Apple has had all the common frameworks running in Ax chips since the beginning of Ax chips. Most executable app code is spent in these frameworks anyway: it may be apocryphal (but I don’t think so), but MS Excel in the PowerPC day is said to have spent 70% of its time in the Toolbox call DrawString(); since Apple had redone QuickDraw in PowerPC Native code, 70% of Excel ran natively with no work from Microsoft. That said, I’m not sure how well Ax chips will do with thermals if called upon to do x64 emulation.
I’m guessing Apple might give a huge lead time for developers and Marzipan, and release a Mac that did no emulation. But probably not.
What about software developers like Adobe? Fonts? Importing old files?
I lived through the Rosetta era and it was brutal. Brutal. Just brutal.
The hardware may have switched in a year, but the software nightmare was real for well over 3.
They will have to do better on the software side.
I think it's very different now. The OS (since X...built from NEXTSTEP) had this concept of "fat binaries" (called "universal" I think) which was basically just a click of a checkbox to have your app cross-compiled to multiple target CPUs. Pretty sure all that's still there. I would think this transition would be the smoothest yet.
That’s still there and devs can upload code to the App Store as bitcode for the last few years. That code doesn’t need a recompile. Most other code just needs a recompile.
Marizpan has nothing to do with this, the software is already abstracted from the hardware. All that marizpan does is allow mac and iOS apps to be ported in software easier (the appikit will probably merge into uikit ).
I will only buy a new MacBook once Apple puts in its own chip.There is no reason for a Mac to start at $1200 while better performing iPads are $800.
Honestly there is no justification for any of Apples prices of late. While I want to see ARM hardware from Apple (assuming an open Mac OS or other operating system) I’m not getting onboard if they continue to display the same greed in pricing.
As for that other operating system as long as Apple leaves the hardware “open” for Linux they can do what they want for iOS. In any event there are many things that an Apple ARM chip could bring to a Mac. One big feature is very low power usage.
"Any ARM shift won't be immediate, and won't span the entire product line in one fell swoop. It will likely start on Apple's low-end, like the MacBook and possibly a Mac mini migration."
That's an assumption. When Apple made the shift from PowerPC to Intel, they transitioned their entire Mac line within 1 to 1.5 years. There's no reason to believe they won't do the same when transitioning to ARM.
Two years is not a fell swoop. It takes about two years from design to execution, a bit less if an architecture already exists. What it won't be is the entire product line released at once.
This could be debated for years but I see Apple doing a transition as quick as possible. Machines like the current MBP will live on for years never updated. Mean while Apple will update new ARM machines as often as possible to put themselves as far as possible ahead of the competition.
As an interesting aside ARM just released a new advanced server architecture that frankly puts the performance issue to rest. While I don’t expect Apple to photo copy the designs there are significant design optimizations that are certainly worth considering. ARM hardware is currently the most interesting place for compute technology.
Does any one remember the Emate 300. I think Apples ARM strategy is in flux. I believe they wanted the MacBook line to move to ARM. However with the recent shift in Pro users and X86 changes this is probably not viable.
New chips should be hitting 6k single and almost 50k multi. Also the new GPUs are kinda out of Apple reach. Every iPad Pro user I have talked to also wants a more MacOS X style interface for Pro apps.
I would expect Apple to release a ultra thin MacBook with an ARM core. I am sure they wanted to make an ultra cheap iMac but this may not happen with out a faster CPU.
The biggest problem for Apple will be that bigger more power hunger ARM core will never really be worth it. Apple may sell lots of iPhones and iPads, but they will sell only 1-2 million high end ARM chips a year. This will make continued development costly. Intel makes 300-400 million chips a year and can easily justify the high end development.
Apple's ARM chips do not have the circuits or tech to really compete in the desktop space. Intel and AMD are moving to high core count chiplet designs on specialized substrates. They are moving to 5 GHZ + chips with ultra fast memory.
Apple's lack of a new iMac and recent rushed iterations of the 15" MacBook Pro probably are the result of an ARM course correction.
How good would a ArmBook look against $2,400 8core intel/Nvidia 2070 MaxQ notebook.
There ARM based products would need to be cheap, really cheap. This is not Apples Style.
There simply is no truth in this post. ARM has some very capable processors with a brand new server architecture just released. These are extremely capable chips and actually run faster than many similar x86 chips. There is nothing actually missing but even there Apple has its own IP to build into the chip.
BS. If it were coming to MacOS then AMD Threadripper and Ryzen would already be here.
How does this statement make any sense? What does AMD have to do with Apple planning to switch to their own ARM chips, AMD using x86/AMD64?
It means Intel is deflecting. Apple needs Thunderbolt, period. It's the only reason they've stuck with Intel after Zen came out. Intel has ZERO threat of ARM supplanting them on the desktop and laptop, never mind the Data Center. They have every concern of AMD and future generations using their superior products for LESS COST.
Apple was ecstatic when Intel announced Thunderbolt would be open sourced. Intel has dragged its feet for nearly 2 years since the announcement and it is still not royalty free and released.
So there is no rational basis for Apple to invest heavily into augmenting their ARM designs for a workstation [Mac Pro], never mind the desktop/laptop [And no iOS is fast because it is very limited in multi-user/multithreaded, multi-core based processing that will be a must on macOS. There are literally hundreds to thousands of processess/threads that are and can be running inside OS X that ARM won't ever supplant what is coming down the pike.
Basic threads and processes on my Macbook Pro 13: 1391 threads, 346 processes. The ARM would get slammed with that and that is nothing when pushing an iMac Pro or Mac Pro.
If you think Apple is going to screw over developers with ARM with the Mac Pro you're effing nuts.
Intel bound Apple when Apple [and as a former NeXT/Apple Engineer I was there] needed a fusion of legitimacy, especially when IBM crapped the bed. At NeXT we made a Quad FAT architecture for the OS because Motorola fucked us over more times than you can imagine on their designs. HP did the same thing. The HP PA-RISC ran circles around x86 at the time. HP did nothing to follow through.
Sun was just a clusterfuck of stupid with regards to the OpenStep initiative. Sun wanted all revenues on the Hardware and to force us to cut the cost of OpenStep licensing. So people were ``shocked'' that didn't take off? Please.
ARM dictates designs. Apple modifies but within those design specs.
You keep believing those pissant benchmarks the mobile world shows as performance figures. Throw 500 processes and 2000 threads at an iPhone and it crashes. There is a reason Apple has very limited subsets of functionality tuned around the tightly coupled hardware constraints.
The Mac Pro is effectively dead. Given that developers really don’t care about architecture as much as the do about performance. It is pretty clear now that ARM has real advantages here. Mainly because they can have cores running at 1-2 watts at hight clock rates than Intel or AMD. This leads to the prospects of a Mac Pro running 50 to 100 cores at far higher cLock rates than can be achieved with x86.
Hm, Threadripper 3 vs ARM. If I start to laugh, I'll die.
Threadripper is a great achievement on Intel’s part but ARM isn’t standing still. ARM just released their new server architecture and it is surprisingly good.
Apple has had all the common frameworks running in Ax chips since the beginning of Ax chips. Most executable app code is spent in these frameworks anyway: it may be apocryphal (but I don’t think so), but MS Excel in the PowerPC day is said to have spent 70% of its time in the Toolbox call DrawString(); since Apple had redone QuickDraw in PowerPC Native code, 70% of Excel ran natively with no work from Microsoft. That said, I’m not sure how well Ax chips will do with thermals if called upon to do x64 emulation.
I’m guessing Apple might give a huge lead time for developers and Marzipan, and release a Mac that did no emulation. But probably not.
Why Emulate?
Just use LLVM op code to flag if code can be ARM optomised (ie. most of system and apple apps, UXkit Apps) and shuffle the rest to a x86 co-processor. Scale the Aseries up and the x86 down each generation. Apple already have a custom scheduler in the Aseries doing this and sending code to different parts of the chip. Add system support to help/encoarge developers to split app into service modules so some modules run ARM some x86 some both.
I know this will sound crazy but I predict that Apple WILL NOT position the A-series as the low-end product, but instead position the A-series as the "High End" product categories.
I think it is already happening with the price of iPad Pros. I think they will retain exclusive features compared to Intel Chips, maybe these will be performance (just look at the current iPad A-series, faster than what 92% of portable computers), or battery life, or specialized application support, or maybe it will just be a spec advantage, like more Ram or thinner products.
Now this could fail, and I could be 100% wrong about how they position the A-series as premium but I find it very hard to believe they would make a transition unless 1. The A-series is better than Intel. and if that is true then why make it a value low-profit product??
Agreed. And if the power is indeed better than the Intel chips, i would expect low volume niche machines like the new Mac Pro to be the test bed for these chips and both prove themselves and scale to consumers via that model.
Except pro level machines rely on pro level software, typically from large, slow moving sources. What’s the point of a super fast workstation if I can’t run Pro Tools on it?
As with the switch to Intel, you deal with it and move on. What was the option? Stay with the Power PC chips to avoid upsetting people short term or work towards the future. Companies die when they can’t make the change.
The root issue is not jumping between different architectures, is if the next one offers better performance, thats how the user see's it.
Apple is looking at it from a very different direction they see owning the CPU allows them to control costs improving the profit not necessarily the users experience!
Then there is issue of physics! There is only so many ways to make a chip while Intel and others are rushing to smaller micro-architectures there is a physical limit on how far we can go! Once you reach that limit it really doesn't matter who's chip you go with as they all fall into the same constraints.
Look at it this way at one point you could get different formulations of gas from the different oil companies. Each was pushing how theirs was better at this or that. Today all gas is exactly the same! The only real difference is the octane and the impurities cheap gas may have. Basically, once something hits its plateau thats it on the most part.
I for one would love to see a AMD Threadripper CPU in the Mac Pro as I want massive number of threads I can't see ARM CPU's competing here. I love my iPad too (iOS) but it is not able to do what MacOS can do so thinking it can is a mistake. All I can see here is loosing performance not even holding the same line or even usability as I get out of MAcOS. Lets not get into the Henry Ford viewpoint http://superinnovator.blogspot.com/2012/02/you-can-have-any-color-you-want-so-long.html
Are the Geekbench scores of ARM CPUs and x86 CPUs apples to apples?
The two processors perform differently, and have strengths and weaknesses, but the Geekbench scores are a measure of the relative speed that the two processors chew through the same calculational loads.
How well this applies, or any benchmark applies, to any given person's use case varies a great deal. Geekbench is a good indicator of general use. It is less so for scientific work, or video transcoding.
Hm, Threadripper 3 vs ARM. If I start to laugh, I'll die.
Threadripper is a great achievement on Intel’s part but ARM isn’t standing still. ARM just released their new server architecture and it is surprisingly good.
I know this will sound crazy but I predict that Apple WILL NOT position the A-series as the low-end product, but instead position the A-series as the "High End" product categories.
I think it is already happening with the price of iPad Pros. I think they will retain exclusive features compared to Intel Chips, maybe these will be performance (just look at the current iPad A-series, faster than what 92% of portable computers), or battery life, or specialized application support, or maybe it will just be a spec advantage, like more Ram or thinner products.
Now this could fail, and I could be 100% wrong about how they position the A-series as premium but I find it very hard to believe they would make a transition unless 1. The A-series is better than Intel. and if that is true then why make it a value low-profit product??
Agreed. And if the power is indeed better than the Intel chips, i would expect low volume niche machines like the new Mac Pro to be the test bed for these chips and both prove themselves and scale to consumers via that model.
Except pro level machines rely on pro level software, typically from large, slow moving sources. What’s the point of a super fast workstation if I can’t run Pro Tools on it?
Large, slow moving sources? Like Nokia, Ericsson, Palm and BlackBerry who were too slow to adapt when the iPhone came out?
Microsoft has Windows on ARM now, with a 32-bit software compatibility layer, so virtualization or even Windows on top of one of these new machines isn't out of the question
One thing that nobody is talking about though is compatibility with Intel software. Sure Windows and some software will run on A series chips. But what about the Mac software that is coded to run on Intel? Are they going to run a Blue Box/Yellow Box strategy for a while? It took years before all critical software was ported from PPC to Intel. Some never was and there were a fair number of people who refused to update their OS for years after Apple dropped Rosetta, because they did not dare lose the old software they depended on. Any idea what Apple is going to do for them?
Move on or get left behind. The backwards compatibility at the expense of moving forward is a Microsoft thing that resulted in the mess that is the Windows world.
But why shift to ARM? What does ARM have to offer MacOS? x86 had libraries & Windows software which could be ported/integrated. Aarch64 has nothing that iOS/Xcode can’t already deliver.
If Apple are moving to 1st-party hardware they should go all in and move to 1st-party CPU ISA as they have with the rest of their hardware.
I wouldn’t be surprised if they already have a 1st-party CPU ISA in play just emulating Aarch64 (like Tegra does). In theory they could emulate x86 to and release I-series for Macs and ultimately ditch the emulation for both.
Windows itself may be of little concern to Apple's plans for the Mac, but Office is a whole different beast. I'm pretty certain that Microsoft and Adobe would be privy to these developments.
I know this will sound crazy but I predict that Apple WILL NOT position the A-series as the low-end product, but instead position the A-series as the "High End" product categories.
I think it is already happening with the price of iPad Pros. I think they will retain exclusive features compared to Intel Chips, maybe these will be performance (just look at the current iPad A-series, faster than what 92% of portable computers), or battery life, or specialized application support, or maybe it will just be a spec advantage, like more Ram or thinner products.
Now this could fail, and I could be 100% wrong about how they position the A-series as premium but I find it very hard to believe they would make a transition unless 1. The A-series is better than Intel. and if that is true then why make it a value low-profit product??
Why are you so adamant about the A-series branding being used in Macs? They don't have the GPU or many other SoC features used in desktop and notebook Macs so why not assume Apple will address this with another ARM-based chip that they designed to meet the needs of their PC line running macOS the same way they designed chips chips for the Apple Watch, AirPods, and controller chips?
“I” series. “A” series emulates ARM, “I” series emulates Intel.
wizard69 said:The Mac Pro is effectively dead. Given that developers really don’t care about architecture as much as the do about performance. It is pretty clear now that ARM has real advantages here. Mainly because they can have cores running at 1-2 watts at hight clock rates than Intel or AMD. This leads to the prospects of a Mac Pro running 50 to 100 cores at far higher cLock rates than can be achieved with x86.
Depends on what you are developing. I own a software development company. For my developers who are designing graphical frontends, the CPU architecture does not matter. But my cloud service developers need to run Docker containers at native speed. AWS, Azure and Google Cloud platform are Intel based only. The moment the last Intel based Macs dies, is the moment my cloud service developers can no longer use Macs as a development machine
I think you’re on the right track. I don’t believe that MacOS is transitioning to ARM. I think it’s more likely that they’ll release iOS-based “Macs” in the future.
Ditching Intel doesn’t make a lot of sense. It’s not in the customer’s best interest to lose compatibility with Windows. Moving to Intel saved the Mac, not because of performance but because of compatibility. I think it would be very foolish of Apple to lose that advantage across the entire product line.
iOS Macs? Makes no sense whatsoever.
iOS has a touch-based UI, and is optimized for using your fingers to interact with a touch-sensitive display.
MacOS is optimized for a mouse/keyboard UI.
When Apple brings ARM to Macs, rest assured it will run MacOS.
Comments
Marizpan has nothing to do with this, the software is already abstracted from the hardware. All that marizpan does is allow mac and iOS apps to be ported in software easier (the appikit will probably merge into uikit ).
As for that other operating system as long as Apple leaves the hardware “open” for Linux they can do what they want for iOS. In any event there are many things that an Apple ARM chip could bring to a Mac. One big feature is very low power usage.
As an interesting aside ARM just released a new advanced server architecture that frankly puts the performance issue to rest. While I don’t expect Apple to photo copy the designs there are significant design optimizations that are certainly worth considering. ARM hardware is currently the most interesting place for compute technology.
Apple is looking at it from a very different direction they see owning the CPU allows them to control costs improving the profit not necessarily the users experience!
Then there is issue of physics! There is only so many ways to make a chip while Intel and others are rushing to smaller micro-architectures there is a physical limit on how far we can go! Once you reach that limit it really doesn't matter who's chip you go with as they all fall into the same constraints.
Look at it this way at one point you could get different formulations of gas from the different oil companies. Each was pushing how theirs was better at this or that. Today all gas is exactly the same! The only real difference is the octane and the impurities cheap gas may have. Basically, once something hits its plateau thats it on the most part.
I for one would love to see a AMD Threadripper CPU in the Mac Pro as I want massive number of threads I can't see ARM CPU's competing here. I love my iPad too (iOS) but it is not able to do what MacOS can do so thinking it can is a mistake. All I can see here is loosing performance not even holding the same line or even usability as I get out of MAcOS. Lets not get into the Henry Ford viewpoint http://superinnovator.blogspot.com/2012/02/you-can-have-any-color-you-want-so-long.html
How well this applies, or any benchmark applies, to any given person's use case varies a great deal. Geekbench is a good indicator of general use. It is less so for scientific work, or video transcoding.
Large, slow moving sources? Like Nokia, Ericsson, Palm and BlackBerry who were too slow to adapt when the iPhone came out?
If Apple are moving to 1st-party hardware they should go all in and move to 1st-party CPU ISA as they have with the rest of their hardware.
I wouldn’t be surprised if they already have a 1st-party CPU ISA in play just emulating Aarch64 (like Tegra does). In theory they could emulate x86 to and release I-series for Macs and ultimately ditch the emulation for both.
iOS has a touch-based UI, and is optimized for using your fingers to interact with a touch-sensitive display.
MacOS is optimized for a mouse/keyboard UI.
When Apple brings ARM to Macs, rest assured it will run MacOS.