Maybe OS X has been leading yet another secret life.... This is one of the reasons I love AI. We have some brilliant people. So, is it possible to rewrite and run OS X on ARM? Software folks?
That's just about what iOS is. It's OS X, or Mac OS as it is now known, somewhat adjusted and compiled for ARM. Underneath the covers, it's the same. Same kernel, same lots-of-other-stuff.
There is absolutely no reason that Apple couldn't build an ARM-based MacBook. As others have said, I expect that they have them running in the lab.
For people who use lots of x86-specific apps, a switch to ARM would be problematic. But for folks who use mostly apps from Apple (Safari, Mail, Pages etc) it would be a no-brainer. Many of the other popular apps would come quickly - it'll just be a recompile and fat-binary. That solution space has been well explored with other transitions (680x0 -> PPC -> x86).
One big question is whether or not there will be a "Rosetta" to allow x86 binaries to run on ARM, the way PPC ones did on x86. Apple certainly has the ability to do this - it's a question of whether or not the investment is worth it.
Microsoft is already running Win 10 on ARM, with their equivalent of Rosetta to execute X86 binaries. Apple has done this twice before so it would be quite surprising if they aren't doing it again in the lab.
Anyone claiming that an X86 is needed to run X86 applications should tell Microsoft they're doing it wrong.
Regarding the role of ARM hardware in Macs, I don't think they'll produce anything that shares MacOS between X86 and ARM. It'll be all or nothing. The rumored hardware in this story handles lid-closed sleep mode things, which doesn't require the entirety of MacOS. When Apple does finally produce an ARM based Mac, I don't think it'll have any Intel hardware in it. It would dilute the perception of A family processors to say "when the going gets tough, we switch to the Intel chip" and I don't see Phil Schiller saying that. Instead, I could see him saying "Our A family processor is faster and more power efficient than any Intel chip we could put in this laptop".
Maybe OS X has been leading yet another secret life.... This is one of the reasons I love AI. We have some brilliant people. So, is it possible to rewrite and run OS X on ARM? Software folks?
That's just about what iOS is. It's OS X, or Mac OS as it is now known, somewhat adjusted and compiled for ARM. Underneath the covers, it's the same. Same kernel, same lots-of-other-stuff.
There is absolutely no reason that Apple couldn't build an ARM-based MacBook. As others have said, I expect that they have them running in the lab.
For people who use lots of x86-specific apps, a switch to ARM would be problematic. But for folks who use mostly apps from Apple (Safari, Mail, Pages etc) it would be a no-brainer. Many of the other popular apps would come quickly - it'll just be a recompile and fat-binary. That solution space has been well explored with other transitions (680x0 -> PPC -> x86).
One big question is whether or not there will be a "Rosetta" to allow x86 binaries to run on ARM, the way PPC ones did on x86. Apple certainly has the ability to do this - it's a question of whether or not the investment is worth it.
Microsoft is already running Win 10 on ARM, with their equivalent of Rosetta to execute X86 binaries. Apple has done this twice before so it would be quite surprising if they aren't doing it again in the lab.
Anyone claiming that an X86 is needed to run X86 applications should tell Microsoft they're doing it wrong.
It's certainly possible, but consider the the differences from back 2006 when Apple announced the transition from PPC to x86. The new HW was faster so in many instances the PPC-based app was able to execute a task more efficiently than on the PPC Mac that was just replaced. With ARM, you're not going to see a 12" MacBook running a 3Ghz ARM with 8GiB RAM besting the latest MacBook Pro. Then consider Apple's philosophy with the new MBPs which didn't even include a single USB-A port. Since we're talking about a low-end machine being the first option for this inclusion and not a direct all-or-nothing replacement since Intel is moving their tech forward, unlike what IBM was doing, I think it's very likely that we won't see another Rosetta this time around. If you think about the changes we've had in over a decade in how Mac apps are developed with Xcode, bitcode, easy compiling, the Mac App Store, and the target market, the demand simply isn't the same.
That's not the right way to look at it. To say it's not feasible until it's as powerful as Apple most powerful Intel-based Mac is not how this will happen. Their ARM designs already outperform the lower-end Macs, and they do it using much less power whilst generating less heat. That's where this still start. There is no all-or-nothing with an ARM-based Mac that will force Apple to forego coming to market if they can't best Intel's fastest chips that could be in the next Mac Pro or iMac. You start with the low-end notebook where cost, size, and weight are important. These are customers that aren't likely to get a MBP, iMac or Mac Pro that they need to run larger, resource-hungry apps for a particular industry. You get those that need a basic email and internet machine and the handful of apps that they can get from the Mac App Store. With the latest advances in how apps are written, their Xcode IDE, and their App Stores titles will be appear quickly. From there Apple can slowly move ARM-based Macs into more Macs if Intel continues to drop the ball in the future.
First they don't outperform any of the low end macs using less power and heat. TDP is about the same. Where ARM outperforms Intel is cost as Intel ISN'T dropping the ball but providing Apple exactly what it wants: better performance per watt.
Second, if all you need is basic email and internet then get the iPad Pro or even plain old iPad. Which already has more than a handful of apps. An ARM based Mac is redundant for this market segment.
Finally, if you don't do all or nothing you end up with a fragmented product line like Windows RT. While XCode can handle generating apps for both ARM and X86 that's only for those apps in current development/maintainence. It also eliminates all folks that need windows apps via parallels or boot camp.
Which is a very large segment of enterprise users.
nht said: An ARM based Mac is redundant for this market segment.
Nope.
Finally, if you don't do all or nothing you end up with a fragmented product line like Windows RT.
Oh, so you're saying that when Apple included fat binaries for 32-bit and 64-bit it was "afragmented product line like Windows" when they had to create two separate versions of Windows? Short answer: Nope. Longer answer: Don't use MS as your guide when Apple has a very long history of proving everything you've said wrong.
For trial, Apple can offer ARM based 12"-13" macbook which basically has browser and apps but heavy processing happens in cloud. Overtime as ARM processors get faster, move those processing to local and one day acts like current macbooks.
For trial, Apple can offer ARM based 12"-13" macbook which basically has browser and apps but heavy processing happens in cloud. Overtime as ARM processors get faster, move those processing to local.
Heavy processing can happen right on the device. The A10X is more powerful than the 12" MB's processor and with a smaller power envelope. If Apple made it beefier as well as designed it for use with a desktop OS it would have a much higher performance and be considerably less expensive to produce.
Soli said:I think it's very likely that we won't see another Rosetta this time around. If you think about the changes we've had in over a decade in how Mac apps are developed with Xcode, bitcode, easy compiling, the Mac App Store, and the target market, the demand simply isn't the same.
You're right that Xcode's bitcode (https://lowlevelbits.org/bitcode-demystified/) obviates the need for Rosetta and we probably won't see emulation in any ARM based Macs, though Microsoft is proving it's both possible and (for them) practical. While Intel may be making progress with the X86, they also earn margin on their IP. If that margin is higher than they earn as a foundry (where they must compete more directly with TSMC and others), Apple can pocket the difference by switching to ARM and shopping the foundries.
Introducing an ARM based Mac with a two tiered approach is risky for sure, but the payoff could be huge as well. As Soli has indicated, think of a thin and light laptop targeted at general consumer, student, and office worker (not programmers or engineers). Like the 12" MB, but for $799 starting price. But more powerful for the apps it supports, better battery life, and with very good media capabilities. Margins could be same or better than current Macs.
Apple would probably end up owning about 20% total PC market share, with the usual 90%+ profit share. This is billions in upside of profit vs current situation. With Apple in full control of the silicon, the advancement of functions would accelerate (not specifically raw power, but specialized functions like media processing, speech recognition, photo manipulation, etc). Apple completes taking the overwhelming majority of profits across all computing platforms.
As as to an iPad addressing the same market, not in my experience. I love my iPad for many things, but it took me about twice as long to write this post on it vs. my Mac. A laptop form factor is simply better for many tasks, just as an iPad is better for many others.
One big question is whether or not there will be a "Rosetta" to allow x86 binaries to run on ARM, the way PPC ones did on x86. Apple certainly has the ability to do this - it's a question of whether or not the investment is worth it.
If you have a hybrid system then why drop x86 processor in the near / foreseeable future?
They could just downsize the x86 in each machine as it slowly gets relegated to a co-processor. First round touch bar, next up network operations to storage, release APP extensions that can run background. Third ARM with beefed up graphic cores drives screen and system and ARM compiled apps. After that pure x86 with for unsupported apps (including virtual Windows).
That's not the right way to look at it. To say it's not feasible until it's as powerful as Apple most powerful Intel-based Mac is not how this will happen. Their ARM designs already outperform the lower-end Macs, and they do it using much less power whilst generating less heat. That's where this still start. There is no all-or-nothing with an ARM-based Mac that will force Apple to forego coming to market if they can't best Intel's fastest chips that could be in the next Mac Pro or iMac. You start with the low-end notebook where cost, size, and weight are important. These are customers that aren't likely to get a MBP, iMac or Mac Pro that they need to run larger, resource-hungry apps for a particular industry. You get those that need a basic email and internet machine and the handful of apps that they can get from the Mac App Store. With the latest advances in how apps are written, their Xcode IDE, and their App Stores titles will be appear quickly. From there Apple can slowly move ARM-based Macs into more Macs if Intel continues to drop the ball in the future.
First they don't outperform any of the low end macs using less power and heat. TDP is about the same. Where ARM outperforms Intel is cost as Intel ISN'T dropping the ball but providing Apple exactly what it wants: better performance per watt.
Second, if all you need is basic email and internet then get the iPad Pro or even plain old iPad. Which already has more than a handful of apps. An ARM based Mac is redundant for this market segment.
Finally, if you don't do all or nothing you end up with a fragmented product line like Windows RT. While XCode can handle generating apps for both ARM and X86 that's only for those apps in current development/maintainence. It also eliminates all folks that need windows apps via parallels or boot camp.
Which is a very large segment of enterprise users.
I wish we had some data on this (what percentage of user or "enterprise users" use Windows at all on their Macs and how frequently). I work at company that is exclusively Windows for desktops. Plenty of staff bring their own laptops for meetings and whatnot. Probably half of those are MacBooks. I don't know any of those people who ever run Windows on those MacBooks. Personally, I would be stunned to learn that even 10% of Mac users install Windows on their Macs. Maybe it's a critical selling point for some people, but I doubt it's very many. So if some Macs couldn't run Windows, I don't think that would hurt Apple in the least.
Intel x86 inside Mac is a must for true full compatibility with the rest of the world (read Windows). Switch Mac to ARM and we will switch to Windows. Sadly. A shame for all.
I don't know why you guys think Apple would introduce new tech at the lower end? They could probably do that now with ARM chips.
The reason they're taking long is because they want to do it right. I don't see ARM Macs being as powerful as Intel but MORE powerful probably with a new custom A-chip. I think Apple will release the whole Mac line with ARM chips when ready to boost adoption, make profit and add new innovations.
Like someone else said I doubt Apple will be on stage saying "and for the low end models we're using our brand new chips".
I doubt that the T1 is the first ARM related component integrated in a Mac. For exemple, ARM core are used in SSD controllers and Ethernet controllers (the Thunderbolt display Ethernet controller has one, for exemple). It is the first that has a visible role, and the first that is Apple produced, probably. But probably not the first Arm core in a Mac.
Dual-processor MacBooks are coming: ARM and Intel.
The ARM will run macOS. The Intel chip will kick in for legacy apps. iLife apps will be recompiled for ARM and save tons of power and battery life compared to running the old apps on Intel.
All transparent to the user.
It will be like having an x86 PCI card in your old PowerMac.
Dual-processor MacBooks are coming: ARM and Intel.
The ARM will run macOS. The Intel chip will kick in for legacy apps. iLife apps will be recompiled for ARM and save tons of power and battery life compared to running the old apps on Intel.
All transparent to the user.
It will be like having an x86 PCI card in your old PowerMac.
This sounds pretty elegant to me. Also, it provides a decent roadmap for Mac developers to migrate apps to ARM.
Dual-processor MacBooks are coming: ARM and Intel.
The ARM will run macOS. The Intel chip will kick in for legacy apps. iLife apps will be recompiled for ARM and save tons of power and battery life compared to running the old apps on Intel.
All transparent to the user.
It will be like having an x86 PCI card in your old PowerMac.
I think this is what will happen. They need thunderbolt integrated with ARM (for all macs) and ability to use external GPU (for some macs) with ARM. Then, just use intel (for next couple years) for legacy apps/windows compatibility; they can buy the slowest clocked version of intel chip with slowest integrated-GPU since still faster than software emulation. For most intel chips buying the slowest clocked version will cut the price in half; more than enough to pay for souped-up Apple-designed ARM chip. Could then announce nearly simultaneously for all macs. After 4-5 years drop the intel chip. By then hopefully most Mac and Windows apps will have transitioned to having ARM versions.
Dual-processor MacBooks are coming: ARM and Intel.
The ARM will run macOS. The Intel chip will kick in for legacy apps. iLife apps will be recompiled for ARM and save tons of power and battery life compared to running the old apps on Intel.
All transparent to the user.
It will be like having an x86 PCI card in your old PowerMac.
I think this is what will happen. They need thunderbolt integrated with ARM (for all macs) and ability to use external GPU (for some macs) with ARM. Then, just use intel (for next couple years) for legacy apps/windows compatibility; they can buy the slowest clocked version of intel chip with slowest integrated-GPU since still faster than software emulation. For most intel chips buying the slowest clocked version will cut the price in half; more than enough to pay for souped-up Apple-designed ARM chip. Could then announce nearly simultaneously for all macs. After 4-5 years drop the intel chip. By then hopefully most Mac and Windows apps will have transitioned to having ARM versions.
There is no need for legacy app compatibility as xCode will handle that. Anything that can still compile in xCode will be built to run on ARM, or Intel. Maybe both in the same executable.
For apps already on the app store ( which on the Mac isnt nearly as universal as iOS) there wont even be the need for a re-compile.
Comments
http://www.theverge.com/2016/12/7/13866936/microsoft-windows-10-arm-desktop-apps-support-qualcomm
Anyone claiming that an X86 is needed to run X86 applications should tell Microsoft they're doing it wrong.
Regarding the role of ARM hardware in Macs, I don't think they'll produce anything that shares MacOS between X86 and ARM. It'll be all or nothing. The rumored hardware in this story handles lid-closed sleep mode things, which doesn't require the entirety of MacOS. When Apple does finally produce an ARM based Mac, I don't think it'll have any Intel hardware in it. It would dilute the perception of A family processors to say "when the going gets tough, we switch to the Intel chip" and I don't see Phil Schiller saying that. Instead, I could see him saying "Our A family processor is faster and more power efficient than any Intel chip we could put in this laptop".
Second, if all you need is basic email and internet then get the iPad Pro or even plain old iPad. Which already has more than a handful of apps. An ARM based Mac is redundant for this market segment.
Finally, if you don't do all or nothing you end up with a fragmented product line like Windows RT. While XCode can handle generating apps for both ARM and X86 that's only for those apps in current development/maintainence. It also eliminates all folks that need windows apps via parallels or boot camp.
Which is a very large segment of enterprise users.
Oh, so you're saying that when Apple included fat binaries for 32-bit and 64-bit it was "afragmented product line like Windows" when they had to create two separate versions of Windows? Short answer: Nope. Longer answer: Don't use MS as your guide when Apple has a very long history of proving everything you've said wrong.
Apple would probably end up owning about 20% total PC market share, with the usual 90%+ profit share. This is billions in upside of profit vs current situation. With Apple in full control of the silicon, the advancement of functions would accelerate (not specifically raw power, but specialized functions like media processing, speech recognition, photo manipulation, etc). Apple completes taking the overwhelming majority of profits across all computing platforms.
As as to an iPad addressing the same market, not in my experience. I love my iPad for many things, but it took me about twice as long to write this post on it vs. my Mac. A laptop form factor is simply better for many tasks, just as an iPad is better for many others.
They could just downsize the x86 in each machine as it slowly gets relegated to a co-processor. First round touch bar, next up network operations to storage, release APP extensions that can run background. Third ARM with beefed up graphic cores drives screen and system and ARM compiled apps. After that pure x86 with for unsupported apps (including virtual Windows).
The reason they're taking long is because they want to do it right. I don't see ARM Macs being as powerful as Intel but MORE powerful probably with a new custom A-chip. I think Apple will release the whole Mac line with ARM chips when ready to boost adoption, make profit and add new innovations.
Like someone else said I doubt Apple will be on stage saying "and for the low end models we're using our brand new chips".
The ARM will run macOS. The Intel chip will kick in for legacy apps. iLife apps will be recompiled for ARM and save tons of power and battery life compared to running the old apps on Intel.
All transparent to the user.
It will be like having an x86 PCI card in your old PowerMac.
For apps already on the app store ( which on the Mac isnt nearly as universal as iOS) there wont even be the need for a re-compile.