perhaps called the Mac Air? I would propose that the Mac Air drop any space for HDDs, drop most connectors, and even use AirPlay2 for its video output. With an ARM CPU it could also omit the fan, just like the iPhone does. Physically it would be smaller than an iPhone and would be powered by a USB-C or Lightning cable. Maybe in lieu of an iPhone "screen" it could be powered by a solar panel and that would be called "a wireless power source" in which case it would have no physical ports at all, not unlike the Apple Watch. Or it could be powered by a new wireless power system. Hey, I know 80% of my ideas are bad, but that's still not 100%.
I suspect Apple would want to move everything at once, or as close as possible. They don't want to be supporting multiple platforms. They don't want to get into a situation where you can run this software on an iMac, but not on a Mini or a Pro. It will IMO be all or nothing.
Who knows, maybe this is why Mac updates have lagged as much as they have. They've been concentrating on ARM versions.
I don't think so. There have been so many advancements in computing since the PPC days that there's both no reason to start at the higher-end, and no reason to move as quickly as possible to make both an entry-level ARM-based Mac notebook and a high-end ARM-based Mac Pro as close together as possible.
Starting at the lower-end will allow for ARM-based chips designed by Apple (which also use Apple GPUs) to allow for faster and less expensive Macs. These are allow the Macs that will have fewer VM and Bootcamp users, as well as fewer app suites which will take longer to port (although not nearly as long as it use to be, especially since Adobe and Microsoft have been heaving supporting and coding iOS for over a decade now.
While I look forward to seeing ARM-based Macs, I suspect that my 15" MBP will be one of the last models to get the change.
There is a big difference between transitioning between different third-party CPU architectures and transitioning to an architecture that is your own design. One thing that is often left out of this discussion is the integrated GPU architecture -- Apple would also be designing that. eGPU support would be essential.
Another component of the Intel transition that is often overlooked is the legal settlement with Microsoft for Office that gave Apple cover to build OS X. I think this article is correct that Apple is big enough now that developers (including Microsoft) will follow, with Xcode leading the way.
Swift programming language will make transitioning to ARM very easy since Swift is already being used on Raspberry Pi which is an Arm architecture. Plus ARM are based on RISC which is similar to the old PowerPC chips. RISC has shown and proven to be superior than Intel in efficiency and speed. So the move will be huge and easy for software developers.
Steve Jobs stated at his annual talks that he had working prototype Macbooks running on ARM and that "it's not that bad". And he was obviously alive then. I think Tim Cook might have referenced ARM based Macbooks as well, but I can't remember for sure. You just need to macrotize your assembly language routines, but that's been easier and easier lately, because we've recently had to upgrade our code to AVX512.
People who complain about Windows emulation are forgetting that Microsoft ALSO wants to port Windows to ARM. At which point you'll need ARM to do high performing Windows emulation or boot camp.
I hope Apple doesn't transition Macs to ARM chips. The benefit of having a POSIX *n*x running on the same hardware as the rest of the world is hard to overstate. The thinking with the transition is that since ARM chips are so powerful sipping such little energy on iOS devices, imagine the workhorses they'd be on desktops? Sure? Maybe?
But this would only be a short-lived advantage until the same physical obstacles affecting Intel come up. The reason to transition is that progression on the Intel architecture has decelerated. But this is universal and will affect the ARM architecture as well. The laws of physics won't give Apple's ARM engineers any advantages over Intel engineers. ARM may have a head start, but it WILL hit the same limits at 4nm process with yield problems, etc.
Apple is always trying to bring things in house. Windows isn’t a threat anymore. Intel can’t produce fast enough. ARM kicks ass and I’ve been holding out for an ARM iMac.
Apple is best at computers and programming. I hope they realize that and make great computers again with software that has no match. I know they can do it. I think the transition to arm is extremely easy (this time) because of high standardization: the APIs are highly standard currently and hidden unofficial APIs are discouraged for a long time, with the app stores I think, killing its use off altogether. Multicore APIs and Metal for GPUs makes the software transition in most cases just a push on the ‘compile’ button. I think its also essential to be compatible on the hardware level, so thunderbolt and same fast (ultra fast) RAM busses and of course USB3/C etc. This is the most difficult part of the new ARM SoC, but since A chips have most already this is probably no problem at all.
I hope Apple doesn't transition Macs to ARM chips. The benefit of having a POSIX *n*x running on the same hardware as the rest of the world is hard to overstate. The thinking with the transition is that since ARM chips are so powerful sipping such little energy on iOS devices, imagine the workhorses they'd be on desktops? Sure? Maybe?
But this would only be a short-lived advantage until the same physical obstacles affecting Intel come up. The reason to transition is that progression on the Intel architecture has decelerated. But this is universal and will affect the ARM architecture as well. The laws of physics won't give Apple's ARM engineers any advantages over Intel engineers. ARM may have a head start, but it WILL hit the same limits at 4nm process with yield problems, etc.
Apple isn't moving to ARM to avoid physical obstacles of CPU design. Apple is doing this to manage processor costs and to control the processor development path. Intel has a terrible record of delivering against their own processor roadmap for some time now. This has been a significant problem for Apple, who doesn't just refresh their Mac lineups with the latest and greatest chips every x months the way some hardware manufactures do. By taking processor control in house Apple will have the same type of advantage they have today with iPhones where they control the timeline and feature set so it is much easier to release on a schedule. Using their own ARM chips will also increase Apples ability to differentiate their products from the competition.
I hope Apple doesn't transition Macs to ARM chips. The benefit of having a POSIX *n*x running on the same hardware as the rest of the world is hard to overstate. The thinking with the transition is that since ARM chips are so powerful sipping such little energy on iOS devices, imagine the workhorses they'd be on desktops? Sure? Maybe?
But this would only be a short-lived advantage until the same physical obstacles affecting Intel come up. The reason to transition is that progression on the Intel architecture has decelerated. But this is universal and will affect the ARM architecture as well. The laws of physics won't give Apple's ARM engineers any advantages over Intel engineers. ARM may have a head start, but it WILL hit the same limits at 4nm process with yield problems, etc.
POSIX is a software standard and so has nothing to do with the hardware. macOS is POSIX compliant but - as I found out myself - this is a very liberal statement. Some parts of libraries can be omitted for example (only the headers must exist but its implementation can be nill) which makes porting of (for example) multithreading POSIX code to the Mac a very difficult exercise. Benchmarks show that Apples A(RM) chips are extremely powerful, its no gamble ‘no sure, maybe’. Every advantage is short-lived, but the advantage is exactly the point. If we wait 1000 years non of it matters, but a two or three year advantage makes the difference selling things or not. Sure the laws of physics apply for A chips, but Intel chips have an inherent disadvantage running a RISC layer in hardware. This means that they will never be as power efficient or fast as the the ultimate ARM chip.
If Apple does make this jump, would it jump all its macOS hardware platforms at the same time, or would it run a trial balloon first on a single platform to gauge public interest? Which platform has the least number of users running Windows? Would that be the Mac Mini or maybe the MacBook Air? Or could the new ARM platform be a brand new Mac device, perhaps called the Mac Air? I would propose that the Mac Air drop any space for HDDs, drop most connectors, and even use AirPlay2 for its video output. With an ARM CPU it could also omit the fan, just like the iPhone does. Physically it would be smaller than an iPhone and would be powered by a USB-C or Lightning cable. Maybe in lieu of an iPhone "screen" it could be powered by a solar panel and that would be called "a wireless power source" in which case it would have no physical ports at all, not unlike the Apple Watch. Or it could be powered by a new wireless power system. Hey, I know 80% of my ideas are bad, but that's still not 100%.
I suspect it will move from the bottom, up, and migrate to the high-end when the market demands.
I agree that makes the most sense. That said I kind of wonder of the new Mac Pro might not sport some kind of hybrid CPU set up. Somehow Apple needs to get a lot of pro software converted and developers could use such a machine for that purpose and pros could beta test it on it.
If Apple does make this jump, would it jump all its macOS hardware platforms at the same time, or would it run a trial balloon first on a single platform to gauge public interest? Which platform has the least number of users running Windows? Would that be the Mac Mini or maybe the MacBook Air? Or could the new ARM platform be a brand new Mac device, perhaps called the Mac Air? I would propose that the Mac Air drop any space for HDDs, drop most connectors, and even use AirPlay2 for its video output. With an ARM CPU it could also omit the fan, just like the iPhone does. Physically it would be smaller than an iPhone and would be powered by a USB-C or Lightning cable. Maybe in lieu of an iPhone "screen" it could be powered by a solar panel and that would be called "a wireless power source" in which case it would have no physical ports at all, not unlike the Apple Watch. Or it could be powered by a new wireless power system. Hey, I know 80% of my ideas are bad, but that's still not 100%.
I suspect Apple would want to move everything at once, or as close as possible. They don't want to be supporting multiple platforms. They don't want to get into a situation where you can run this software on an iMac, but not on a Mini or a Pro. It will IMO be all or nothing.
Who knows, maybe this is why Mac updates have lagged as much as they have. They've been concentrating on ARM versions.
I can see your logic but everything all at once and almost no software other than Apple's own at best would seem a bit risky. Something needs to lead the way as a machine for the software developers I think (see my last post above). Unless Apple has already got some sort of Intel emulation running very fast under ARM ...?
Apple will need to keep something to ensure compatibility for Windows based applications.
They are already have the same problem they had with Power PC. Intel has lagged Moore's Law for at least half a decade now. Monolithic CPU's are not going to push the bar forward at anything like the pace they had back in the 1990's. Memory has been the leading edge of new process technology because it is not as complex. AMD has stolen the march on Intel in this next phase. I am not saying Intel is not about to respond. It is just obvious that the who computing system needs an overhaul.
My own take for where this is going:
Apple will replace intel with it's own core (90% likely).
There will be an edge system for web based intel technology similar to the system being built by Google for their online gaming platform that will allow cloud based support for x86. (50%)
Apple is waiting on 5G to deploy this system which will use AI to improve latency. (80%)
Security will be handled by a separate chiplet or a dedicated part of the ARM core. (95%)
AMD has a shot at producing what ever Apple uses for their replacement for Intel. (50% or less)
Privacy will be their core product and it will not come cheaply.
If Apple does make this jump, would it jump all its macOS hardware platforms at the same time, or would it run a trial balloon first on a single platform to gauge public interest? Which platform has the least number of users running Windows? Would that be the Mac Mini or maybe the MacBook Air? Or could the new ARM platform be a brand new Mac device, perhaps called the Mac Air? I would propose that the Mac Air drop any space for HDDs, drop most connectors, and even use AirPlay2 for its video output. With an ARM CPU it could also omit the fan, just like the iPhone does. Physically it would be smaller than an iPhone and would be powered by a USB-C or Lightning cable. Maybe in lieu of an iPhone "screen" it could be powered by a solar panel and that would be called "a wireless power source" in which case it would have no physical ports at all, not unlike the Apple Watch. Or it could be powered by a new wireless power system. Hey, I know 80% of my ideas are bad, but that's still not 100%.
I suspect it will move from the bottom, up, and migrate to the high-end when the market demands.
Wow, I can't recall the last time an AI staff member responded to me (and kinda agreed). I feel honoured. My question might have been provocative, because seconds after Mike Wuerthele responded, DAalseth responded with what appears to be an opposite opinion. I think someone else opined that maybe the new Mac Pro would include (a module for?) an ARM processor. I'm inclined to concur with that. Perhaps the ARM module for the Mac Pro will include an A7, A8, A9,... etc so that developers can test their code on each ARM model. Pure speculation but not irrational.
What a strange article. Somewhat of revisionist history to suggest that Apple's architecture moves were made because Apple was on the bleeding edge and needed new processors to push the limits. In reality their reliance on technology that in the end, just couldn't keep up with Intel. x86 has been around since the late 70s, ARM since the mid-80s. I don't know how moving from one old technology to another old technology can be considered a 10 year lifespan. In 1988 there is no way Steve Jobs envisioned the current computing environment. Just take a look at what the top 500 super computers are running...
“When it does so, Apple will be making an enormous change that will definitely require extensive work by software developers. They will have to convert their apps to run the new ARM processors and in some cases that's going to take significant effort.”
For anybody who isn’t using assembly and is using Xcode ( c,c++,swift or objective C using the standard libraries) it’s a recompile. Today when you build for iOS you can run in a simulator which is an intel x64 version of your app. It’s not emulated, it’s basically a Mac app. The same process would create ARM binaries for a Mac application. If people have to write #if ARM it #if INTEL in any high level code then Apple will have failed.
For devs who have already uploaded using the bitcode option they have nothing to do. Expect Apple to enforce this option for Mac apps pretty soon and then you can really start to speculate on ARM.
If Apple closes the door on Intel, AMD and NVIDIA - (boy did they close the door on NVIDIA!!), it might be difficult to open that door again.
By moving everything (including GPU) in-house, Apple solve most of their supply woes...
...but at what cost?
Apple have only just completed adding eGPU support it would seem odd to even bother if they are just about to change hardware that would then rule that out. Seems to me Apple wants to make it all more flexible so they can be self reliance.
"Mac shifting to ARM may come as soon as ..." ... ore more likely, it may not. It's a dumb idea that breaks the Windows compatibility that has been Apple's hook into bringing people over from the dark side, and will involve huge amounts of angst for everyone, with no 'noticeable' difference (other than your software won't work). By noticeable, it needs to run 40 times faster for anyone to care.
Comments
Toasted
Starting at the lower-end will allow for ARM-based chips designed by Apple (which also use Apple GPUs) to allow for faster and less expensive Macs. These are allow the Macs that will have fewer VM and Bootcamp users, as well as fewer app suites which will take longer to port (although not nearly as long as it use to be, especially since Adobe and Microsoft have been heaving supporting and coding iOS for over a decade now.
While I look forward to seeing ARM-based Macs, I suspect that my 15" MBP will be one of the last models to get the change.
Another component of the Intel transition that is often overlooked is the legal settlement with Microsoft for Office that gave Apple cover to build OS X. I think this article is correct that Apple is big enough now that developers (including Microsoft) will follow, with Xcode leading the way.
People who complain about Windows emulation are forgetting that Microsoft ALSO wants to port Windows to ARM. At which point you'll need ARM to do high performing Windows emulation or boot camp.
Apple is best at computers and programming.
I hope they realize that and make great computers again with software that has no match.
I know they can do it.
I think the transition to arm is extremely easy (this time) because of high standardization: the APIs are highly standard currently and hidden unofficial APIs are discouraged for a long time, with the app stores I think, killing its use off altogether. Multicore APIs and Metal for GPUs makes the software transition in most cases just a push on the ‘compile’ button.
I think its also essential to be compatible on the hardware level, so thunderbolt and same fast (ultra fast) RAM busses and of course USB3/C etc. This is the most difficult part of the new ARM SoC, but since A chips have most already this is probably no problem at all.
Benchmarks show that Apples A(RM) chips are extremely powerful, its no gamble ‘no sure, maybe’.
Every advantage is short-lived, but the advantage is exactly the point.
If we wait 1000 years non of it matters, but a two or three year advantage makes the difference selling things or not.
Sure the laws of physics apply for A chips, but Intel chips have an inherent disadvantage running a RISC layer in hardware. This means that they will never be as power efficient or fast as the the ultimate ARM chip.
They are already have the same problem they had with Power PC. Intel has lagged
Moore's Law for at least half a decade now. Monolithic CPU's are not going to push
the bar forward at anything like the pace they had back in the 1990's. Memory has been
the leading edge of new process technology because it is not as complex. AMD has stolen the
march on Intel in this next phase. I am not saying Intel is not about to respond. It is just obvious
that the who computing system needs an overhaul.
My own take for where this is going:
Apple will replace intel with it's own core (90% likely).
There will be an edge system for web based intel
technology similar to the system being built by Google
for their online gaming platform that will allow cloud based support
for x86. (50%)
Apple is waiting on 5G to deploy this system which will use
AI to improve latency. (80%)
Security will be handled by a separate chiplet or a dedicated part of
the ARM core. (95%)
AMD has a shot at producing what ever Apple uses
for their replacement for Intel. (50% or less)
Privacy will be their core product and it will not come cheaply.
By moving everything (including GPU) in-house, Apple solve most of their supply woes...
...but at what cost?
For anybody who isn’t using assembly and is using Xcode ( c,c++,swift or objective C using the standard libraries) it’s a recompile. Today when you build for iOS you can run in a simulator which is an intel x64 version of your app. It’s not emulated, it’s basically a Mac app. The same process would create ARM binaries for a Mac application. If people have to write #if ARM it #if INTEL in any high level code then Apple will have failed.
For devs who have already uploaded using the bitcode option they have nothing to do. Expect Apple to enforce this option for Mac apps pretty soon and then you can really start to speculate on ARM.
https://www.quora.com/What-is-Apple-Bitcode