Apple Considering Switch From Intel Chips in Future Macs

Renzatic said:

I think it's funny how Google didn't have a map app until Apple released theirs. Now they're griping about their knock off not getting approved on the App Store.

Well, maybe if you innovated a little more this wouldn't be a problem for you, Google. Apple invented vectors for their maps. what have you done? Huh? Yeah. Thought as much.

Google is desperate.

Click to expand...

Here is the thing. I also use Nokia maps. I have had none of the problems I have had with Apple maps. People might say my experiences are anecdotal. Well, they aren't so anecdotal with all the complaints I have seen.

Bottom line, I don't care about vectors. I am not sure why I SHOULD care about vectors when the damn map isn't taking me where I need to go.

Also, when the CEO of Apple makes a PUBLIC APOLOGY for the failure which is Apple Maps, you can bet it took more than some anecdotal evidence to get him to that point.

----------

WilliamLondon said:

I think people need to relax about this, the uproar in these two threads over a rumour is so melodramatic.

Click to expand...

Half the fun of the threads is the [polite] uproar (at least for me)

VenusianSky said:

Virtualization will make this possible. I wouldn't be surprised if Microsoft and VMWare are trying to virtualize ARM in HyperV/ESX. That will allow ARM applications to run in a virtual environment on an x86/x64 platform.

Click to expand...

I like the way you think. But honestly if they could make a chip to run it all that would be great.

Renzatic said:

True, but Grand Central Dispatch doesn't make creating a multithreaded application easy, just easier. If you have to offload even the most basic app to multiple CPUs to offset the fact it's single core performance is below par, no one is going to want to use it.

If you have to offload even the most basic app to multiple CPUs to offset the fact it's single core performance is below par, no one is going to want to use it.

Click to expand...

Programmers who are creating simple applications don't use heavy multithreading,
because of low system requirements - and, as result, GCD is not needed and rarely used.

Those who are creating complex applications, which use heavy multithreading,
have enough skill to successfully use GCD and improve the overall performance greatly.
Most complex applications could be multithreaded in a big way, and, thanks to GCD, it is not that hard to accomplish.

So, in cases when GCD is really needed, it is easy to use for those who need it

KnightWRX said:

Sure, but Apple itself can't make every instruction independant on the other. Sometimes, code just needs to wait for other code to finish. Nothing anyone can do about that, and no amount of tools or frameworks is going to change that. Blocking code is blocking.

Click to expand...

Surely, there are some tasks which couldn't be multithreaded (in example, some hash functions calculation)
but, in most of these cases, the bottleneck is the performance of some other system component
(in this example, the hard drive).

nuckinfutz said:

Seriously?

Click to expand...

If Apple would screw me over like they did with Samsung, LG, Sharp, etc (companies which made millions or billions of dollars with their iDevices) they can do it to me anytime LOLLL

----------

nutmac said:

If true, and I am quite certain Apple has had OS X running on ARM for some time now, I suspect Apple won't be switching their entire Mac line-up to ARM overnight. Specifically, MacBook Airs will be the initial seed for ARM-based processors, offering iPad-like battery life and fan-less operation.

As for MacBook Pros, power users depend on virtualization (to run Windows and Linux), so I think 15" MacBook Pros, 27" iMac, and Mac Pro may stick with Intel CPUs for some time, especially if Intel manages to greatly reduce power consumption while improving performance.

Click to expand...

Not completely sure if they are still aiming that hard for the pro base, mate.... they make such tons of money with their idevices, they killed the macbook pro 17, they "iOS'exed" the OS X...

lordofthereef said:

Snip

Click to expand...

It's so hard being a smartass on the internet these days.

AppleMacFinder said:

Programmers who are creating simple applications don't use heavy multithreading,
because of low system requirements - and, as result, GCD is not needed and rarely used.

Those who are creating complex applications, which use heavy multithreading,
have enough skill to successfully use GCD and improve the overall performance greatly.
Most complex applications could be multithreaded in a big way, and, thanks to GCD, it is not that hard to accomplish.

So, in cases when GCD is really needed, it is easy to use for those who need it

Click to expand...

Before we go any farther, what exactly do you think multithreading is, and what do you think constitutes "heavy" multithreading?

Surely, there are some tasks which couldn't be multithreaded (in example, some hash functions calculation)
but, in most of these cases, the bottleneck is the performance of some other system component
(in this example, the hard drive).

Click to expand...

How would the HDD be a bottleneck in this situation? If it's being processed by the CPU, it's already cached in ram or L1 & L2. The harddrive has nothing to do with anything here.

And the reason you can't multithread hash functions is because they're single thread processes. You can't distribute the work of one process between two CPUs.

This will be awesome!

I think it is inevitable and it *could* be awesome. Check out the geekbench scores for the A6x - puts some laptops from just few years back to utter shame. While Intel processors have been going at ~30% increase YoY, Apple Ax chips (and other's mobile chips) seem to be advancing at ~100% YoY, in terms of performance. Given that trajectory, I won't be surprised if Apple can put in 4 or more of those multi-core chips and achieve on par or better performance compared to Intel desktop/laptop chips. I wonder if this will also open up the floodgates of the quarter million iPad apps to easily port into more powerful desktop class apps...

spydr said:

While Intel processors have been going at ~30% increase YoY, Apple Ax chips (and other's mobile chips) seem to be advancing at ~100% YoY, in terms of performance.

Click to expand...

Percentages, what an awesome comparison base.

Tell me, is 30% of 10,000 worse or better than 100% of 800 ?

KnightWRX said:

Percentages, what an awesome comparison base.

Tell me, is 30% of 10,000 worse or better than 100% of 800 ?

Click to expand...

100% !! It has more percentages.
-Apple's Benchmarking team

Renzatic said:

It's so hard being a smartass on the internet these days.

Click to expand...

I got where you were coming from the first time. Re-reading my response, it wasn't really even a response to yours. It was simply a continuation of mine. LOL.

namethisfile said:

except, apple is also a software company. i am sure they can program a mass amount of cores to good use in an operating system. my question is what cpu architecture is better for this mass-core/multi-core computing? is it ARM or INTEL? i know intel chips are more powerful today. but what if ARM in a couple yrs has a system where one can add cores willy-nilly to a laptop and the laptop becomes a multi-core beast as easy as one plugs in an external hdd or say add RAM to their computer?

Click to expand...

Both ARM and Intel are equally bad for massive parallelism. Fundamentally they are the same thing. The future belongs to neither of these architectures. The debate over which ISA is superior is mostly irrelevant to multicore computing.

iWolf00 said:

So Apple is planning to drop intel processors and expect intel to watch? Intel shall deliver more powerful processor than apple can do. I think we will see harsh competition between both companies in term who can build more powerful and faster chip ,,,, At the end All Hail Power Mac!!

Click to expand...

I wouldn't think of it in those terms. If Apple drops Intel, then Apple is going more like an Xbox 360 route, where there is a dedicated box with custom silicon and optimized software to provide the best performance.

So yes Intel can make a more powerful chip, but Apple can make a better performing product by making it custom and purpose specific.

Off the shelf chips are typically an easy solution, but not always the path to best performance or design goals.

If intel doesn't just watch. and make "better" CPU's we will see a "war of the chips"

ARM may be short lived in their laptops., if they do switch, in such a case, if Intel gets Apple's interest.

I don't think Intel would be too happy if Apple said "they developed Thunderbolt"

I'd kinda like to go back to the bad ole days..

I'd welcome a change in processors and architecture

G51989 said:

Yeah guess what, Intel is the best Chip maker in the world, they've already proven they can match ARM chips in power consumption, now the next step is to rape them in performance, which haswell will do.

Click to expand...

There was never any doubt that Intel can "rape" ARM in performance. The major task ahead of them is to match ARM in power consumption and efficiency. That won't happen with Haswell, but it could happen with Broadwell or Skylake or whatever the next iteration is. If you don't think that ARM can make shocking gains in the next five years then you haven't been paying attention to how quickly the field can change. Like I said, Intel is racing down in power consumption, ARM is racing up in performance. They could potentially arrive at the same place around the same time, but there's usually a clear winner and loser. Who it will be you (nor I, nor anyone) knows at this point.

If Apple wasn't working on an ARM solution for OS X at this point, they'd be stupid. You have to be ready for whatever shakes out.

brdeveloper said:

Well, in computer science classes, we have several lessons of parallel computing with specific semantics, but these concepts never reached the market in large scale. Maybe a several-core ARM processor would be the first chance where we could explicitly call these parallel primitives, which would bring new ways of thinking programming like non-deterministic-like development.

If Apple launched iOS in 2007 and it's relatively mature after 5 years, I wouldn't doubt that Apple would be capable of bringing this concept to an upcoming OS X+.

Click to expand...

Possibly, but Apple isn't the only one to be concerned about. 3rd-party programmers who write primarily for Windows, which is mainly run on 2 or 4 core computers, would have to write their programs very differently for Apple's 32-core computers. Maybe it would be OK for Apple if everyone did it.

Meaningless worry

Such woe and dread about something that is several years away, maybe even a decade. "Bought my last Mac" eh? Can't you imagine owning one or two more in the decade we wait for this to come about?

As for changing processors ruining Apple, that's exactly what Mac lovers said about the Intel switch. They saw it as Apple giving up to Windows, and none of their software would work.

As for Rosetta (OS 10.4 to 10.6) running slowly, rubbish. It was faster than Classic Mac OS, and removing PPC code with third-party software did not make a huge speed jump. Yes, dropping Rosetta (OS 10.7) added a tiny bit of speed, but it wasn't an earth-shaking improvement. Windows apps are often bloated with "reach-back" code to allow the app to run in XP, Vista, and 7. No one seems to care about that.

Ten years from now, I think we will see Windows (desktop / mobile) and Apple (OS X, iOS) holding about 30% each of the "computing device" (smart phone, tablet, notebook) market share, and "other" (Android, Linux, etc.) about 40%. Most office workers will be using tablets for their work. Is this shocking? Ten years ago, most office workers had large desk phones with big plunky buttons. Now most of them use their mobile phone for a large part of office communication. Here in China, I know no one's "land line" number. Most people don't have one. That's progress. Apple is just so far ahead of the curve we don't see where they are going until they arrive and we say "Of course! It's obvious now."

ncaissie said:

This is going to be very bad for real games and not these mini games they have on the iOS now. Good luck running Diablo 3 on an ARM processor.

Click to expand...

Yep, I bet Crytek will port Crysis 3 on ARM.

----------

dukebound85 said:

google has had vectors in their maps for some time......

Click to expand...

I think he's being sarcastic.

----------

AppleMacFinder said:

1) They have already advanced quite far at the ARM processor development route.
Already five generations in Ax family.

2) x86-64 architecture has a lot of unnecessary legacy stuff,
which takes space that could be used for better purposes.

Click to expand...

ARM is pretty old too.

And could you elaborate on the "legacy" stuff?

milo said:

Because it would break compatibility with all apps and require all new builds, just like the PPC transition all over again.

Click to expand...

There is much less difference between x86 and ARM then the difference between x86 and PowerPC. I ported a few hundred thousand lines of code from x86 to PowerPC, and the only thing that needed changing were a few bits of code that assumed "if it's not Intel, then it must be PowerPC" which was wrong. No other changes needed.

Unless you have x86 assembler code in your product, it will just work on an ARM processor.

linkgx1 said:

Um.....you sound like an ANGRY old man and...

Click to expand...

Well, I used to sound like an Angry Young Man, but I guess we all get older. I guess late '30s would be old to teenagers.

WHAT THE HELL IS A PLAYSTATION 360?

Click to expand...

It's a Playstation 3 sitting on top of an X-Box 360.

Imagination Technologies buys MIPS, Apple buys Imagination Technologies and TSMC

Steve jobs turning on hes grave

gnasher729 said:

There is much less difference between x86 and ARM then the difference between x86 and PowerPC. I ported a few hundred thousand lines of code from x86 to PowerPC, and the only thing that needed changing were a few bits of code that assumed "if it's not Intel, then it must be PowerPC" which was wrong. No other changes needed.

Unless you have x86 assembler code in your product, it will just work on an ARM processor.

Click to expand...

You're talking at a developer level. He was talking at a user level. Virtualization packages, game ports that use WINE wrappers, multi-booting OSes through Bootcamp, etc... the usual culprits have all been named.

And frankly, while for most developers it's a recompile away, most won't bother, so we'll be stuck with the same "Rosetta" situation we were with the PPC->Intel move. Except an emulator might not work so well for Intel on an ARM platform, with the ARM stuff being much less capable than the Intel stuff.

Then there's the fact that Apple slimmed down OS X a lot when they removed the PPC stuff. Fattening it up again by shipping ARM/Intel versions of the frameworks and other stuff would just be daft.

So yes, compatibility is an issue. Think at other levels than just the code. The same high level frameworks were there in the PPC days, most devs that didn't just assume endianess or had assembly code didn't have any issues either then too just hitting "compile".

So, then would this mean "Rosetta" would be back ? and used mostly ?

Not in the traditional sense, but in much the same Rosetta could run PPC, ARM would use intel (of those developers who wish to just "work" on ARM ?)

Is true, then this could also mean say 75% of people may always be running on x86 in the chip, and not running ARM native... Hence, those 75% wouldn't be getting the full benefits out of ARM.

Just like always running Rosetta apps all the time wouldn't gain the benefits of Intel.

roxxette said:

Steve jobs turning on hes grave

Click to expand...

He's laying on top of his grave?

miniroll32 said:

He's laying on top of his grave?

Click to expand...

OMG