ARM-based OS X machine ready for prime time next year?

[Neodym]

During some research i stumbled over an image of the A6X CPU. People got used to the explanation that the "X" denotes a more powerful variant of the respective iPhone CPU, intended to be used in the iPad.

However, looking at the picture i figured the "X" was using a thicker font than the preceding "A6", as if to emphasize it. And suddenly i had to think of OS X and wondered, if the "X" series of Apple's "Ax"-CPU's was originally designed with the aim in mind to eventually use those variants for a full-blown OS X machine (like that rumored ARM-based MacBook Air).

The A6X more than doubled its Geekbench score (1776) over the A5X (781), being in the ballpark of a 2004 PowerMac G5 Dual-2GHz (1733) and close to a 2010 MacBook Air Core2Duo (2022).

If the A7X could manage to again double its score to ~3550, that would be about the power of a 2010 Mac mini or 2009 MBP and finally approaching the grounds of a 2011 core-i5 MBA (4558), which is still a powerful machine for low-to-mid level workloads even by today's standards.

Come A8X and an ARM-based OS X machine may suddenly be ready for prime time. Perhaps debuting in the 11" MBA, which is probably more often used for light mobile tasks than its bigger brothers and therefore the majority of its users would not need the power of a current (pricey!) i5 or i7.

Some reasons why Apple could pursue this direction:

• Second source for CPU's (currently they are completely dependant on Intel, as AMD lacks performance)
• Lower prices (though its unclear whether it would lead to lower end-customer prices or rather help Apple's margin),
• Further merging of platforms (iOS and OSX),
• More influence on CPU/GPU design (e.g. efficiency, design for target machine requirements, better integration with the OS etc.)
• In the future even inhouse production, thus better secrecy and improved value added chain (keeping more margin in the company)

 

[maflynn]

Microsoft's arm based tablets are awful, in performance and volume of applications. This is probably the best and biggest reason why moving to ARM for OSX is an awful idea.

 

[mattferg]

Several reasons why Apple will not switch to ARM soon and why it's a terrible idea:

1. Have to recode the entire OS and all their apps
2. No performance gains from switching to ARM, in fact performance loss.
3. With Haswell, no power gains from switching to ARM
4. Breaks compatibility with all 3rd party apps, forcing another Rosetta situation onto users
5. Breaking compatibility with all older Macs, meaning Apple will have to release dual binaries for several years until the Intel models are phased out.

In short, why would they even bother? And as for the "second source for CPUs" argument, they'd never sell both Intel and ARM machines at the same time, that'd be a marketing nightmare. See "Windows RT".

So unless they can release an ARM machine that's capable of being as powerful as the new Mac Pro, it ain't happening.

 

[maflynn]

5. Breaking compatibility with all older Macs, meaning Apple will have to release dual binaries for several years until the Intel models are phased out.

This is assuming that they plan on phasing out the intel chipsets. They make do a Microsoft move and have both. This makes #5 last not years but forever.

In short, why would they even bother? And as for the "second source for CPUs" argument, they'd never sell both Intel and ARM machines at the same time, that'd be a marketing nightmare. See "Windows RT".

Agreed, and hopefully Apple if they were thinking of doing this will see the issue Ms is having and drop the subject post haste.

 

[mattferg]

This is assuming that they plan on phasing out the intel chipsets. They make do a Microsoft move and have both. This makes #5 last not years but forever.

Agreed, and hopefully Apple if they were thinking of doing this will see the issue Ms is having and drop the subject post haste.

Yeah the only way they'd release both is if Apple protected themselves and made it for mobile devices, made it have it's own OS and ecosystem. They should call the OS something else too. I know, iOS!

But yeah, Apple has no reason to run two CPU architectures, they dropped PowerPC pretty swiftly after the Mac Pro was released, right? I see MS dropping RT soon too. I think ARM only really has a chance of survival on phones, I see Atom taking over in tablets tbh.

 

[B...]

Yeah the only way they'd release both is if Apple protected themselves and made it for mobile devices, made it have it's own OS and ecosystem. They should call the OS something else too. I know, iOS!

But yeah, Apple has no reason to run two CPU architectures, they dropped PowerPC pretty swiftly after the Mac Pro was released, right? I see MS dropping RT soon too. I think ARM only really has a chance of survival on phones, I see Atom taking over in tablets tbh.

Since Baytrail will see such an improvement this year approaching Core ix speed, Atom is a conceivable chip in 2016+'s Air.

 

[Stetrain]

Several reasons why Apple will not switch to ARM soon and why it's a terrible idea:

1. Have to recode the entire OS and all their apps

Just have to recompile. They've been practicing with compiling OSX components, frameworks, and apps for ARM for years now with iOS.


I do agree though that it's very unlikely, mostly due to Intel's improvements with Haswell and the new version of Atom coming out later this year. There would be some power usage advantage by switching to ARM for low end devices but it isn't worth dealing with the compatibility headache for a few percentage points.


I'm sure it's a contingency that Apple has been considering for a long time, just like how they kept Intel compatibility with OSX from the beginning just in case, but they won't pull that trigger as long as Intel keeps improving the power consumption of their mobile chips.

 

[thejadedmonkey]

Microsoft's arm based tablets are awful, in performance and volume of applications. This is probably the best and biggest reason why moving to ARM for OSX is an awful idea.

I'd have to disagree. My Surface RT is on par with a slower Windows machine. It really doesn't have any slowdown that impacts me on a daily basis, and my understanding is that Windows "Blue" is supposed to fix it anyway.

However, there's a much much larger picture to look at, regarding Microsoft.

Microsoft yesterday took a 900 million loss on the Surface RT. This means they expect to not sell over 600 million tablets. Either Microsoft was incredibly naive about sales projections, or they dropped almost 1 billion dollars into scaring Intel into making sure that their CPU's were power-friendly (and that ARM doesn't take off).

Now why would they want Intel to make more power efficient CPU's? Because if they don't, there's a chance that Apple would go ARM - meaning an entire ecosystem of Windows apps can't be run on Bootcamp. That's even more scary for Microsoft.

I think Apple was planning on going over to ARM, and I think Microsoft paid almost $1,000,000,000 to stop it from happening.

 

[B...]

I'd have to disagree. My Surface RT is on par with a slower Windows machine. It really doesn't have any slowdown that impacts me on a daily basis, and my understanding is that Windows "Blue" is supposed to fix it anyway.

However, there's a much much larger picture to look at, regarding Microsoft.

Microsoft yesterday took a 900 million loss on the Surface RT. This means they expect to not sell over 600 million tablets. Either Microsoft was incredibly naive about sales projections, or they dropped almost 1 billion dollars into scaring Intel into making sure that their CPU's were power-friendly (and that ARM doesn't take off).

Now why would they want Intel to make more power efficient CPU's? Because if they don't, there's a chance that Apple would go ARM - meaning an entire ecosystem of Windows apps can't be run on Bootcamp. That's even more scary for Microsoft.

I think Apple was planning on going over to ARM, and I think Microsoft paid almost $1,000,000,000 to stop it from happening.

Or MS dropped the RT price $150 recently and they are sitting on a stock of 6 million unsold devices, so they adjusted the shares $0.07 each for a total of a $900 million write down.

 

[TC25]

In summary, Apple going to an ARM-based processor is a stupid idea.

 

[alexrmc92]

Several reasons why Apple will not switch to ARM soon and why it's a terrible idea:

1. Have to recode the entire OS and all their apps
2. No performance gains from switching to ARM, in fact performance loss.
3. With Haswell, no power gains from switching to ARM
4. Breaks compatibility with all 3rd party apps, forcing another Rosetta situation onto users
5. Breaking compatibility with all older Macs, meaning Apple will have to release dual binaries for several years until the Intel models are phased out.

In short, why would they even bother? And as for the "second source for CPUs" argument, they'd never sell both Intel and ARM machines at the same time, that'd be a marketing nightmare. See "Windows RT".

So unless they can release an ARM machine that's capable of being as powerful as the new Mac Pro, it ain't happening.

1. Remember when OS X for intel was first announced? They openly stated that every version of OS X had been compiled and tested on intel processors behind the scenes. There is nothing stopping them from doing it with ARM either, so it has probably already been done.

2. True, but look how fast the performance of ARM processors are climbing. Still not on par with todays intel chips, but the future may change that.

3. ARM is still more power efficient than haswell, but haswell has a better performance per watt ratio.

4. Not necessarily. Maybe for some apps this would be the case, but apples libraries and API's have grown quite a bit since Tiger. Applications that don't have any custom cpu specific code in them would simply have to be recompiled after apple ports over their API's and libraries, which is fairly simple to do.

5. Again not a big deal if they have already been compiling them for years.

 

[maflynn]

1. Remember when OS X for intel was first announced? They openly stated that every version of OS X had been compiled and tested on intel processors behind the scenes. There is nothing stopping them from doing it with ARM either, so it has probably already been done.

True, and there were rampart rumors of apple keep a dual code base for intel chips. We've not heard of any of this, other then they might be moving there.

2. True, but look how fast the performance of ARM processors are climbing. Still not on par with todays intel chips, but the future may change that.

They're getting better, but so isn't intel chips and software is only getting more demanding. I don't think ARM chips will get ahead of the curve in this case.

3. ARM is still more power efficient than haswell, but haswell has a better performance per watt ratio.

Perhaps (I really don't know) but with Haswell we have an all day battery with the MBA.

 

[mattferg]

I'd have to disagree. My Surface RT is on par with a slower Windows machine. It really doesn't have any slowdown that impacts me on a daily basis, and my understanding is that Windows "Blue" is supposed to fix it anyway.

However, there's a much much larger picture to look at, regarding Microsoft.

Microsoft yesterday took a 900 million loss on the Surface RT. This means they expect to not sell over 600 million tablets. Either Microsoft was incredibly naive about sales projections, or they dropped almost 1 billion dollars into scaring Intel into making sure that their CPU's were power-friendly (and that ARM doesn't take off).

Now why would they want Intel to make more power efficient CPU's? Because if they don't, there's a chance that Apple would go ARM - meaning an entire ecosystem of Windows apps can't be run on Bootcamp. That's even more scary for Microsoft.

I think Apple was planning on going over to ARM, and I think Microsoft paid almost $1,000,000,000 to stop it from happening.

Yes that's right, Microsoft spent a billion dollars to make sure that Apple computers, 7% of the PC market and shrinking, could still run Windows apps on the percentage of Mac users who even do that (so like 4% of the PC market). Microsoft doesn't care that much about such a small percentage of the market (smaller than anyone running 8, 7 or XP)

More than likely MS was hedging its bets incase Intel wasn't able to improve the x86 architecture for power consumption.

----------

1. Remember when OS X for intel was first announced? They openly stated that every version of OS X had been compiled and tested on intel processors behind the scenes. There is nothing stopping them from doing it with ARM either, so it has probably already been done.

2. True, but look how fast the performance of ARM processors are climbing. Still not on par with todays intel chips, but the future may change that.

3. ARM is still more power efficient than haswell, but haswell has a better performance per watt ratio.

4. Not necessarily. Maybe for some apps this would be the case, but apples libraries and API's have grown quite a bit since Tiger. Applications that don't have any custom cpu specific code in them would simply have to be recompiled after apple ports over their API's and libraries, which is fairly simple to do.

5. Again not a big deal if they have already been compiling them for years.

So much misinformation in this post.

1. There's no benefit of going to ARM at all.
2. ARM is increasing in performance more than Intel because Intel already has the performance. ARM won't ever succeed Intel, it's just not that great an architecture.
3. Haswell is but one step towards better power consumption. Intel chips will be as power efficient as ARM long before ARM is comparable in performance to Intel.
4. Exactly, developers would have to recompile their apps, and thus everyone would have to update/purchase new versions, just like the Intel switch aaaaannnd BREAKING COMPATIBILITY.
5. The whole point of this wasn't the dual binary it was that they'd effectively be giving an end of life to all their current existing machines.

So all in all a terrible idea for Apple.

 

[magbarn]

With Broadwell expected to bring even more power savings with a die shrink, why would you even need ARM? Do we need a 24 hour MBA 13?

 

[jdechko]

1. There's no benefit of going to ARM at all.

While I agree that there is no benefit, it still doesn't mean that Apple couldn't have an ARM-compatible version of OS X in the labs, just like they did with the x86 version. We know that Apple likes to own their core technologies. They don't want to be in a position like they were with PPC (unable to upgrade or hit certain speeds).

Likewise, I'm sure they've at least got a prototype of an ARM laptop somewhere in the lab.

In fact, we know that at least part of OS X has already been ported to ARM, because that's the foundation for iOS.

 

[Beachguy]

What a depressingly bad idea. Dropping powerful chips and the entire software base for a slower, more limited chip design.

 

[alexrmc92]

While I agree that there is no benefit, it still doesn't mean that Apple couldn't have an ARM-compatible version of OS X in the labs, just like they did with the x86 version. We know that Apple likes to own their core technologies. They don't want to be in a position like they were with PPC (unable to upgrade or hit certain speeds).

Likewise, I'm sure they've at least got a prototype of an ARM laptop somewhere in the lab.

In fact, we know that at least part of OS X has already been ported to ARM, because that's the foundation for iOS.

You right about apple liking to own everything. This is why i would think if intel would switch at all they would switch to AMD. With AMD now allowing Sony and Microsoft to license it's chip designs and customize them apple could easily build their own x86 chip based on an AMD chip.

There are already talks of apple fully designing and building all of their ARM chips, whats stopping them from doing the same with AMD chips other than the fact that they will have some work ahead of them redesigning the AMD chips to be as efficient and Intel chips.

 

[mattferg]

While I agree that there is no benefit, it still doesn't mean that Apple couldn't have an ARM-compatible version of OS X in the labs, just like they did with the x86 version. We know that Apple likes to own their core technologies. They don't want to be in a position like they were with PPC (unable to upgrade or hit certain speeds).

Likewise, I'm sure they've at least got a prototype of an ARM laptop somewhere in the lab.

In fact, we know that at least part of OS X has already been ported to ARM, because that's the foundation for iOS.

They started testing an x86 version of OS X when x86 processors started to show significant improvements over the PPC architecture in both performance and power consumption. So far ARM hasn't shown any benefits at all over the current Mac/Intel range.

 

[Moonjumper]

Microsoft's arm based tablets are awful, in performance and volume of applications. This is probably the best and biggest reason why moving to ARM for OSX is an awful idea.

Microsoft tablets have always been awful, yet Apple managed to do a very nice tablet.

Do not judge Apple's potential from Microsoft's failures.

 

[mattferg]

Microsoft tablets have always been awful, yet Apple managed to do a very nice tablet.

Do not judge Apple's potential from Microsoft's failures.

Weird, I thought Microsoft only started doing tablets last year. And that the Surface Pro was actually very well reviewed and received, just overpriced.

 

[alexrmc92]

Weird, I thought Microsoft only started doing tablets last year. And that the Surface Pro was actually very well reviewed and received, just overpriced.

I agree, i like the surface pro and it performs decently. Sales are low because its just a standard windows 8 computer that costs a lot more than similarly spec windows 8 computers.

 

[Moonjumper]

Weird, I thought Microsoft only started doing tablets last year. And that the Surface Pro was actually very well reviewed and received, just overpriced.

Awful is an exaggeration for the current Windows tablets, but I was referencing the word in the post I quoted. There have been Windows tablets for more than a decade, but they were really poor, and were used as a reason why the iPad would fail. And we know how that turned out! The post I quoted was using similar flawed logic.

 

[scaredpoet]

However, looking at the picture i figured the "X" was using a thicker font than the preceding "A6", as if to emphasize it. And suddenly i had to think of OS X and wondered, if the "X" series of Apple's "Ax"-CPU's was originally designed with the aim in mind to eventually use those variants for a full-blown OS X machine (like that rumored ARM-based MacBook Air).

I think you're reading too much into the thickness of some text that few people will ever actually see in their daily use of the products carrying them.

 

[thejadedmonkey]

Or MS dropped the RT price $150 recently and they are sitting on a stock of 6 million unsold devices, so they adjusted the shares $0.07 each for a total of a $900 million write down.

I really really hope Balmer isn't that stupid.

 

[B...]

I really really hope Balmer isn't that stupid.

If you search Surface RT on Google, lots of results pop up showing that is what happened. It was in the NY Times Business section today.