[AppleInsider] Apple may abandon Intel for its Macs starting with post-Broadwell

[R0H1T]

wouldnt OSX kernal need to be re-optimized all over again from scratch to support the ARM?

Then wouldnt apple need to invest in R&D on both tablet/Phone + Desktop CPU's which would cost a gignormous amount of resources?

I dont see how apple would do that when R&D is done by Intel, at a level which no other vendor can compete against.

It doesn't make sense and i dont see the saving from cost of having your OWN cpu.

Also who wants a desktop which is just as fast as your phone/tablet?

AFAIK OS X is UNIX based so porting it over to ARM is no biggie, certainly not for Apple.

How so? The majority of their R&D money has already been well spent on developing the Ax SoC's & software recompilation is just a one time thing, third party apps have to do it on their own if they are to continue using OS X as a platform. Apple will just have to do the legwork for some of their important applications like the iWork suite, the rest can be (probably) subsidized by Apple as far as certain other major third party apps are concerned.

 

[aigomorla]

AFAIK OS X is UNIX based so porting it over to ARM is no biggie, certainly not for Apple.

How so? The majority of their R&D money has already been well spent on developing the Ax SoC's & software recompilation is just a one time thing, third party apps have to do it on their own if they are to continue using OS X as a platform. Apple will just have to do the legwork for some of their important applications like the iWork suite, the rest can be (probably) subsidized by Apple as far as certain other major third party apps are concerned.

because a mobile cpu is not a desktop cpu, even in the case of arm.
They are very specific to what they do and what environments it would run.

You cant take a mobile cpu, make it higher tdp, increase clockspeed / cores and then stomp it on a desktop.

That type of evolution improvement requires a ton of resources on a massive scale, with validations, and yield optimization.

Apple would lose money overall vs having a company like intel do it for them.


When you look at the creation of any cpu a lot of factors are put in.
Material used in silicon, the litho process, and yield optimization.
It requires a LOT of resources to pull all that together, which is why AMD sold all it fab's because it could no longer handle the weight of such process.

Apple has too many products to spread that yield out. To burden there financial position with a new process node doesn't sound financially sound.

 

[R0H1T]

because a mobile cpu is not a desktop cpu, even in the case of arm.
They are very specific to what they do and what environments it would run.



When you look at the creation of any cpu a lot of factors are put in.
Material used in silicon, the litho process, and yield optimization.
It requires a LOT of resources to pull all that together, which is why AMD sold all it fab's because it could no longer handle the weight of such process.

Apple has too many products to spread that yield out. To burden there financial position with a new process node doesn't sound financially sound.

Does this include a timeline, say when they're within striking distance of Intel core's IPC, OR do you think it can never happen, let's just forget about whether it'll happen or not?

You cant take a mobile cpu, make it higher tdp, increase clockspeed / cores and then stomp it on a desktop.

That type of evolution improvement requires a ton of resources on a massive scale, with validations, and yield optimization.

Apple would lose money overall vs having a company like intel do it for them.

That isn't entirely true, an efficient uarch can always scale up, all you need is an good interconnect like QPI, HT or NVlink plus the uarch has to be good enough to justify this investment. Maxwell's shown that it can be done, so has Jaguar to an extent & there;s no reason Ax can't do it. The opposite isn't always true though, probably the reason why Intel couldn't get the core line to be down to Atom's level of power.

 

[Eug]

I'm not a coder but...

It should be noted that Apple has been dumbing down some of their consumer OS X applications so that they are much more like the iOS versions. I'm referring to iWork, iLife, and Airport Utility.

Those of us with Mac OS X laptops and desktops hate it, but I would guess that the similarities are not just aesthetic. There is probably much more shared code base than there used to be.

 

[aigomorla]

Does this include a timeline OR do you think it can never happen, let's just forget about whether it'll happen or not?

i dont think it will ever happen.

i honestly think intel showed all its competitors what happens when you poke a sleeping ogre. :biggrin:

If Apple was to do such a thing, they would not only be poking a sleeping ogre, but would also be shoving a stick at a sleeping dragon.

And well, when i used to be active with intel partners, they had a LOT of stuff which never got released due to the "need in demand" yet was shelf'd for future references.

My Sponsor also told me:
"There is no failed projects in Intel, its either shelf'd for future development, or a learning step ladder to implement in future products."

A perfect example of such is hyper threading.
When the original Pentiums had hyper threading, it was considered a failed innovation in the eyes of many.
However u wont find a single i7 which doesn't benefit from hyper threading, and its now released in all the current i7 cpu's.

Also with today's clock speed and performance increase, who knows if netburst variant wont even make a come back.

I'm not a coder but...

It should be noted that Apple has been dumbing down some of their consumer OS X applications so that they are much more like the iOS versions. I'm referring to iWork, iLife, and Airport Utility.

Those of us with Mac OS X laptops and desktops hate it, but I would guess that the similarities are not just aesthetic. There is probably much more shared code base than there used to be.

then the way i see is it they dont want to have a spread of different code across the two set of platforms.
It would make more sense then if they optimized the hell out of one set of code so it runs a ton more efficient on 1 platform, and then migrate to that platform on a pure code level.
If you had a poor cpu, yet excellent optimized code, it would run just as fast as a super processor with very poor code.

If apple is taking this route, then i can see x86 being replaced due to the unified code and the optimization of such code.

 

[R0H1T]

If apple is taking this route, then i can see x86 being replaced due to the unified code and the optimization of such code.

And this is what I've been saying all along ()

 

[aigomorla]

And this is what I've been saying all along ()

then i am in complete agreement with you. :whiste:

if it isnt difficult to make x rasberry pi's into a server like this:

it wouldn't be difficult to make a mac with 5-10 arm processors on 1 board with highly optimized code..

 

[kimmel]

That isn't entirely true, an efficient uarch can always scale up

Always?

 

[aigomorla]

Always?

ADD MOAR (pun on more) CORES!!! :thumbsup:

Ideally ARM uses such low power, you could just add one more, and if that isnt enough, add another, and keep on adding while optimizing the code to handle the multiple arm cores.
And due to arm's low power again, putting 5 - 10 - 15 - 20 isnt something very difficult... its implementing the code and fetch/handle which would be the hard part.

But if ur apple and dumping that much money into unified code, who knows if it isnt possible.

 

[Eug]

then the way i see is it they dont want to have a spread of different code across the two set of platforms.
It would make more sense then if they optimized the hell out of one set of code so it runs a ton more efficient on 1 platform, and then migrate to that platform on a pure code level.
If you had a poor cpu, yet excellent optimized code, it would run just as fast as a super processor with very poor code.

Yes, that's what I'm getting at. However the reason I hate it is because Apple is actually REMOVING features in the OS X versions.

If apple is taking this route, then i can see x86 being replaced due to the unified code and the optimization of such code.

I think they're just hedging their bets, in that that isn't necessarily the ultimate goal, but it is a possibility. If they keep developing their OS X and iOS applications completely separately, then it becomes much more difficult to introduce crossover products.

 

[III-V]

They'll save billions that they spend on Intel's processors, you think that isn't incentive enough then I have a bridge to sell you.

Yeah, that's not how economics work.

 

[aigomorla]

I think they're just hedging their bets, in that that isn't necessarily the ultimate goal, but it is a possibility. If they keep developing their OS X and iOS applications completely separately, then it becomes much more difficult to introduce crossover products.

that would be there biggest problem.

there would be very very little crossover, and emulation / translations would seriously be impacted due to the performance of such arch.

 

[Eug]

that would be there biggest problem.

there would be very very little crossover, and emulation / translations would seriously be impacted due to the performance of such arch.

It's possible to do this because it's gotten to the point where Apple ARM CPUs CAN run these consumer applications.

On an A8X iPad it isn't actually the speed that holds you back, but sometimes the interface does.

 

[aigomorla]

It's possible to do this because it's gotten to the point where Apple ARM CPUs CAN run these consumer applications.

On an A8X iPad it isn't actually the speed that holds you back, but sometimes the interface does.

i was more talking in the aspect of using a machine to write code for another machine and then testing the validation of said code on the same machine...

so example.. ur using a apple to write code for a Xbox, or a PS4, or even a windows environment....

Each of those developer tools would suffer from translation / emulation due to such a arch.
You would be strictly limited to apple only devices when you wrote the code and tested the code...

Infact how the hell does apple get away from anti trust laws, and microsoft cant still amazes me, i guess microsoft didn't have google to compete against.

 

[kimmel]

Yeah, that's not how economics work.

After the "Why doesn't Intel buy Qcom?" thread it's pretty clear that quite a few people don't have a good grasp on economics.

On an A8X iPad it isn't actually the speed that holds you back, but sometimes the interface does.

The interface is setup in the manner you describe, precisely because of the performance of the platform. Even the slick slide to unlock interface was an amazing bit of UI to hide how slow the hardware was.

 

[erunion]

Personally I don't think there will be a big push from Intel to ARM any time soon for the Macs, but as mentioned I won't rule out the possibility of a low end Mac tablet/laptop hybrid with ARM, esp. since that's a product that I'd actually consider buying myself.

Two problems with that.

An ARM Mac laptop would split the mac ecosystem in two, a la windows RT. It would have to be all or nothing.

The product you're describing could exist, but it would be an iPad with a keyboard. So definitely wouldn't count as moving OSX to ARM.

The second problem is that Apple has ridiculed windows convertibles. Of course, they did the same thing to large phones until the iphone6+ came out. So it may not really indicate anything

 

[ShintaiDK]

They'll save billions that they spend on Intel's processors, you think that isn't incentive enough then I have a bridge to sell you. IMO they just need 80~90% of the top end (Intel) perf & their ecosystem synergies + profits will be massive.

I'll reiterate though that for such a thing to happen they must have a certain vision in mind, besides their greed & mountain of cash will eventually lead them there

Seems someone sold you a bridge.

To save billions Apple would more or less have to develop and make 20 millions of them for free.

 

[el etro]

So basically, Apple is willing to:

1) drop x86 support
2) jeopardise x86 support for hundreds of thousands of apps on Mac OS X, save their own suite of apps (iWork, iLife, Safari etc.)
3) sacrifice iGPU and pure compute power of Intel ULV chips that are now reaching desktop class

all because they:

1) want to save negligible $ p. product sold
2) reduce lead times that aren't even a problem (?????)

I really hope you guys aren't taking this seriously.

No, greatnoob, they want only to build everything they can "In da House". There's no intent of compete with Intel. Apple just want to be as much auto-sufficient as they can.

 

[tential]

They'll save billions that they spend on Intel's processors, you think that isn't incentive enough then I have a bridge to sell you. IMO they just need 80~90% of the top end (Intel) perf & their ecosystem synergies + profits will be massive.

I'll reiterate though that for such a thing to happen they must have a certain vision in mind, besides their greed & mountain of cash will eventually lead them there

So I guess R&D, Developing Fabs, etc. is all not only cheap, but free!

If you look only at the retail cost to us of Intel's chips then lol...

 

[tential]

then i am in complete agreement with you. :whiste:

if it isnt difficult to make x rasberry pi's into a server like this:

it wouldn't be difficult to make a mac with 5-10 arm processors on 1 board with highly optimized code..

Ya, it definitely is easy to make multithreaded programs. Every program out there is very good at using multiple threads. Even better, every third party software developer loves to make code for a 2-4 threaded processor (most of the Windows base) and a 10+ thread (for your new apple processors).

Jesus... you guys are thinking of building it, not making it actually work in the real world.

But it's a tech forum, not a sales or economics forum so who can blame it.

 

[Red Hawk]

I say good riddance, it would restore the differentiation apple had on the pc space when they sported ibm cpus. To buy a mac that has idebtical hardware to what you can get in a pc its kinf of absurd, speciallu for the Premium they ask for their products

Nonsense. Differentiation for the sake of differentiation with no performance advantage is worthless to the end user. It sounds like...it sounds exactly like Apple, to be quite frank. :\

 

[PPB]

Nonsense. Differentiation for the sake of differentiation with no performance advantage is worthless to the end user. It sounds like...it sounds exactly like Apple, to be quite frank. :\

Aaaand someone finally grasped the concept. Absolute performance is useless for apple. Its all about product differentiation. The lack of software ecosystem is a failed argument, apple has a greater influence over the development of its software ecosystem than microsoft does, being a closed platform and all.

 

[ShintaiDK]

Aaaand someone finally grasped the concept. Absolute performance is useless for apple. Its all about product differentiation. The lack of software ecosystem is a failed argument, apple has a greater influence over the development of its software ecosystem than microsoft does, being a closed platform and all.

Yet Apples entire success is based on moving to performance.

 

[Rakehellion]

Lack of an SSD controller is a compelling reason. By not having one, they are forced to spend at least $30 more than they need to on storage.

So they're going to make a multi-billion dollar decision based on a SSD controller.

 

[ninaholic37]

I guess you missed probably the most important news from this year's CES ~
http://www.anandtech.com/show/8811/nvidia-tegra-x1-preview

Now imagine an SoC twice (maybe even thrice) as powerful as that, on 14nm, & it's GPU being at or slightly above par as compared to the X1. This coupled with active cooling ~
http://www.anandtech.com/show/8872/...p-technologies-and-cooler-master-join-forces-

Now that'll give you close to an i3m level of performance & a much better GPU, I doubt Intel has anything close to such a solution atm. If Apple were serious in doing something similar then you can bet ya they can replace the entire i3/i5 (u/m) variants from their lineup by just making OS X run on ARM & porting critical apps to it.

I wonder, with the rise of "good enough" computing and high-end CPU stagnation, if ARM will ever reach the stage that is acceptable for 95% of laptop/desktop users. I think Apple keeping their options open and preparing for / exploring the possibility is a good thing, at the very least, it's always good to have a backup plan to play around with.