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

Page 12 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Idontcare

Elite Member
Oct 10, 1999
21,118
59
91
I could see them continue to scale-up their iPad technology into the laptop-lite realm.

Heck all they need to do is take their iPad 2 Air and slap on a keyboard ala your standard clamshell laptop design of today and they'd have something that a certain percentage of existing mac users would find to be a satisfying and suitable replacement for their mac.

There is no reason this has to be an either/or fork in the road. Apple can continue making high-end macs with x86 based chips while making arm-based laptops that are scaled up ipads in the same way that ipads are scaled up iphones.
 

III-V

Senior member
Oct 12, 2014
678
1
41
I could see them continue to scale-up their iPad technology into the laptop-lite realm.

Heck all they need to do is take their iPad 2 Air and slap on a keyboard ala your standard clamshell laptop design of today and they'd have something that a certain percentage of existing mac users would find to be a satisfying and suitable replacement for their mac.

There is no reason this has to be an either/or fork in the road. Apple can continue making high-end macs with x86 based chips while making arm-based laptops that are scaled up ipads in the same way that ipads are scaled up iphones.
Yeah, I'd imagine that if Apple wants to replace Intel, it'll be in phases. They won't completely oust them overnight... for them to do that would be quite the feat, to say the least. It makes much more sense for them to slowly work their way up the TDP spectrum.
Binary translation would have been an acceptable solution for Apple in the PPC days when Macs were a niche, but given how popular MacOS is today among "laymen" and how many MacOS users like the ability to run Windows, this would be a disaster, IMHO.
Very, very good point. I'd completely forgotten about bootcamp... yeah, that's going to be a deal breaker for a lot of people, including myself.
 

krumme

Diamond Member
Oct 9, 2009
5,956
1,595
136
The ip6 have eg pretty good browsing speed for my subjective impressions. And thats in a pretty small tdp form factor.

How big is apple laptop market anyway and forecast?
 

RampantAndroid

Diamond Member
Jun 27, 2004
6,591
3
81
I could see them continue to scale-up their iPad technology into the laptop-lite realm.

Heck all they need to do is take their iPad 2 Air and slap on a keyboard ala your standard clamshell laptop design of today and they'd have something that a certain percentage of existing mac users would find to be a satisfying and suitable replacement for their mac.

There is no reason this has to be an either/or fork in the road. Apple can continue making high-end macs with x86 based chips while making arm-based laptops that are scaled up ipads in the same way that ipads are scaled up iphones.

Some might prefer that, but I'd argue they already HAVE it then, with the ipad keyboard cases (costing less than a Macbook Air.) I tried this route myself for a bit...and it got maddening at times. No multitasking, switch apps and the other is now killed etc.

For this to work, iOS needs to move towards OSX, which I doubt it will really.
 

TreVader

Platinum Member
Oct 28, 2013
2,057
2
0
I could see them continue to scale-up their iPad technology into the laptop-lite realm.

Heck all they need to do is take their iPad 2 Air and slap on a keyboard ala your standard clamshell laptop design of today and they'd have something that a certain percentage of existing mac users would find to be a satisfying and suitable replacement for their mac.

There is no reason this has to be an either/or fork in the road. Apple can continue making high-end macs with x86 based chips while making arm-based laptops that are scaled up ipads in the same way that ipads are scaled up iphones.


Good point, it is possible they will keep Intel for their Mac Pro and maybe even their MacBook Pro lines and just replace the MBA.

I agree. I have an iPad Air 2 and it is plenty fast, faster than my 2011 MacBook Air.
 

tential

Diamond Member
May 13, 2008
7,355
642
121
I could see them continue to scale-up their iPad technology into the laptop-lite realm.

Heck all they need to do is take their iPad 2 Air and slap on a keyboard ala your standard clamshell laptop design of today and they'd have something that a certain percentage of existing mac users would find to be a satisfying and suitable replacement for their mac.

There is no reason this has to be an either/or fork in the road. Apple can continue making high-end macs with x86 based chips while making arm-based laptops that are scaled up ipads in the same way that ipads are scaled up iphones.

One of my first posts in this thread addressed this. If Apple was to start to bring it's ARM processor into it's other lines, they'd start with an Ipad Air convertible design, and test the waters there.
 

jpiniero

Lifer
Oct 1, 2010
14,844
5,457
136
There is no reason this has to be an either/or fork in the road. Apple can continue making high-end macs with x86 based chips while making arm-based laptops that are scaled up ipads in the same way that ipads are scaled up iphones.

I haven't seen a percentage breakdown, but I suspect that Apple's desktops are a very tiny percentage of sales, and the iMac is basically a laptop anyway with a higher TDP processor in it.

Apple sold 5.5M Macs last quarter.
 

imported_ats

Senior member
Mar 21, 2008
422
63
86
You don't get my point do you? You take a Haswell core & then add more Haswell cores, & cache, depending on the workload. That's why more Mobile chips generally have 2 cores, 4 cores for the desktops & more for server chips. The fact is you can always work around a good design & pretty much always will do better with (more) good cores.

There is a lot more to it than just stamping out cores. The designs of the mobile/desktop part are significantly different from the workstation/server parts.

There is a reason why the metal api was released just for the iOS & that the A8X CPU cores occupy such a vast majority of that die when pretty much everyone else, including Intel is moving in the opposite direction with more die space for IGP.

The cores on the A8X do not occupy the majority of the die. In fact, they occupy less die than the GPU or the system logic. Your basis thesis is flawed.



Tell me it ain't so & that what I'm proposing is simply not feasible or better yet that it's out of the realms of possibility.

Anyone bringing geekbench into a discussion about actual performance pretty much loses all credibility, esp when that discussion has centered around real world parallel scaling. First of all, geekbench is a horrible benchmark suite in pretty much any regard. Second, 50% of the score of the composite is derived from completely unrealistic embarrassingly parallel running of the basic kernels. Geekbench composite scores are therefore completely irrelevant to the actual real world and real applications. It would be like comparing mobile/laptop/desktop processors using SpecRate. Its the wrong benchmark to estimate performance for the workloads actually run.

My only point is that it can be done so long as Apple's gunning for it, the rest is just a technical limitation & not a hard set physical one.

No, you are confusing parallel performance for actual real world performance. Here's a hint, most real world application are pretty much dependent on single context performance in the mobile, laptop, and desktop world. Unless the underlying workloads can actually make significant use of additional hardware contexts, adding more cores simply will result in lower actual delivered performance.
 

imported_ats

Senior member
Mar 21, 2008
422
63
86
I haven't seen a percentage breakdown, but I suspect that Apple's desktops are a very tiny percentage of sales, and the iMac is basically a laptop anyway with a higher TDP processor in it.

Apple sold 5.5M Macs last quarter.

Which is all the more reason they won't make the switch at least for quite some time if not ever. To support even somewhat performance competitive designs for even the Macbook Air, would require basically a new core design and associated chip. The volumes for the Mac line, unlike the iOS line cannot support the costs required.

Even if we assume that the volume for the new part is 5.5M, you are looking at ~1B in NRE for those parts. That works out to ~$180 per part just in NRE. Add in production, test, packaging and assorted ancillary costs, and you are looking at a raw cost in the $250-$350 range. At that point you aren't saving anything over buying parts from Intel. And you have significantly more risk and overhead.

The reason that the A parts work is that they are able to spread the NRE over 100s of millions of chips. Just like Intel is able to spread the costs for their NRE over 100s of millions of chips for mobile, laptop, and desktop. The parts that Intel makes that can't be spread over such large volumes, have significantly higher prices to cover the NRE required.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Intel loses maybe 5-8% revenue short term and gains it back as people abandon Macs in droves... Apple was close to 90% of IBMs volume. Apple is at best 10% of Intel's volume and less in revenues.

Don't underestimate the impact of losing a halo, and high-margin vendor. No doubt some still believe Intel still has a chance of winning Apple designs, not just fabbing for their chips. IMO, Apple abandoning Intel will kill ALL hopes, even foundry.

How much emotional impact it'll have on other vendors? That'll translate into financial impact sooner or later.
 

Mondozei

Golden Member
Jul 7, 2013
1,043
41
86
Anyone bringing geekbench into a discussion about actual performance pretty much loses all credibility, esp when that discussion has centered around real world parallel scaling. First of all, geekbench is a horrible benchmark suite in pretty much any regard. Second, 50% of the score of the composite is derived from completely unrealistic embarrassingly parallel running of the basic kernels. Geekbench composite scores are therefore completely irrelevant to the actual real world and real applications. It would be like comparing mobile/laptop/desktop processors using SpecRate. Its the wrong benchmark to estimate performance for the workloads actually run.

Interesting, which benchmark would you prefer and why?
 

Lonyo

Lifer
Aug 10, 2002
21,939
6
81
How about real world application benchmarks for the sort of program you would be using a Mac for rather than an iPad? i.e. content creation rather than consumption, including basic things like manipulating 2D images, encoding/editing videos, performing large/complex Excel calculations etc.
 

krumme

Diamond Member
Oct 9, 2009
5,956
1,595
136
Which is all the more reason they won't make the switch at least for quite some time if not ever. To support even somewhat performance competitive designs for even the Macbook Air, would require basically a new core design and associated chip. The volumes for the Mac line, unlike the iOS line cannot support the costs required.

Even if we assume that the volume for the new part is 5.5M, you are looking at ~1B in NRE for those parts. That works out to ~$180 per part just in NRE. Add in production, test, packaging and assorted ancillary costs, and you are looking at a raw cost in the $250-$350 range. At that point you aren't saving anything over buying parts from Intel. And you have significantly more risk and overhead.

The reason that the A parts work is that they are able to spread the NRE over 100s of millions of chips. Just like Intel is able to spread the costs for their NRE over 100s of millions of chips for mobile, laptop, and desktop. The parts that Intel makes that can't be spread over such large volumes, have significantly higher prices to cover the NRE required.

Where do you get the 1b?

Idc: What is your estimate for nre?
 

turtile

Senior member
Aug 19, 2014
618
296
136
I think a custom part from AMD could be used to drive a super high resolution display. I'm not sure it's worth converting Mac OSX to run on ARM (along with all of the other software).

Intel has already made the IrisPro...
 

imported_ats

Senior member
Mar 21, 2008
422
63
86
Interesting, which benchmark would you prefer and why?

Honestly none, outside of browser benchmarks the mobile space is a cesspool of bad benchmarks and benchmarking practices. Part of that is that until recently mobile space didn't even have enough memory to even think about porting over more robust suites from the laptop/desktop space. Then there is also the issue of what are actual useful workloads for mobile phones and laptops...

Browsers are somewhat covered but even there you have significant issues with the benchmarks not fully replicating actual user use patterns, esp wrt multiple tabs. And lets be honest, the vast majority of mobile apps are glorified browsers...

Games make generally poor benchmarks in mobile because there are generally lots of differences between platforms. Within platforms they can be a decent proxy, but they mostly stress the GPU.

Ideally, something like Spec2006 would be run as that give a decent variety of interesting workloads and application patterns.
 

TreVader

Platinum Member
Oct 28, 2013
2,057
2
0
Honestly none


That's because there are no benchmarks which support your argument that apple is behind at all. Not in performance per watt. That's what matters.


Geekbench always gets hated on by the Intel Support Group because it's one of the few benchmarks that can compare apple Intel AT ALL. Intel should never be compared, according to them, because they are without peers.
 
Last edited:

imported_ats

Senior member
Mar 21, 2008
422
63
86
Where do you get the 1b?

Idc: What is your estimate for nre?

1B is basically an educated wag. Its unlikely you are doing a competitive laptop/desktop processor or less than a Billionish these days. You are likely looking at a 4-5 year development cycle from inception to ship. You'll probably looking at an average of 200+ people. You'll have various licensing costs as well. Etc. Its somewhere in the range of 500m-1b in NRE. A lot of its going to depend on fully loaded costs per person. Plus you'll have all the fab related NRE costs.
 

imported_ats

Senior member
Mar 21, 2008
422
63
86
Don't underestimate the impact of losing a halo, and high-margin vendor. No doubt some still believe Intel still has a chance of winning Apple designs, not just fabbing for their chips. IMO, Apple abandoning Intel will kill ALL hopes, even foundry.

How much emotional impact it'll have on other vendors? That'll translate into financial impact sooner or later.

What's the halo? And I'm pretty sure that Apple isn't nearly as high margin as you think it is wrt Intel.

Intel foundry is pretty orthogonal to Intel CPUs being in macs.

And it would have basically no emotional impact on other vendors.
 

Nothingness

Platinum Member
Jul 3, 2013
2,769
1,429
136
Honestly none, outside of browser benchmarks the mobile space is a cesspool of bad benchmarks and benchmarking practices. Part of that is that until recently mobile space didn't even have enough memory to even think about porting over more robust suites from the laptop/desktop space. Then there is also the issue of what are actual useful workloads for mobile phones and laptops...

Browsers are somewhat covered but even there you have significant issues with the benchmarks not fully replicating actual user use patterns, esp wrt multiple tabs. And lets be honest, the vast majority of mobile apps are glorified browsers...
Browser benchmarks are terrible to compare CPUs. The variation across different browsers on the same CPU hints at that.

Games make generally poor benchmarks in mobile because there are generally lots of differences between platforms. Within platforms they can be a decent proxy, but they mostly stress the GPU.

Ideally, something like Spec2006 would be run as that give a decent variety of interesting workloads and application patterns.
What part(s) of SPEC CPU 2006 are representative of what users are doing?
 

imported_ats

Senior member
Mar 21, 2008
422
63
86
That's because there are no benchmarks which support your argument that apple is behind at all. Not in performance per watt. That's what matters.

Your assumption of perf/w being all that matters is hilarious and ignorant. If perf/w was all that matters, phones wouldn't being using customer cores, A9/12/15/et al. They would be using the lower end ARM chips which are much more efficient at perf/w.

And there are plenty of benchmarks that do support it. Even bad one. None of the mobile products are on the same performance levels as the laptop/desktop products currently in the market. And they aren't really even that close.


Geekbench always gets hated on by the Intel Support Group because it's one of the few benchmarks that can compare apple Intel AT ALL. Intel should never be compared, according to them, because they are without peers.

No, geekbench is hated because its about as useful as Dhrystone. Its a collection of largely uninteresting kernals that have little to no impact on actual application performance and its omnibus number is completely useless as it doesn't tell you anything because its trying to combine too many disparate aspects including roughly half its value being pointless embarrassingly parallel runs. Its working sets are far too small, even by mobile standards, to be useful. It doesn't model at all any real application load. Nor does it have any actual research behind the including of its parts or tuning of those parts. And finally, the whole thing is incredibly opaque.

If you want to do actual comparisons, run Spec with various compilers or use real application based benchmarks.
 

imported_ats

Senior member
Mar 21, 2008
422
63
86
Browser benchmarks are terrible to compare CPUs. The variation across different browsers on the same CPU hints at that.

Um, that's because they are running different code sets as should be obvious. Given the same browser however, its a reasonable approximation of what the majority of phone applications actually do.

What part(s) of SPEC CPU 2006 are representative of what users are doing?

GCC
xalancbmk
astar
sjeng
gobmk
mcf
bzip2
perlbench
sphinx3
 

TreVader

Platinum Member
Oct 28, 2013
2,057
2
0
Your assumption of perf/w being all that matters is hilarious and ignorant. If perf/w was all that matters, phones wouldn't being using customer cores, A9/12/15/et al. They would be using the lower end ARM chips which are much more efficient at perf/w.

And there are plenty of benchmarks that do support it. Even bad one. None of the mobile products are on the same performance levels as the laptop/desktop products currently in the market. And they aren't really even that close.




No, geekbench is hated because its about as useful as Dhrystone. Its a collection of largely uninteresting kernals that have little to no impact on actual application performance and its omnibus number is completely useless as it doesn't tell you anything because its trying to combine too many disparate aspects including roughly half its value being pointless embarrassingly parallel runs. Its working sets are far too small, even by mobile standards, to be useful. It doesn't model at all any real application load. Nor does it have any actual research behind the including of its parts or tuning of those parts. And finally, the whole thing is incredibly opaque.

If you want to do actual comparisons, run Spec with various compilers or use real application based benchmarks.
A8X/A8/A7 all run exceptionally well in real-world benchmarks. I'm sure you've used an iPhone 5s or 6, or iPad air 2 and noticed how quickly they run. They also beat intel in perf/W in javascript benchmarks. There are literally NO cross platform benchmarks that intel wins when comparing A8X at 3.5-4W and Haswell at 15W in perf/W. Please show me where I am wrong here.


You are correct that Apple has not, and probably can't for the moment match intel in peak performance. That's not what this topic is about, we are talking about a Macbook Air 12" Retina. The discussion is whether or not a A9 ARM Apple designed processor could serve to give adequate performance. In what way is the iPad Air 2's performance not adequate for that kind of device? This would obviously have a higher clocked, higher TDP processor than an iPad.


Apple probably won't ditch intel completely, not all at once anyways. Intel still offers performance that peaks at 5-10x that of peak A8X in their multi-processor Xeon systems. However, that's at 20-50x the peak power draw. So they are not suitable for mobile stuff.


There is no reason for apple to continue using intel in a 7MM 12" form factor. It is, in fact, a poor engineering choice to put a 15W Broadwell or even "4.5W" Broadwell (as evidenced by insane throttling in Yoga 3 pro, a 12" 8mm tablet, even at 3.5W) in this new macbook Air.


There is just no way to engineer your way around that loss in battery volume (over 50%, remember the macbook air was 13" this is 12" and this is also half the thickness) and TDP capacity. Look at what Lenovo did. They are not retards.


IF apple do manage to fit broadwell-U into that form factor with good performance, I will be amazed. That would be an epic engineering feat.
 

Nothingness

Platinum Member
Jul 3, 2013
2,769
1,429
136
Um, that's because they are running different code sets as should be obvious. Given the same browser however, its a reasonable approximation of what the majority of phone applications actually do.
Has iOS Safari been ported to other OS? That'd be a prerequisite to compare Apple CPU against other since as far as I know Apple doesn't allow JIT on its OS, so other browsers are limited in what they can do.

GCC -> compiler
xalancbmk -> XML
astar -> path finding
sjeng -> Chess
gobmk -> Go
mcf -> Simplex
bzip2 -> compression
perlbench -> PERL scripts
sphinx3 -> voice recognition

Odd choice of benchmarks for sure: at least two of astar, sjeng and gobmk should go away or you are over-representing games. Simplex, well no comment... PERL, gcc, almost no one use those unlkess they are developers.

I'm surprised you didn't include h264 which is likely more representative than mcf, perl or gcc, even though few people do software encoding.

BTW most of these SPEC benchmarks put *very* little pressure on instruction caches which is an issue. As an example mcf is less than 3K lines of code. And none of them does runtime code generation which has become very important (though that'd be partly covered by browser benchmarks).

Benchmark selection is a real pain :biggrin:
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |