Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

YoitsTmac

Cancelled
Original poster
Aug 30, 2014
270
534
I'm a part of a BMW forum and we do hypotheticals like this a lot. This is something I've been thinking about for a long time and I'd love to hear what others think. A series chips are "low powered" and are getting increasingly fast every year. They are 64-bit, just like Mac. An iPhone 7 multicore Geekbench score is 5538 and an iPad Pro scores a nice 9178. These all have no cooling systems so could probably be a little quicker in a larger body. A current base 13" MBP scores a 8970 and a current regular macbook scores a 6700. Even if Apple didn't change these chips, their A series chips are right about on par with what intel is bringing. I also imagine that when apple designs the chip and the software, even greater optimization could be taken place. I don't know, what do you guys think? I just really want to here what other people think of this possibility if Apple decides to continue to care enough about the Mac.
 
This has been rumored for years, ie. https://www.cultofmac.com/289773/ex-apple-exec-says-macs-run-arm-processors-2016/

"Over the years, I’ve seen rumor after rumor that Apple would eventually abandon Intel chips in favor of ARM chips.", says an article from 2014.

One of the more recent rumors:

https://www.theinquirer.net/inquire...xt-gen-macbooks-with-arm-based-low-power-chip

Yep talked about for years. Likely won't happen because of architecture difference between processors. They'd have to license Intel stuff or run an emulator. Otherwise they'd force software makers to make specific MacOS apps from the ground up, and nobody has time or resources for that these days.
 
Yep talked about for years. Likely won't happen because of architecture difference between processors. They'd have to license Intel stuff or run an emulator. Otherwise they'd force software makers to make specific MacOS apps from the ground up, and nobody has time or resources for that these days.

Microsoft are going to do it and they only have access to inferior ARM processors. I believe they’re going the emulation route.
 
Last edited:
They'll do it. Won't run x86 apps though. Xcode will upload fat binaries or llvm byte code to the MAS, which will handle distributing the appropriate binaries, at least for MAS purchases. They're in a much better position to get it done than they were with the Moto to PPC or PPC to x86 transitions.
 
They'll do it. Won't run x86 apps though. Xcode will upload fat binaries or llvm byte code to the MAS, which will handle distributing the appropriate binaries, at least for MAS purchases. They're in a much better position to get it done than they were with the Moto to PPC or PPC to x86 transitions.

MAS is pretty crappy though. No photoshop, no office, no x86 virtualization apps, no weird obscure apps (corporate VPN clients, CA PAM client etc.) and very little chance of timely ports or even any ports in some cases. They might push devs into it if, and only if, they go all in (like they did with x86) but I don’t think they can do that because they would be bifurcation an already small market. Lastly, there simply isn’t enough power even on Apple’s A* chips to accomodate the very high end. No Mac Pro, no iMac pro and probably a pretty crippled MBP.
 
MAS is pretty crappy though. No photoshop, no office, no x86 virtualization apps, no weird obscure apps (corporate VPN clients, CA PAM client etc.) and very little chance of timely ports or even any ports in some cases. They might push devs into it if, and only if, they go all in (like they did with x86) but I don’t think they can do that because they would be bifurcation an already small market. Lastly, there simply isn’t enough power even on Apple’s A* chips to accomodate the very high end. No Mac Pro, no iMac pro and probably a pretty crippled MBP.

As a long time CPU designer, I guarantee you that if Apple wants a CPU for the pro macs, it can design one. The fact that the current A series don't meet that need is immaterial. They don't yet because it would be overkill for iPad and iPhone.
 
  • Like
Reactions: Feenician
As a long time CPU designer, I guarantee you that if Apple wants a CPU for the pro macs, it can design one. The fact that the current A series don't meet that need is immaterial. They don't yet because it would be overkill for iPad and iPhone.

You’re probably right. They have practically humiliated other ARM SOCS and even Intel on that low power (and I mean that in terms of power consumption, not speed) end. It still doesn’t resolve the software issues though. I’m just not sure the Mac market, especially a split one, is enticing enough for some devs to put the effort in.
 
You’re probably right. They have practically humiliated other ARM SOCSm and even Intel on that low power (and I mean that in terms of power consumption, not speed) end. It still doesn’t resolve the software issues though. I’m just not sure the Mac market, especially a split one, is enticing enough for some devs to put the effort in.

That's always the risk, but they did it twice before and they have more market power now than the last two times (true, they had ppc emulation for awhile, but most everyone eventually came on board).

The software hook will be that you'll be able to target a single binary at iOS devices and mac devices if you want (with some form of mouse-pointer-based uikit on arm mac).

I also wouldn't put it past apple to go to some new risc architecture they invent themselves. Code in apple's language (swift) using their compiler, distribute through their App Store to run on devices using their CPU with their instruction set architecture.
 
Yep talked about for years. Likely won't happen because of architecture difference between processors. They'd have to license Intel stuff or run an emulator. Otherwise they'd force software makers to make specific MacOS apps from the ground up, and nobody has time or resources for that these days.

Unless I'm missing something, I don't see this being much of an issue as long as the SDKs are at parity - fully compliant OSX apps would just need a recompile for a new binary. Those using undocumented APIs or low-level optimizations would likely need to have porting done.
 
  • Like
Reactions: T'hain Esh Kelch
Unless I'm missing something, I don't see this being much of an issue as long as the SDKs are at parity - fully compliant OSX apps would just need a recompile for a new binary. Those using undocumented APIs or low-level optimizations would likely need to have porting done.

Old, established code-bases like Creative Suite are a train-wreck. Mishmashes of optimized assembly, C, C++, C#, Objective C, Flash(!), depending on platform and chock full of x86 “optimizations” (read: hacks). Office has been cleaned up over the years though, and Microsoft already had a near full version on ARM on Windows, with a lot of shared code across platforms, so they are probably a bit nearer.

These are just two examples I picked on for being the 800lb gorillas but every type of business has their own. Add that to things like I mentioned earlier. Fortinet only recently delivered a usable VPN client on iOS, god only knows how long it would take CA to get their act together for PAM.

In my field, enterprise software, I absolutely rely on virtualization to build POCs and test environments (you cannot always rely on access to labs). An ARM Mac would, I’m sorry to say, being useless to me for work. I realize that I am somewhat niche in that but I think virtualization is fairly large, as niches go.

I’m very split on this. As much as I’d like to see what Apple would do, freed of Intels glacial roadmap, it would render Macs pretty useless for me for work and, I think, damage the application software ecosystem for Macs.

YMMV
 
Old, established code-bases like Creative Suite are a train-wreck. Mishmashes of optimized assembly, C, C++, C#, Objective C, Flash(!), depending on platform and chock full of x86 “optimizations” (read: hacks). Office has been cleaned up over the years though, and Microsoft already had a near full version on ARM on Windows, with a lot of shared code across platforms, so they are probably a bit nearer.

These are just two examples I picked on for being the 800lb gorillas but every type of business has their own. Add that to things like I mentioned earlier. Fortinet only recently delivered a usable VPN client on iOS, god only knows how long it would take CA to get their act together for PAM.

In my field, enterprise software, I absolutely rely on virtualization to build POCs and test environments (you cannot always rely on access to labs). An ARM Mac would, I’m sorry to say, being useless to me for work. I realize that I am somewhat niche in that but I think virtualization is fairly large, as niches go.

I’m very split on this. As much as I’d like to see what Apple would do, freed of Intels glacial roadmap, it would render Macs pretty useless for me for work and, I think, damage the application software ecosystem for Macs.

YMMV

For awhile the high end macs would likely have intel chips too. But in the end, intel isn't going to hit the power efficiency and performance benchmarks apple wants, and more and more of apple's ecosystem involves functionality intel won't give them, so...
 
The software hook will be that you'll be able to target a single binary at iOS devices and mac devices if you want (with some form of mouse-pointer-based uikit on arm mac).

I hadn’t thought of that. That may heavily tip the scales for developers.

It probably wouldn’t, at least in the short term, solve my virtualization challenges. Cross platform virt is not a solved problem and, while QEMU for example can do it, it is VERY slow.
[doublepost=1499664347][/doublepost]
For awhile the high end macs would likely have intel chips too. But in the end, intel isn't going to hit the power efficiency and performance benchmarks apple wants, and more and more of apple's ecosystem involves functionality intel won't give them, so...

That could work. Intel, under that circumstance I suspect, would be very hostile. There’s always AMD in the interim though...

Intel have been a big part of the problem (though not all of it) with the Mac roadmap. I am 100% certain though that Macs with CPUs and GPUs that Apple control would be developed and improved much more aggressively than they have been. All that engineering R&D would be used in various tiers in all the hardware.
 
Last edited:
I hadn’t thought of that. That may heavily tip the scales for developers.

It probably wouldn’t, at least in the short term, solve my virtualization challenges. Cross platform virt is not a solved problem and, while QEMU for example can do it, it is VERY slow.

Yep. If you need a machine that executes x86 instructions, a risc machine ain't likely gonna cut it. Or maybe they do a deal with intel - in exchange for letting them emulate x86 (either with a hardware instruction decoder wrapped around an arm core or by software), intel gets 100% of the baseband chip business (bye, Qualcomm) and/or gets to be the fab.
[doublepost=1499664673][/doublepost]
I hadn’t thought of that. That may heavily tip the scales for developers.

It probably wouldn’t, at least in the short term, solve my virtualization challenges. Cross platform virt is not a solved problem and, while QEMU for example can do it, it is VERY slow.
[doublepost=1499664347][/doublepost]

That could work. Intel, under that circumstance I suspect, would be very hostile. There’s always AMD in the interim though...

Intel have been a big part of the problem (though not all of it) with the Mac roadmap. I am 100% certain though that Macs with CPUs and GPUs that Apple control would be developed and improved much more aggressively than they have been. All that engineering R&D would be used for in varios tiers in all the hardware.

Remember that intel is willing to contract its fabs now, and also wants apple's broadband chip business. So apple has leverage. Plus AMD would be a possibility (i am not a fan, even though I designed their cpus for a decade :)
 
  • Like
Reactions: Feenician
I'll be honest, a lot of this is out of my know how, but there have been a number of things that people have deemed as difficult that Apple has somehow transitioned smoothly. Most memorably is 32bit to 64 bit on iOS and updating apps to be unique between iPhone 6 and 6 Plus.
 
I'll be honest, a lot of this is out of my know how, but there have been a number of things that people have deemed as difficult that Apple has somehow transitioned smoothly. Most memorably is 32bit to 64 bit on iOS and updating apps to be unique between iPhone 6 and 6 Plus.

Changing between 32bit/64bit is not nearly as difficult as entirely changing the architecture of a processor. As 32bit apps can run on a 64bit system, it is a gradual process. But changing the architecture is like rebuilding a building using different bricks/techniques, all new services, pipes and electrical... everything; and then using entirely new things throughout, whereas 32bit/64bit is more like switching to LED lighting.

Also uniqueness in Apps is very simple, Apple use a set resolution on their devices that scales easily, so it's just a case of having set break-points in a responsive UI.

The only transition would have to be through emulation, but this would not be sufficient enough for pro apps or users. However, if Windows went to an ARM based CPU, and MacOS followed onto an ARM based CPU (Most SoC are ARM based), then it would be simple for software developers to code for both. But that would completely cut Intel out of the loop, so it would seem more probable that they would just get Intel to make the chips.

If Intel can't make it, then it'd take years for anyone else to develop just to catch up, yet alone overtake them.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.