[TweakTown] Intel claims Apple would be lost without their chips

[Gikaseixas]

Intel CFO Stacy Smith states that Apple would "have to take a big step off performance to step off our architecture"

With over 10 years of cooperation, Intel have been a mainstay in Apple hardware - taking over from Apple's co-designed chips with PowerPC due to them reportedly having issues with slimming down their laptop range, turning to the processor giant for some help.

n a recent interview with Business Insider, Intel's CFO Stacy Smith stated "for a customer like Apple you'd have to take a big step off performance to step off our architecture. That is what in essence enables us to win across different customers."

This is a rather big statement from Intel, but lets be honest here - Intel is at the top and it will stay that way for a long period of time no matter what happens.

There has been some rumors that Apple are looking to move on from Intel in order to utilize ARM architecture in their products, however nothing tangible has come to light as of yet.


Read more at http://www.tweaktown.com/news/43529/intel-claims-apple-lost-without-chips/index.html

 

[Yuriman]

All Apple software would need to be recompiled, but that has happened before. People wouldn't be happy, but they'd take it anyway because Apple.

Performance would be worse, and there aren't many reasons to use ARM in the desktop space other than price, but I don't think too many people buy Apple computers for their raw performance anyway. A sufficiently high clocked big ARM chip would provide reasonable desktop performance, and GPU acceleration would help a lot with heavy lifting.

I think Apple could pull it off.

 

[krumme]

What a crapload. CFO should generally just shut up and get back to the desk.

Look this autumn for Apple A9x. Probably a quad core on 14nm Samsung finfet process. Then go compare the benchmarks to similar intel product within tablet or reasonable tdp form factor.

Then this dude will know its not perf or perf/watt and especially battery life that hinders a shift - because intel will get beaten here - but total indirect cost and a stagnating and even decreasing laptop market that will - perhaps - make sure the shift will not happen.

Is Intel starting to think we are idiots? An cfo should know what total cost is for customers - and respect that because its what makes Intel money - instead of this bragging about perf when competitors is racing past them for perf on the mobile side and their own 14nm is late and leaking (or is it just the bw arch leaking - same result) .

 

[witeken]

Why do I have to click 2 (or 3) links to get to the actual source?

http://uk.businessinsider.com/intel-and-apple-relationship-2015-2?r=US

 

[ShintaiDK]

Apple could kiss their entire OSX series of products goodbye without Intel.

 

[coercitiv]

Then this dude will know its not perf or perf/watt and especially battery life that hinders a shift - because intel will get beaten here - but total indirect cost and a stagnating and even decreasing laptop market that will - perhaps - make sure the shift will not happen.

So they could develop a better high performance chip, but they can't due to cost... and the Chief Financial Officer of Intel did not figure this out by himself.

 

[scannall]

This is all related to the rumors that Apple will go to their own CPU's for their desktop and laptop lines. While I do like and appreciate Apple's products, this is one rumor that seems more like hot air to me. Or perhaps click baiting.

While Apple is likely a little unhappy with Intel, and their constantly moving shipment days I'm not seeing a real dollar cost benefit here. To develop their own CPU for mobile makes sense. Since the volume is so high, and they have specific targets to hit with it in performance and battery life.

But the volume of Macs is a small fraction that of mobile, so I'm not seeing a significant savings there. Plus losing Bootcamp would cause a loss of sales.

 

[ElFenix]

building ARM cores in the double-digit power range wouldn't get you any real performance advantage over intel - intel is going to have whatever fab partner beat and intel has gobs more experience building high-performance cores. ARM would probably be behind quite a bit, in fact. ARM's instruction set isn't magic. intel has already beaten everyone else trying to offer products in the double and triple digit power ranges - pretty much all of which offered processors with similar instruction sets to ARM.

 

[krumme]

building ARM cores in the double-digit power range wouldn't get you any real performance advantage over intel - intel is going to have whatever fab partner beat and intel has gobs more experience building high-performance cores. ARM would probably be behind quite a bit, in fact. ARM's instruction set isn't magic. intel has already beaten everyone else trying to offer products in the double and triple digit power ranges - pretty much all of which offered processors with similar instruction sets to ARM.

Well its kind of clickbait article so who knows what products its about lol but the new ultra thin air is mentioned. I recon its fanless and more like 5w tdp area than the usual 10-15w tdp.

But anyway. That segment is not what it used to be. Even apple tablets is in decline. So its not that interesting anyway even if software was easy to port and cost was leveled. The discussion is uninteresting.

The interesting question is if there is any more strategic advantage for apple to go arm and use their own cores?

 

[svarog19]

Apple made a mistake 10 years ago by choosing Intel and they are now suffering because of that mistake, if they have choose AMD instead of Intel then they would't have suffered at all...

That rumor that Apple is going to make own CPU is possible though there will be people mad when they switch then again AMD has opened section for their customers to order chips that comply to their needs...

If Apple switches to AMD then it would need a lot less work then if they switched to ARM, only thing they would need to do is recompile and optimizations for AMD CPU's and AMD did gave a hint about products being in place for 2016 and AMD's "Zen" is out in 2016... So Apple could switch to AMD which would use sub 20nm process nm process for Zen which is a completely new CPU that AMD is designing... There is no doubt that it will approach Ivy or Haswell in IPC or even exceed it since veterans that created CPU's that outperformed Intel's CPU's by a wide margin.

 

[witeken]

mistake suffering

Wait, are you talking about the richest company on earth?

Apple's a premium company, so it makes sense that they buy premium chips from a premium chip company.

 

[Pheesh]

Apple made a mistake 10 years ago by choosing Intel and they are now suffering because of that mistake, if they have choose AMD instead of Intel then they would't have suffered at all...

....
So Apple could switch to AMD which would use sub 20nm process nm process for Zen which is a completely new CPU that AMD is designing... There is no doubt that it will approach Ivy or Haswell in IPC or even exceed it since veterans that created CPU's that outperformed Intel's CPU's by a wide margin.

So Apple chooses AMD and they get subpar performance at higher power...and they made a mistake opting for better perf/watt by going w/ Intel?

Does not compute.

 

[avAT]

Could someone explain to me how meaningful benchmarks like Geekbench are in comparing Intel chips to A-series chips for lightweight use and for more professional use?

Purely from a performance standpoint and going by Geekbench, I have a really hard time believing Apple would ever use a Core-M processor in a notebook like they are rumored to later this year. It seems like the (currently nonexistent) A9x would outperform the Core-M chips that we have seen.

I just don't understand if I'm misguided about this. Again, this is not factoring in software compatibility, cost, etc.

 

[dahorns]

Could someone explain to me how meaningful benchmarks like Geekbench are in comparing Intel chips to A-series chips for lightweight use and for more professional use?

Purely from a performance standpoint and going by Geekbench, I have a really hard time believing Apple would ever use a Core-M processor in a notebook like they are rumored to later this year. It seems like the (currently nonexistent) A9x would outperform the Core-M chips that we have seen.

I just don't understand if I'm misguided about this. Again, this is not factoring in software compatibility, cost, etc.

I'll let use others address how meaningful the comparison is. But, even assuming it is perfectly comparable, current Core-M devices still have ~40% higher single-core performance than the A8x.

 

[avAT]

I'll let use others address how meaningful the comparison is. But, even assuming it is perfectly comparable, current Core-M devices still have ~40% higher single-core performance than the A8x.

I admit I'm a bit optimistic about the type of increase the A9x will bring, but it's irrelevant if the benchmark isn't as meaningful as I have been thinking anyway.

 

[krumme]

Could someone explain to me how meaningful benchmarks like Geekbench are in comparing Intel chips to A-series chips for lightweight use and for more professional use?

Purely from a performance standpoint and going by Geekbench, I have a really hard time believing Apple would ever use a Core-M processor in a notebook like they are rumored to later this year. It seems like the (currently nonexistent) A9x would outperform the Core-M chips that we have seen.

I just don't understand if I'm misguided about this. Again, this is not factoring in software compatibility, cost, etc.

As i have seen this the last 20 years there have been a fight for benchmark methology and always will be. Traditionally between amd and intel. Anand wrote it took years to agree on the benchmarks. I think he misfrased it. Intel won and amd had to adapt. Thats how it is. Like adult makes agreements with children lol

The strong supplier have means to get own interst in top and define what is important. And as follows what ahould be meassured. You lobby for that (dont nessesary think its always bad because its also valuable info and knowledge). You pay for it because thats how the game is.

Now eg samsung and apple is competing with Intel. Then there is the usual benchmark behind the scenes fight. imo i think when we see spec suite at AT its because its influenced and sponsered by Intel. Not in a direct way- ofcouce not - but it can bend that way. Relationships you know And eg Samsung spares no means to also outright cheat - and they favor other types of bm that makes their products look good.

I dont know how geekbench meassures, and can therefore not answer yojr question, but i am pretty sure the tactics to get spec suite into mobile will stop at a few sites. And eg Intel will not get their hands into geekbench so to speak. And if A9x is meassured here i am as you pretty sure some skywell at the same formfactor will lose big time both for perf perf/w and battery life. The use of bm suites is therefore highly important so i think its a good question to debate.

 

[coercitiv]

I admit I'm a bit optimistic about the type of increase the A9x will bring, but it's irrelevant if the benchmark isn't as meaningful as I have been thinking anyway.

It doesn't matter how meaningful the benchmarks are, year over year we're seeing significant jumps in performance from the A chips, and they are doing it with headroom to spare (frequency wise). Meantime Intel keeps peeling away from the power consumption of their strongest cores, dropping watts every year. It's a collision course.

We keep imagining a clash of the titans, Apple's big arm core vs Intel's mighty quad, when the real fight will be fought for the tablet, or to be more precise the device that will enable laptop like productivity with tablet like portability. The winner takes it all.

 

[krumme]

I admit I'm a bit optimistic about the type of increase the A9x will bring, but it's irrelevant if the benchmark isn't as meaningful as I have been thinking anyway.

I am pretty sure its more meaningfull than eg spec fp numbers.
But take a apple ip 6 and browse with it. Its damn fast imo. Whats your subjective take on it?

As for a9x its 14nm finfet. Its going to give a tremendous lift for excactly perf. But as with core the apple cores are big compared to eg A57 so cost start to become an issue. I am sure Samsung will take their part.

 

[svarog19]

So Apple chooses AMD and they get subpar performance at higher power...and they made a mistake opting for better perf/watt by going w/ Intel?

Does not compute.

Don't be ignorant... If Apple choose AMD instead of Intel in 2005 which is in period when AMD was ahead of Intel. If Apple choose AMD over Intel, they would get better performance per watt as you say so how is that not computing? Even when AMD was ahead of Intel, they were still cheaper.

Money that AMD could have earned from Apple would have allow them to invest more into foundries and not risking into a debt. Yet again, Apple made same mistake as Microsoft did, choose Intel over AMD and that's how Half Life 2 ran like garbage on original Xbox despite AMD had capability to deliver a CPU which is more than twice as powerful. If Microsoft have chose AMD then Half Life 2 would't ran like garbage and it would't be weaker than Gamecube.

 

[shady28]

Apple could kiss their entire OSX series of products goodbye without Intel.

That's kind of broad.

There are all kinds of things Apple could do to get off of Intel if you think about it, and not all of them have to do with ARM or their A8/A9 etc chips.

For example, Apple could buy AMD's entire market cap as a wholly owned subsidiary for about 2% of their cash holdings or about 3 weeks of their average earnings.

The real question is, why would they want to? I think it's a nutty proposition to get off Intel right now, unless Intel is milking them in some way.

Much more likely they'll wind up putting Intel into their tablets and phones one day, if Intel can get mobile that's good enough that is (so we know that's at least a few years off). That would make a lot more sense to me.

 

[Sweepr]

Don't be ignorant... If Apple choose AMD instead of Intel in 2005 which is in period when AMD was ahead of Intel. If Apple choose AMD over Intel, they would get better performance per watt as you say so how is that not computing? Even when AMD was ahead of Intel, they were still cheaper.

Money that AMD could have earned from Apple would have allow them to invest more into foundries and not risking into a debt. Yet again, Apple made same mistake as Microsoft did, choose Intel over AMD and that's how Half Life 2 ran like garbage on original Xbox despite AMD had capability to deliver a CPU which is more than twice as powerful. If Microsoft have chose AMD then Half Life 2 would't ran like garbage and it would't be weaker than Gamecube.

The first Xbox had a reasonably powerful NVIDIA GPU and could easily pull better graphics than GC/PS2. They wouldn't be able to run Half Life 2, Doom 3, Far Cry, Conker Live and Reloaded, Halo 1 and 2 and a bunch of titles like the Xbox 1 did. If anything the GPU and limited RAM (64MB) were the major bottlenecks in later gen 2004-2006 titles, not the CPU. Anyway, that's 100% off-topic.

AMD wasn't far ahead of Pentium M and Core Duo (Yonah) performance and performance per watt and by the time Apple ditched IBM and used Intel in actual products (early 2006) Conroe was right around the corner for desktops.

 

[Enigmoid]

Could someone explain to me how meaningful benchmarks like Geekbench are in comparing Intel chips to A-series chips for lightweight use and for more professional use?

Purely from a performance standpoint and going by Geekbench, I have a really hard time believing Apple would ever use a Core-M processor in a notebook like they are rumored to later this year. It seems like the (currently nonexistent) A9x would outperform the Core-M chips that we have seen.

I just don't understand if I'm misguided about this. Again, this is not factoring in software compatibility, cost, etc.

Geekbench is incredibly terrible.

For integer, there is a ton of encyption which is rather useless. Tons of compression, very little actual integer tests.

Memory tests are pure garbage. Nothing tests latency or bandwidth per transfer size. Basically the same bandwidth test with a (add number, copy number, etc).

http://forums.anandtech.com/showthread.php?t=2404965&page=4

Not too sure about FP but I will say (and you can look back on forum posts) that they do no to little optimization. For example SGEMM and DGEMM are terribly implemented and not at all indicative of real world performance. For example, AVX let alone AVX2 is not supported to any real extent despite having launch day support for ARM8 on the A7.

With AVX Ivybridge can perform 8 DP operations per clock. Thus for a chip operating at 3.5 ghz the result is a peak performance of 28 GFLOPS per core, double this for haswell (56 GFLOPS core/ 224 GFLOPS total). The result (DGEMM) on Geekbench? About 3.3 single core and 15.5 GFLOPS MT on a 4770. The problem is that generally written code with AVX extensions will have no trouble hitting 50% of peak flops, with the best implementations around 80%.

As said by one of the developers of GB on this forum (paraphrased): "We designed our code to simulate real-world usage."

Yet what they have done is most undoubetedly so. Trying to run the same code, without any care for its implementation (ie GB runs the same DGEMM code on x86 and ARM according to him) is stupid; no body who actually runs DGEMM on an x86 CPU is going to code their application that way.

http://support.primatelabs.com/disc...performance-of-4770k-across-linux-and-windows

We certainly agree that the same hardware should perform similarly and we strive for this. We want the scores to represent the hardware performing well, but we also intend the scores to reflect execution performance of real-world application code. We expect that a programmer will write their code once and compile that same code for each platform that he supports. We choose not to use optimized vendor libraries for workloads such as GEMM and FFT since we expect such libraries to be optimized for each target architecture making a direct comparison of the score troublesome. Furthermore, if the libraries are proprietary we don't know exactly what optimizations the library performs.

This is complete Bollocks. Most programmers are not coding cross platform x86 and ARM (the only ones who do so are app developers). Programmers have no trouble using and finding good libraries for speedups. They are also specifically unoptimizing code that in the real world great care is taken to optimize.

This completely shows why GB is trash.

To make it even worse GB, as I said before, supports ARMv8 on the day of launch while leaving AVX and AVX2 behind.

 

[Blitzvogel]

Don't be ignorant... If Apple choose AMD instead of Intel in 2005 which is in period when AMD was ahead of Intel. If Apple choose AMD over Intel, they would get better performance per watt as you say so how is that not computing? Even when AMD was ahead of Intel, they were still cheaper.

Money that AMD could have earned from Apple would have allow them to invest more into foundries and not risking into a debt. Yet again, Apple made same mistake as Microsoft did, choose Intel over AMD and that's how Half Life 2 ran like garbage on original Xbox despite AMD had capability to deliver a CPU which is more than twice as powerful. If Microsoft have chose AMD then Half Life 2 would't ran like garbage and it would't be weaker than Gamecube.

I've never seen anything on GC impress me more than the Xbox except maybe Rogue Squadron. P3s were powerhouses, and the amount of physics driven gameplay in many Xbox titles trumped most of the GCs library by a pretty large amount. And as said by someone else, HL2's problem more likely stemmed from limited RAM, not the CPU. Also, the AMD chip originally set for the Xbox was no faster than the P3, however AMD did design the north bridge if I remember correctly, and it ended up still being used in the system.

As for Apple and ARM............mmmmm not anytime soon. It's certainly feasible they could be designing big ARM cores in secret, but what they have with Intel is just as good if not better for now. Apple buying AMD would be a cheap purchase, where not only would they gain access to AMD's CPU tech, but their graphics tech which could prove useful in making GPGPU a truly integrated part of computing. Apple has the cash and clout to make that a reality. Not sure how well dual FirePro D700s are working out for the new Mac Pros, but there is no doubt that there is a **** ton of capability there that hopefully Apple is investing time and energy into understanding and making use of.

 

[Zodiark1593]

Intel CFO Stacy Smith states that Apple would "have to take a big step off performance to step off our architecture"

With over 10 years of cooperation, Intel have been a mainstay in Apple hardware - taking over from Apple's co-designed chips with PowerPC due to them reportedly having issues with slimming down their laptop range, turning to the processor giant for some help.

n a recent interview with Business Insider, Intel's CFO Stacy Smith stated "for a customer like Apple you'd have to take a big step off performance to step off our architecture. That is what in essence enables us to win across different customers."

This is a rather big statement from Intel, but lets be honest here - Intel is at the top and it will stay that way for a long period of time no matter what happens.

There has been some rumors that Apple are looking to move on from Intel in order to utilize ARM architecture in their products, however nothing tangible has come to light as of yet.


Read more at http://www.tweaktown.com/news/43529/intel-claims-apple-lost-without-chips/index.html

Considering the perceived potential threat from ARM, I don't think Intel would make such a boast heedlessly if they weren't absolutely sure they could back it up, especially toward a formidable tech giant such as Apple.

 

[Arachnotronic]

There is no doubt that it will approach Ivy or Haswell in IPC or even exceed it since veterans that created CPU's that outperformed Intel's CPU's by a wide margin.

I don't even...