Intel Expecting Apple to Transition to Custom ARM-Based Chips Starting in 2020

[ph001bi]

iOS and macOS are both Unix-based operating systems. Yes, iOS was optimized for smaller devices (mostly in terms of having less ancillary code running, as well as not having to support a lot of normal expected Unix services), but when you're running a computation-based benchmark, you don't care about all that other stuff, and, indeed, you try to minimize its impact on the benchmark (e.g. you don't run other intensive software while the benchmark is running). Computation-based benchmarks aren't measuring anything directly related to the OS, they're going to run roughly the same on an iPhone vs a MacBook, once you've quieted down anything that might interfere with the benchmark.


First, it's not at all clear that that is true. Second, putting an iPhone CPU in a "real" computer is not what Apple would do. One of Apple's huge advantages with their A-series of chips, is they can have their engineers target precisely the needs of their iPhones and iPads. The mobile phone environment is severely limited in terms of how much power you can use, and how much heat you can generate.

Those limits are much different in the laptop arena, and many orders of magnitude different in the desktop arena. So why do you assume that they would use chips designed for a phone in their other computers? Especially given that a huge advantage they have is the ability to precisely target their own needs? Other companies have to scrounge around for processors that fit (and mostly come in two sizes: too small and too large), while Apple has the equivalent of their own tailor making exactly what they need.

There's nothing stopping them from making a desktop ARM-based CPU with 5x as many cores, running at twice or more the clock speed. You haven't seen a chip like that from them yet, because it's more computational power than they need in a phone, and it couldn't stay within the thermal and power limits of the phone environment. But in a laptop, or desktop, it'd be fine. And I wouldn't be at all surprised if they've had at least designs for such chips in the lab for a few years.

Exactly, Apple designs chips according to a workflow. The A12X in the iPad Pro is not the same as the A12 in the XS. Different workflows, different designs. They will design a Laptop chip and a desktop chip. Just like Intel has got an Atom, a Celeron, a Pentium, an I3,5,7,9 core or a Xeon according to the requirement. There will be an A13, A13X, A13XX etc based on what it is they want the chip to do. Same thing for the GPU. I am pretty sure they won't stick to the intel integrated Iris or whatever it's called.
Microsoft has done a very good job by just adding a Neural Engine to the Qualcomm 8CX. Imagine what Apple could do.

 

[adamw]

Not sure where to stand on this. So much software is still Windows only, even today!

If Apple hopes to get this right, without alienating loyal customers, they should be prepared to offer us something much better than Intel, with some level of very good compatibility for older Windows and Mac apps, in addition to offering the support for iPad and iPhone apps working on the Mac.

 

[cmaier]

Not sure where to stand on this. So much software is still Windows only, even today!

If Apple hopes to get this right, without alienating loyal customers, they should be prepared to offer us something much better than Intel, with some level of very good compatibility for older Windows and Mac apps, in addition to offering the support for iPad and iPhone apps working on the Mac.

There will be no support for Windows apps.

 

[DevNull0]

The fact that people don’t understand CPUs and offer opinions like this is humorous. Things ignored:

This was a great post. You really are a master of self-referential humor.
[automerge]1571687725[/automerge]

There will be no support for Windows apps.

Most mac apps will also not be supported unless the dev updates them. So all software that's no longer being updated won't run, and you'll have to re-buy most of what is updated. I guess it will be a great opportunity for every dev to start pushing SaaS crap. There is just zero upside to the consumer.

 

[Ledgem]

Not sure where to stand on this. So much software is still Windows only, even today!

Not just Windows - x86-based Windows. Don't forget that when Microsoft created their Surface line of hybrid tablet-laptops they also had an ARM-based Surface running a special version of Windows called "Windows RT." The Surface line is still alive and well today, but they're purely running x86 processors now. Windows RT has been discontinued, and there are no more ARM-based Surfaces. The purported advantages that the ARM advocates talk up apparently weren't good enough to entice consumers and developers, and lack of software support will suffocate any new platform.

The only thing Apple has going for it today that Microsoft did not when it made the Surface RT is the fact that iOS exists on ARM, and Catalyst is supposed to help developers to bring those applications over to the Mac. But people don't buy Macs to run iOS apps, and last I'd checked none of my "heavy" applications have iOS versions.

 

[cmaier]

Not just Windows - x86-based Windows. Don't forget that when Microsoft created their Surface line of hybrid tablet-laptops they also had an ARM-based Surface running a special version of Windows called "Windows RT." The Surface line is still alive and well today, but they're purely running x86 processors now. Windows RT has been discontinued, and there are no more ARM-based Surfaces. The purported advantages that the ARM advocates talk up apparently weren't good enough to entice consumers and developers, and lack of software support will suffocate any new platform.

The only thing Apple has going for it today that Microsoft did not when it made the Surface RT is the fact that iOS exists on ARM, and Catalyst is supposed to help developers to bring those applications over to the Mac. But people don't buy Macs to run iOS apps, and last I'd checked none of my "heavy" applications have iOS versions.

There was a time when "heavy" applications didn't have intel versions. There was another time when "heavy" applications didn't have PPC versions.

Let's wait and see how apple handles the transition - they did a good job on two prior occasions.

 

[Brien]

Looking at the beast that is the A12X in the iPad Pro, I can see why.

As a customer looking to buy an Apple desktop this year, I'm really torn on what to do now. Is the lifespan of my Intel-based Mac going to be shortened by this? Seems likely. I also wonder if the transition will start with portables or desktops? You would think with a major modular redesign of the Mac Pro this year that it will be supported for a long time, but now I'm not sure.

I would assume so. I'd imagine that much like legacy 68k, PPC or 32-bit support, Apple will be chomping at the bit to drop Intel support as quickly as possible.

I could see them rolling out macOS 11 before 2025 and with it dropping support for all non-ARM hardware... and I'd imagine macOS 11 would ditch legacy APIs entirely so the entire stack is shared across mac/i/iPad/tv/watch OS's. A truly universal binary at that point.

 

[ASentientBot]

This was a great post. You really are a master of self-referential humor.
[automerge]1571687725[/automerge]


Most mac apps will also not be supported unless the dev updates them. So all software that's no longer being updated won't run, and you'll have to re-buy most of what is updated. I guess it will be a great opportunity for every dev to start pushing SaaS crap. There is just zero upside to the consumer.

With Apple's previous PowerPC to Intel transition, their Rosetta translation layer allowed us to run PPC apps on x86 machines for a couple years. I'm sure they will do the same thing this time, with likely an even longer grace period since there are so many more programs now to be updated.

The upside to the consumer would be faster, more efficient computers, and possibly the ability to run iOS apps without a recompile for Catalyst/Marzipan. There will be some issues (no chance of patching the OS onto older Macs or hackintoshes, loss of old hardware drivers) but 95% of user applications will most likely work.

 

[cmaier]

With Apple's previous PowerPC to Intel transition, their Rosetta translation layer allowed us to run PPC apps on x86 machines for a couple years. I'm sure they will do the same thing this time, with likely an even longer grace period since there are so many more programs now to be updated.

The upside to the consumer would be faster, more efficient computers, and possibly the ability to run iOS apps without a recompile for Catalyst/Marzipan. There will be some issues (no chance of patching the OS onto older Macs or hackintoshes, loss of old hardware drivers) but 95% of user applications will most likely work.

i am not so sure this time they’ll have an emulation later. Lots of patent problems in the x86 space. We will see.

 

[Unregistered 4U]

The fact that people don’t understand CPUs and offer opinions like this is humorous. Things ignored:

1) the ARM cpu in the mac won’t be an iPhone chip just stuck into a mac
2) even the ARM cpu in your iPhone is capable of laptop-like speeds if given a sufficient thermal solution so that it doesn’t have to keep throttling the frequency down. And it just so happens that sticking it in a bigger case, like a laptop, allows you to have such a solution
3) there is absolutely nothing that makes ARM ISA chips inherently slower than AMD64/x86-64 chips. In fact, given identically-talented designers, and modern compiler technology, ARM has a slight advantage since you can use a significantly simpler instruction decoder, which takes less space, eliminate certain critical paths, and run the pipelines faster.
4) Apple’s CPU designers are much better than Intel’s. At AMD we never were worried about the designers at Intel - we were worried about their excellent fabs.
5) TSMC‘s fabs are at least as good, if not better, than Intel’s now.

Additionally, I think the biggest thing usually left out of the equation is that, for an ARM Mac, Apple would control not ONLY the makeup of the processor BUT the instruction set as well (recent updates to the ARM license indicates that companies like Apple can extend the instruction set). So, imagine a processor that is tuned to execute, say, much utilized Final Cut Pro and Logic Pro workloads faster than any Intel processor could EVER execute them (Think Apple’s Mac Pro “AfterBurner” card at the CPU level).

“Benchmarks don’t tell the whole story” works in both directions. Some generic benchmarks may still indeed put ARM behind Intel in some specific tasks, but, in the end, if the software I run is many times faster on ARM because it’s tuned for ARM, then you can keep your benchmarks, and I’ll keep the time saved rendering
[automerge]1571714209[/automerge]
Someone with the skills, write a simple app for iPad OS that searches for primes. Use Mac Catalyst to get it running on the Mac (select the “Mac” checkbox in the project settings of that iPad app to create a native Mac app. You could then enhance it further, but don’t, just take what Catalyst spits out).

Post the performance comparison. I mean, it’s all conjecture, but if the thing I want to do is search for primes, and I can quickly create an app that runs better on an iPad Pro than a MacBook, then it doesn’t matter Intel vs. ARM, I’d use the fastest device I have available to me which MIGHT be an iPad Pro.

 

[Unregistered 4U]

Not just Windows - x86-based Windows. Don't forget that when Microsoft created their Surface line of hybrid tablet-laptops they also had an ARM-based Surface running a special version of Windows called "Windows RT." The Surface line is still alive and well today, but they're purely running x86 processors now. Windows RT has been discontinued, and there are no more ARM-based Surfaces.

Microsoft announces Surface Pro X

It’s the first ARM-based Surface from Microsoft since the Surface 2

www.theverge.com

Available on November 5th. So, there are none NOW, but there soon will be

But people don't buy Macs to run iOS apps, and last I'd checked none of my "heavy" applications have iOS versions.

This is an interesting point that I wanted to take further, not only do people not buy Macs to run iOS apps, MOST people don’t buy Macs for anything! Yes, they still sell millions per year, but when you consider the entire computing world outside of MacOS (including iOS and all of the Linuxes and Windows) the vast majority of people doing any kind of computing is doing that computing without MacOS. And, just like many folks today are using MacOS hardware to run other OS’s, I wouldn’t be surprised if a decent percentage one of those OTHER OS’s in the future is iOS.

 

[Ledgem]

Microsoft announces Surface Pro X

It’s the first ARM-based Surface from Microsoft since the Surface 2

www.theverge.com

Available on November 5th. So, there are none NOW, but there soon will be

All right, you got me there. It remains to be seen if it'll be successful or if it will be a repeat of the Surface RT. I'm skeptical, but I'm not going to predict its demise before we see if it offers any advantages, and how the market reacts.

This is an interesting point that I wanted to take further, not only do people not buy Macs to run iOS apps, MOST people don’t buy Macs for anything! Yes, they still sell millions per year, but when you consider the entire computing world outside of MacOS (including iOS and all of the Linuxes and Windows) the vast majority of people doing any kind of computing is doing that computing without MacOS. And, just like many folks today are using MacOS hardware to run other OS’s, I wouldn’t be surprised if a decent percentage one of those OTHER OS’s in the future is iOS.

You're right, but macOS still represents a very important piece of the ecosystem for Apple. Apple's devices are strong on their own, but the real unique draw is the way their devices seamlessly integrate with one another. It's not only a strength, but a barrier to exiting.

I use a Mac, an iPhone, an iPad, and have a host of other Apple devices. Some of those devices would work rather poorly with anything non-Apple, and non-Apple devices wouldn't patch in and work as seamlessly. If the Mac becomes unusable because the software that I need isn't available, then I'm forced to go with a different operating system and computer. Once I do that, the iPhone and iPad lose some of the strengths that differentiate them from any other smartphone or tablet. Suddenly I go from upgrading my iPhones and iPads once every few years without looking at any other non-Apple options to being a free agent.

I don't think that I'm unique in that, although I don't have sales data to indicate how unique or not my case is. I'd imagine Apple has data to show whether the people switching back and forth between iOS and Android have multiple Apple devices or not, though. They're likely aware how carefully they need to tread in this category.

 

[DevNull0]

With Apple's previous PowerPC to Intel transition, their Rosetta translation layer allowed us to run PPC apps on x86 machines for a couple years. I'm sure they will do the same thing this time, with likely an even longer grace period since there are so many more programs now to be updated.

Rosetta was possible because the x86 machines were immensely faster than the PPC CPUs they replaced (which was the whole reason for the transition to x86 -- PPC had fallen many years behind Intel in performance).

Modern desktop and laptop x64 processors are much faster than any arm processor currently in existence. Even the strongest proponents of a switch to ARM say that ARM can catch up quickly and desktop class CPUs can be created. Even if that is true, there is no way an ARM processor will have the power to interpret x64 code in real-time any time soon.

And that ignores the fact that as good as Apple's mobile CPU engineers may be, they're not going to out-desktop AMD and Intel on a short time horizon. ARM is not going to beat x64 any time soon.

The upside to the consumer would be faster, more efficient computers, and possibly the ability to run iOS apps without a recompile for Catalyst/Marzipan. There will be some issues (no chance of patching the OS onto older Macs or hackintoshes, loss of old hardware drivers) but 95% of user applications will most likely work.

Faster more efficient computers assumes someone *will* come up with an ARM CPU that can beat x64. Right now, that's Sci-Fi. And if Apple wanted, they could allow you to run an iOS app on x64 without a recompile, because x64 is so much more powerful than any iPhone or iPad A?? chip.

And when you say 95% of user applications will most likely work, that's a huge citation needed. I'm not going to call myself a typical user, but more than half my stuff wouldn't work. Heck, I still use Adobe CS6 (legitimately bought) a lot on my Mac because I refuse to pay for SaaS and won't touch CC with a 10-foot pole. Even that will be broken if Apple switches architectures. I also do a lot of IoT stuff on semi-open source platforms (particle.io, RPi, thread, etc). The thought of having to recreate my toolchains on a new architecture is enough to give me nightmares.

 

[chucker23n1]

The Surface line is still alive and well today, but they're purely running x86 processors now. Windows RT has been discontinued, and there are no more ARM-based Surfaces.

Windows RT has been discontinued indeed, but Windows on ARM has not, and ARM Surfaces are a thing again. https://en.m.wikipedia.org/wiki/Surface_Pro_X
[automerge]1571727867[/automerge]

Modern desktop and laptop x64 processors are much faster than any arm processor currently in existence. Even the strongest proponents of a switch to ARM say that ARM can catch up quickly and desktop class CPUs can be created.


Faster more efficient computers assumes someone *will* come up with an ARM CPU that can beat x64. Right now, that's Sci-Fi.

Apple Ax is already faster than many of their current MacBooks, at least in short bursts. What remains to be seen is if sufficient cooling will allow it to sustain that performance. If so, it’s perfectly adequate to replace Core-Y and Core-U today. Maybe -H next year.

 

[Unregistered 4U]

All right, you got me there. It remains to be seen if it'll be successful or if it will be a repeat of the Surface RT. I'm skeptical, but I'm not going to predict its demise before we see if it offers any advantages, and how the market reacts.

It was pretty close. I actually double checked the date on your post before I replied Apple and Microsoft will both try and try again in this area because the future belongs to ultraportable systems and, because that’s not Intel’s bread and butter (they make WAY more money on CPU’s that are ok to have big ol‘ heat sinks), Intel’s fine to continue to miss their estimates forcing system builders to make do with years old technology. And really, the Surface Pro X doesn’t even have to be a screaming fast machine. Heck, pretty much ANY computer an average person buys today is way more power than they’ll ever use... we’re in a world where being half as fast as a competing Intel chip is actually “good enough” for most computer users.

You're right, but macOS still represents a very important piece of the ecosystem for Apple.

See, I don’t think it’s very important as someone can spend thousands of dollars a year on Apple products AND services, yet never own a MacOS system. Things were different when you needed to sync your iOS device with a Mac, but once they removed that requirement, people who NEED MacOS are those that primarily needed MacOS in the past and still use it as a result. Or, who need to do the increasingly few things that can’t be done on iOS, (like code development for iOS).

 

[Rotten_Apple_]

My only concern with this is will this change still allow us to run Boot Camp and Parallels Desktop on ARM based Macs?

Can't promise you anything regarding money but if you want to run Windows 10 on an arm based Mac, cloud computing is an avenue you can consider. ShadowPC, for example, let's you stream a Windows desktop instance to your mac or iOS device, since it's built for gaming it can more than handle anything office related, only downside is the price, at $25 a month with a yearly commitment, but it's certainly an option, and probably a better alternative to installing "Master Race" on your mac.

 

[0388631]

much utilized Final Cut Pro and Logic Pro workloads faster than any Intel processor could EVER execute them (Think Apple’s Mac Pro “AfterBurner” card at the CPU level).

The card is an FPGA enhanced with ARM. Intel already has processors with built in ARM and FPGAs. They also have Xeons with built in FPGA. There already exists FPGAs meant for video editing available for Windows workstations. They've been around for a while now. FPGAs are only now, in the last few years, becoming affordable to manufacture for the semi-masses. RED's Red Rocket X is an example of an FPGA tuned for their files (.r3d).

 

[poorcody]

Can't promise you anything regarding money but if you want to run Windows 10 on an arm based Mac, cloud computing is an avenue you can consider. ShadowPC, for example, let's you stream a Windows desktop instance to your mac or iOS device [...]

I'm starting to come to this idea as well. I've been reading some on how Microsoft has been working on Windows on ARM. They have 32-bits apps emulated fairly well (not so much because of great Intel emulation, but because the apps are calling native-ARM Windows API). But Microsoft has said they will never support x64 apps. From what I've read, emulation of a modern Intel processor is a lot harder than the older processors of previous generations, both technically and legally.

I've actually been streaming to a home Windows server to run NVIDA code that no longer runs on Mac. It actually works fairly well. Kind of forget I am not running natively. I am considering replacing my Windows/Parallels emulation with this.

So if ARM-Apple can't run Windows, and I decide so stick with Mac at that point, this seems like a viable option.

 

[Unregistered 4U]

Intel already has processors with built in ARM and FPGAs. They also have Xeons with built in FPGA. There already exists FPGAs meant for video editing available for Windows workstations.

Not talking FPGA’s specifically, more something like this:
IOS JavaScript Performance

Changes Apple made to their memory subsystem yielded a dramatic increase in JavaScript performance. Such that iOS at the time benchmarked faster than an iMac Pro (and the A13 may even do better).

This shows that Apple could tune not ONLY their CPU (at the hardware level, not by using FPGA’s) but the entire motherboard such that nothing else in the world could possibly run Final Cut Pro X faster... especially when you consider that they could create a version of FCPX HIGHLY optimized for their own processor and subsystems. If a user’s entire world is in FCPX, Intel or ARM doesn’t matter, they just want the one that will allow them to finish the job faster.

 

[0388631]

Not talking FPGA’s specifically, more something like this:
IOS JavaScript Performance

Changes Apple made to their memory subsystem yielded a dramatic increase in JavaScript performance. Such that iOS at the time benchmarked faster than an iMac Pro (and the A13 may even do better).

This shows that Apple could tune not ONLY their CPU (at the hardware level, not by using FPGA’s) but the entire motherboard such that nothing else in the world could possibly run Final Cut Pro X faster... especially when you consider that they could create a version of FCPX HIGHLY optimized for their own processor and subsystems. If a user’s entire world is in FCPX, Intel or ARM doesn’t matter, they just want the one that will allow them to finish the job faster.

FPGA is a hardware component, not a software one. If Apple did have an FPGA in their mobile devices, it would be tuned as you said it would, at the hardware level during the development phase. Not all FPGAs are designed the same way. An "off the shelf" solution wouldn't perform the same as one Apple designed themselves and tuned themselves later on in software.

Furthermore, Apple controls their entire stack as opposed to competitors. Qualcomm processors aren't quick as Apple's because Qualcomm simply doesn't care because the cost to break into ARM development is very, very high. Apple can afford it with how much they spend on R&D yearly. Not to mention Qualcomm litigating when it can to prevent others from entering the market.

When you have Apple money, you can buy your suppliers or third party software developers out and move everything in-house. Samsung did this with their Corephotonics purchase a year ago, a supplier that supplied to many companies including Apple. They don't have Apple money, obviously. In the last few years, Apple purchased quite a few augmented reality companies. What is Apple offering on their new iPhone these days. Buy and improve. That's the motto of any successful company. Spend as little as on scratch development unless it'll provide large return. Apple spends a fortune on their A~ processors, and for good reason, too. The ROI on this via new devices is staggering.

For us Samsung fans, we don't know when we'll see full integration with their product line.

 

[Unregistered 4U]

FPGA is a hardware component, not a software one.

Sorry, I specifically meant non-FPGA hardware component.

 

[rlhamil]

I'll still want to be able to virtualize various Intel OS's (Windows, Linux, Solaris, others) on a MacBook Pro (not to spend a lot of time in, but to have something I can carry with me to try things out on), and not merely run them in a much slower emulator.

If some lower-end systems like the MacBook Air were running a non-Intel-compatible chip...whatever, not my problem. But I worry, because although two different CPU architectures (PowerPC+Intel, or 32-bit and 64-bit) can be supported, the extra bother and expense for the development and support is such that the temptation will always be to dump one when they can.

 

[Unregistered 4U]

I'll still want to be able to virtualize various Intel OS's (Windows, Linux, Solaris, others) on a MacBook Pro (not to spend a lot of time in, but to have something I can carry with me to try things out on), and not merely run them in a much slower emulator.

The great thing is that this isn’t something Apple can do overnight. So, there will be plenty of advance warning so you can pick up the last Intel MacBook Pro before they go all ARM.

 

[Ledgem]

The great thing is that this isn’t something Apple can do overnight. So, there will be plenty of advance warning so you can pick up the last Intel MacBook Pro before they go all ARM.

"Advance warning" for me would mean, begin the transition back to Windows or Linux-based Intel systems. The only time I hoarded systems and ran archaic systems into the ground was when running specialty scientific applications that really couldn't be run on any other system. Migrating everything over to a new operating system is a huge pain, but in the consumer space it's extremely rare to have applications that truly lock you to a certain system or operating system.

Apple's impressive ecosystem, the way that you can fluidly move files, conversations, and tasks between devices with ease, is the major draw that prevents me from even bothering to take more than a passing glance at competing products. But if I can't get my tasks done, then there's no point in having the product. At that point, Apple's ecosystem strength crumbles and even becomes a liability: at that point there's really nothing holding me back from looking at replacing every single one of my Apple devices and services (I have iCloud and Apple Music subscriptions that would be near-useless without Apple products) with products made by someone else.

 

[EnderBeta]

This will be a disaster. The virtual machines that I run are going to crawl under any kind of emulator that Apple produces, and I'm not all that sure they'll even throw us that bone. Apple has been giving the Mac short shrift for years now and this puts the final nail in the coffin.

But many games come to iOS so this opens up more games to Macs. My virtual machine needs are for office software products and those should run fine with a emulator.
[automerge]1572464913[/automerge]

Absolutely.
I wish I could run GeekBench for a living.

There is more to performance than raw "speed".
Ergonomics, screen size, colour accuracy — external storage… hell, there is quite a list where a desktop or laptop still beats an iPad. Hands down. Any race. Any time.

In the hypothetical scenario where a MAC uses the arm chips I would hope it has all the things you’re listing since it is supposed to be the same form factor.

An interesting prospect for Pro models is Apple could add additional cores and maybe have dual processors with more cores if they hit a wall on implementing cores into the socket with reliably high yields.