Apple in advanced discussions to adopt AMD chips - Page 7

Ksec

I do not believe some of your points are correct.

3. Ati has not lagged behind Nvidia in GPU recently. In fact the leadership crown has been flip flopping between the two companies for a while.

5. AMD Is a lot cheaper than Intel at certain levels. See my first post in this thread.

6. AMD Fusion is still superior. Better GPU with better support for OpenCL. I dont see much reason to put up with Intel's slow IGP.

AMD makes sense at the low end. People buy Mac mini or Macbooks because they represent good value & performance for the money. AMD would do well for these lines.

Have you not heard of ARM A9 SoC? I think Nvidia calls them tegra or something. Can't say they're supercomputer capability chips, but they are sibling/upgrade in design to Apple's A4 SoC.

Were we talking about mobile supercomputers? For the required function for mobile computer, all of them will do fine, including VIA and Nvidia.

I'm not following you here.

To be clear, ARM A9 (Core) = Cortex (Family) = ARM v7 (Architecture Version).

Also, Tegra first arrived using ARM11 = ARM v6 ≠ Cortex.

The next release is Tegra 2 and will ARM Cortex-A9 MPCore, which is 2-4 cores. Apple's A4 is ARM Cortex-A8, which is a single core. Both are ARMv7-A.

•

http://www.nvidia.com/object/tegra_250.html
• http://en.wikipedia.org/wiki/Tegra#Tegra_2xx_series
• http://en.wikipedia.org/wiki/ARM_Cortex#ARM_cores

I am referring to GPGPU aspect. Not Gaming. GPGPU are more about Drivers and Software rather then the Actual Hardware. ATI is lacking in this area especially on the Mac platform. Both Nvidia and ATI is equally OpenCL compliant at this time.


Certain Levels. Yes, that is Workstation and Servers. As your first post stated. Intel has clear advantage in that area because server Admin are reluctant to choose AMD over Intel. Therefore Intel can Mark the price what ever they want.
In Desktop / Laptop OEM Market, Intel as a Platform ( Chipset, Wireless ) is not that much more expensive then AMD. People continue to think AMD are cheaper due to its retail pricing.


There is the Key problem to Fusion. ( One which everyone seems to miss )
It uses old CPU core and Old APU ( GPU ) core.
Its CPU is the current phantom and not Bulldozer with some tweaks for lower power consumption.
Its APU / GPU is the OLD Radeon 4000 Series. A lot of people still find it strange that a CPU coming out in 1 years time will be using technology that is 3 years old when it launch. And more importantly Radeon 4000 is actually not OpenCL 1.0 hardware compliant. While there is OpenCL driver for it, there are certain function which uses CPU as emulation. ( Much like how Intel provide Shader Support on its IGP ). Therefore OpenCL on Radeon 4000 is very slow.

Thanks for clearing that up.

http://www.anandtech.com/show/2871/3




Fusion is not out yet so there's no comparison to be made but AMD/ATI is certainly going to evolve quickly whether its CPU advances or GPU advance. Intel's lack of IGP technology means they have to focus soley on CPU improvements because they continue to fail in the GPU arena.

So, we are settle that first gen fusion wont work on Apple. By 2012 AMD may have a chance of a product that suit apple. 2 years is quite a long time in technology industry. 2012 we have IvyBridge, which offer FMA as well as Double the Graphics of SandyBridge.

Fusion, by look of it, would work like Intel's tick tock model. Where they upgrade the core of APU and CPU every alternate year.

So if the best happen that 2nd Gen Fusion have both new Bulldozer CPU Core and Radeon 5000 APU / GPU. Intel 's IvyBridge, will have a better CPU and unknown GPU performance.

Side Note. The real problem with Intel IGP has never been its hardware. But Intel 's Drivers.

Bulldozer will support AVX, FMA4, XOP & CVT16. AMD has never lacked for supporting the SIMD stuff. Intel's lead is basically in process technology, power and fab capability with strength in Integer.

Bulldozer, as you say before, should rectify the Integer performance and if their GPU technology continues to evolve nicely they won't have to worry about FPU performance.

Apple would be smart (IMO of course) to bring in AMD on the low end in 2011 and remain competive.

$999 for the Macbook isn't worth it. I think it's a $899 at best laptop.

In 2012 I'd like to see.

• 5x7 iPad model for $349
• 9.7 iPad models starting at $499 with Dual Core ARM based processing
• Mac mini Fusion - $499
• Macbook Fusion - $799

That brings Apple in at every price point.

I don't think Apple has any problem with remaining competitive. They are making money hand-over-fist with excellent growth each quarter. As a shareholder I wish the growth was slower to keep the expectations down a bit to help sustain growth.


If it was at $899 wouldn't people just assume it's overpriced and say "I think it's a $799 at best laptop." That seems to be the way these things works.


That is 20% lower than the current MacBook price. Does AMD Fusion really account for that or are there are aspects needed to lower the price, like a cheaper LCD, no LED backlight, smaller, less dense battery, etc.?

Then there are the economic issues to deal with. Apple would have to get a much higher profit per unit to maintain the same profit per unit at the $999 price. That isn't as easy to do the cheaper you go. They can make up for it with increased unit sales but that isn't something you choose to do until you saturate a market at a certain price point. The next tier would then be $899, not $799, and as you enter into a lower price point the market gets larger and it will likely take longer before you need to expand again. So far, I have seen no plateauing of the Mac notebook line to account for a drop in price by 20% to attract new buyers.

Just hopped on Best Buy

Dell
Sony
HP

They all have $999 units with Core i5 and some i7 models. 4GB of RAM and 500GB HDD on the avg.

I thought to myself "well Apple's not doing badly with the Core i3 and the GPU switching tech" and then it dawned on me that I was thinking about the MBP 13" specs. (oops
the MBP 13 is still a C2D)

Here's the newest Macbook


This is not getting it done. I was all set to have the "talk" with my mother about looking at the Macbook assuming that it'd be $799 based on these specs leaked from Microcenter. When it came out at $999 I knew I wasn't going to make a peep.

I'll recommend a Mac mini instead and mate it with a HD display. Apple mailed it in with this last Macbook refresh.

I'm not understanding your point. It's not that Apple can't offer a cheaper system and/or a system that makes less profit, it's that they don't have to because they offer things other vendors can't touch which sets them apart from the rest.

It's clear that very few PC buyers understand or care about these HW specs. They just want a machine that does what they need it to do.

All things being equal Core-i7 for $999 is better than C2D for $999, but all things aren't equal. The MB/MBPs tend to be thinner, having larger batteries yet weight considerably less, have superior construction, have tech support that promotes consumer confidence, and use an OS that consumers tend to find friendlier.


Before we start comparing Dell, Sony and HP to Macs why don't we first compare a $999 Core-i7 Dell, Sony or HP with a more expensive Core-i7 PC from the same vendor. Case in point, Sony Viao Z w/ Core i5. Starts at $1899. Want Blu-ray, add 500 (yet a full Blu-ray player for your HDTV is under $100). Want Core-i7, that's $150 more (yet it's the same exact socket as Core-i5 and only costs $75 more according to Intel's per/1000 unit price list).


The 13" MBP is C2D with Nvidia 320M, which is was the right way to go this time when you look at the specs for the Core-i3. Next time, though, they'll have to move to Core-i, but I suspect it will be Core-i5 and require a major revision.

I'm surprised that everyone is looking at the AMD option so narrowly. I'd imagine Apple would go for a major license investment and custom build x64 processors with AMD for certain machines. I would imagine this is what would be required for a well designed touch iMac or Mac Mini. A custom x64 would also be really nice for a future version of OSX redesigned with touch in mind. That would be they only way AMD could match Apple's design philosophy of building the best computing machines out there.

Any other scenario wouldn't make sense with Apple's design philosophy. Cost isn't the issue when people happily pay $2,000+ for a MacBook Pro. Apple would need a particular advantage over Intel, which AMD does not offer right now. Acquiring even standard AMD processors for low end machines would mean a significant performance per watt advantage over Intel to be the best machine.

How this ALL ties into AMD is this.

The MBP are more either more expensive than they need to be or spec'd lower than they need to be because adding two GPU to a computer IMO is a waste. If Intel's IGP graphics were decent then there would be no need to expend the finances and engineering resources on adding in discrete graphics.

Now granted discrete graphics have always been preferrable to IGP but I wonder if that will be the case going forward as we move from 32nm process and down (within the context of portable computing)


AMD may be able to offer some custom stuff for Apple depending on how much of Apple's business they'd get in on. I'd love to have further confirmation that AMD and Apple are having advanced discussion.

I don't believe Bulldozer is fully compliant to the final AVX spec, partially due to time/design constraints on AMD's part since Bulldozer was originally designed for their defunct SSE5 and partially because Intel was still tweaking the AVX spec as Sandy Bridge's design evolved so AVX was finalized late. This will probably require additional effort on the part of Apple in the compiler and software developers in general to support the different implementations. As well, Bulldozer's AVX implementation will probably be slower than Sandy Bridge's because Bulldozer does not have native 256-bit execution units and instead requires combining the 2 128-bit units in each module (2 cores). Sandy Bridge has multiple 256-bit execution units per core.


Why would Apple need to move to AMD in order to offer models with lower price points? If Apple doesn't want to reduce prices by reducing their profit margins on existing models and components, Intel also offers a variety of lower-end components that Apple could use in cheaper models.


I really hope Apple isn't considering moving to AMD simply to eliminate the need for discrete GPUs. On die IGPs are going to get faster but they are not going to outperform discrete GPUs. The transistor, heat, and power budget of on die IGP are always going to be constrained compared to discrete GPUs so an IGP won't truly replace mid-high end discrete GPUs. Fusion is a good reason to improve 13" MacBook Pros where you are space constrained but not to waste the additional thermal headroom that the larger 15" and 17" form factors afford to put in more powerful discrete GPUs.

It remains to be seen how Bulldozer's duoply 128-bit units function. I think for Apple it would be trivial to support both depending on how much of Clang/LLVM Apple's able to integrate into Xcode 4.x and on. I'm pretty confident that both AMD and Intel will be strong here. I'm betting that Wolfram is pretty excited about what 2011 is going to bring for serious number crunching.



Intel's lower cost items perform almost unacceptably low in graphics. Basically you get a great CPU and so so GPU. Now I'm still not sold on OpenCL (primarly because i've not seen one decent demo that shows me the promise of GPGPU computing) but i've got to think that we are on the cusp of seeing the GPU finally become a peer processing unit for general users. If that's the case AMD closes the gap with Intel in performance (due to superior GPU) and we know they'll come in cheaper overall in price. Performance/Watt though is going to be very important.

I agree with you. I'd love to see Fusion on lower end space constrained product and Intel on the midrange and perhaps both on the high end.

It behooves us all to have two CPU vendors that can push X86 forward since PPC is done as a high volume desktop platform.

Not in multithreaded situations.

Clock speeds and price also play a part.

Melgross;

I see that you have refused to respond to my legitimate questions before; in case you missed them, here is the link:
http://forums.appleinsider.com/showp...0&postcount=99

These are not rhetorical questions, and warrant real answers based on your claims.

If there is no response, I guess we will have to take that your writings in this thread are largely baseless and simply come from your personal misconceptions.

Yes, i already said Bulldozer version of Fusion will compete with IvyBridge, not SandyBridge. Which makes the matter worse. And by AMD's track record, they have been missing target deadline all the time.


As someone point out. Apple doesn't need to lower price to Stay competitive. As a business model, Apple are like LV and Gucci. Did you ever see LV and Gucci goes on Sale? Lower Price Point? These are premium brand. Smaller Mid Class Brand may make 10 - 20 times the volume and revenue, but they still make LESS TOTAL profits then Premium brand.


The problem are not within the component price. The highest cost of BOM inside all Macbook Pro and iMac is actually aluminum. Did you see ANY Price Range of Laptop out there build with a Single Block of Aluminum? Are there any laptop that gives you a feeling this thing is not going to break easily, because it is METAL!. Not Plastic.

I think all analyst were too busy with iPhone sales and iPad. Has no one realize the impact of Aluminium price risen by more then 50% in one years time?? Aluminium was a more expensive material to start with, and its price has only gone up. The unibody manufacturing process cost has also gone up.The Fixed Cost of MBP are much higher then other Laptop to Start with. That is why MBP cant get much lower price. And using AMD fusion wont save you $100 dollars.

For Lower Cost Computer, that is where Apple A4 ( ARM ) and iPad comes into play.


Again as stated before. Two GPU in a computer is now an ideal way to go forward. Because of the way GPU works and designed, unless there is a major breakthrough you can never get a Powerful GPU to acceptable Idle power. You need a GPU for Low power ( GPU inside CPU ) and a discrete GPU for GPGPU usage and Heavy Graphics Lifting. Now it is not important if the GPU inside CPU is powerful or not ( In apple preparative ), they want the GPU inside CPU to be very power efficient. And Intel is doing a great job on this end.

Forgot to mention another point, Apple will still need to work with Intel on Lightpeak. And there are Intel software engineers on LLVM.

Ivy Bridge will probably neutralize AMD's core count advantage unless Llano's successor ups its core counts. Then again they'll probably still be behind in GPU.

Intels strengths are there but that is often over stated. AMD chips still do pretty good in testing / benchmarking. Yes that depends upon the benchmark but what is notable here is that where AMD really fails in the benchmarks would make little difference to Apple as they have OpenCL to make up the difference.

Actually it will be interesting to see how Bulldozer actually behaves on real code. I actually see AMD as really innovating here. Intel has SMT but that is really an old concept. Further Bulldozer ought to allow for a lot more cores on a die or lower overall power.

Actually I don't think Apple gives a crap about where the AMD chips go. They need to knock heads with respect to NVidia and Intel. Intels behavior is absolutely counter productive considering the direction Apple wants to go. Integrating a GPU onto the CPU package is a bit to early in my estimation even if they had decent GPU?'s to integrate. They don't of course so Intel is really thumbing its noise at its customers by not fulfilling their needs.

I don't know about that. As long as they sell Apple has little to worry about. Besides they certainly are a better value than many of the low cost machines they compete with.

Why 2012? Much of what you and I want can be had this year or early next.

• Actually I'd be willing to pay more if they address the short comings in the current iPad. The biggest being the utter lack of RAM. I'm very much interested in this size iPad but I don't want it compromised performance wise. In any event i could see this device coming with the fall iPod refresh.
  
  Quote:
• 9.7 iPad models starting at $499 with Dual Core ARM based processing
  I was hoping for dual core in the current machine but sadly it isn't there and even worst there is not enough memory (RAM) in the iPad to even bother with dual core.
  
  Quote:
• Mac mini Fusion - $499
  I believe Apple could do a much better job with other AMD hardware. AMD/ATIs integrated GPU's are really just the nuts as they don't have to deal with RAM traffic, but more so I believe they have an integrated chip that actually has its own video memory. It might be a task fitting all this into a mini but it certainly would make for a nice machine.
  
  In any event what I' getting at is that the Mini needs another massive boost in performance to keep the gulf between it and the iMac's reasonable. I don't see fusion filling that roll right away.
  
  Quote:
• Macbook Fusion - $799
  That might actually work as long as there isn't a performance regression.
  

I'm more concerned about performance points than price points. While it would be nice to see an entry level Mini that sports good performance it isn't as important as having some sort of well performing machine at a reasonable price point. This brings up the ole X Mac of the past.

That is Apple needs to offer something bigger than a compact box that allows for very good performance at reasonable prices. AMD hardware can give Apple a performance advantage in a low cost machine, it just won't be in a compact box.

Dave

LLVM could be a big win for Apple here. Byte code could be compiled at run time to optimize for the hardware the code is running on.

Do You really think that Apple cares? Really look at the processor they use now in the majority of their hardware. We aren't talking Intels top of the line hardware here. Sure Apple will likely support Sandy Bridge in their high end hardware, but that leaves a lot of open space in the lower end of the hardware lineup. Further Apple is loosing customer because of glaring gaps in their hardware lineup, Bulldozer or other AMD hardware could offer Apple a way to fill those gaps. Mots importantly a way to fill those gaps with economical systems with good GPU choices.

The future doesn't look bright with Intel integrating crap GPU's into their CPU's. Besides Intels low end simply can't compete with AMD's current generation hardware.

This may be very true today but the landscape is changing very quickly. With the move to 20nm or so there will be lots of room on a chip to keep the performance leading edge. At least compared to middle of the road GPU's. High end Discrete GPU's will just keep getting better but how many Apple products use those today?

I actually believe that Apple could have produced a better 13" MacBook Pro with AMD parts today. Mainly because of the way system components are distributed around the the chips in AMD systems. You might loose a little bit in battery life but performance could very well be equal of better than Core 2. Of course there are all the depends that go with such statements.

The other thing people seem to mis in this discussion is that Apple has had a lot of experience with AMD's bus systems using Hyper Transport in Power Macs. Plus the bus has ben adopted by a number of other big name systems builders. AMD is very much quality hardware and for a long time had the best bus going. I've seen others in this thread dismiss AMD as a second tier supplier but that is garbage. They have in many ways been responsible for Apples PPC success.

Dave

This is getting pretty silly. \

VIA is practically the company that defined the low power x86 compatible CPU to begin with, by virtue of their coming to own IDT's Centaur division and whatever was left of Cyrix when National Semiconductor got done with them.

VIA's CPUs were focusing on low power before Intel ever had any intention of creating the Atom CPU. Some are even part of solutions that are completely passively cooled--including the chipset--long before Intel had developed the NM10. VIA's CPUs also have some unique features, including their longstanding "Padlock" hardware security engine. You really ought to see some of their pico-ITX (and smaller!) motherboards. Yes, some of them are smaller than the logic boards Apple is designing today.

While I cannot speak for VIA, I don't know that they really care about having a bleeding edge processor to compete with Intel. They are diversified and have a huge market in terms of supporting devices (USB, Firewire, Audio, SATA, networking) for use in computers. And many of them work perfectly well...

With Apple putting so much emphasis on battery life for all the portable devices, I have a really hard time believing they're going to switch to AMD anytime soon. Their mobile chips power consumption is about twice as bad as a comparable Intel chip. Sorry, but I don't buy into that rumor for a change.

What does "anytime soon mean" please provide some context. Are we talking about 3 months, 6 months a year?

I don't think anyone is going to buy into Apple using AMD processors that are available now and AMD is most likely not in Cupertino talking about the current processors which Apple couldn't use anyways because it'd take 6 months at last to design, test and QA anything from AMD in the first place.

So we need to "skate to where the puck is going to be" and that means we need to be looking at what AMD will be offering in 1H 2010 and that is going to be Llano which is their first Fusion APU.

It'll be 32nm and integrate a capable GPU right in the same silicon with improved power Management features. I think AMD is quoting TDP of rough 20 -59 watts TDP (source)

So let's see the impact here. Take the 2.53 Mac mini with Nvidia 9400M graphics. Here's a list of TDP for various CPU and a couple GPU

The 2.53Ghz C2D processor would be roughly
25W

The Nvidia 9400M would be roughly
12W

So we're approaching 40 Watts TDP just for the CPU and GPU/N+S Bridge for a dual core processor.
Llano will likely come in Dual Core, Triple Core and Quad Core. Which is why their TDP range goes from 20 Watts to almost 60 (which is likely the Quad core).

So it's easy to see "how" the 2011 AMD APU fits into Macs and should not yield much of an impact on battery life as compared to what's in the Macbook line right now.

Quad-core goes to as low as 30 W but that'll probably be a low clock speed like their 1.6 GHz 25 W mobile Champlain coming soon. There's also dual-core at 30 W.

Marketing be damned I'd gladly take a Quad 2Ghz Fusion APU with a solid GPU. We haven't seen the fruits of OpenCL yet but it's coming.

I just downloaded the Siggraph OpenCL preso from AMD's Justin Hensley Ph.D
A Khronos PDF on OpenCL and surprisingly and PDF from AMD, Intel & Nvidia on OpenCL.

OpenCL 2.0 is supposedly due in 2012 with 1.1 being the next intermediary step if it's not already here.

This is interesting as well.

http://arstechnica.com/apple/news/20...i5-upgrade.ars

I kind of agree with Jobs' statement


We're rapidly approaching a new era in which consumers will have to stop thinking about the CPU as the "Central" processing unit. We now have 48-core GPU which is a 3x improvement over the Nvidia 9400M and in a couple of years we'll likely be at 128 cores for a midrange computer.

Of course AMDs future products will have a better TDP, but so will Intel's. And frankly, so far AMD doesn't exactly have a great track record for engineering power efficient CPUs... so H1 2011 may still be too optimistic.

Based on your extensive knowledge of microarchitectures or just your "hunch" ?

Yes Sandy Bridge is coming with a new core and IGP but that IGP is still going to be a separate piece of silicon and few expect it to be of the same class of Llano GPU.

So the question isn't really about battery power (though that's important) but how Apple intends to leverage the GPU with OS X and OpenCL/OpenGL and how they plan to leverage Grand Central Dispatch as more Macs go beyond Dual Core.

No need to go ad hominem. Obviously I'm not an engineer and AMD may very well surprise the industry, like they have before. It's certainly not a bad idea for Apple to keep their options open, I'd just be rather surprised if AMD managed to come out with something as significantly better as Intel's future solution as you make it sound like. Or do you know exactly, what Intel has up its sleeve in the next 24 months? Because I don't.

Agreed my apologies.

Guess i'm just rooting for the underdog here. Intel got the memo about Power/Watt and I hope AMD has been working really hard on the same. Apparently Fusion is sampling right now and so the next several months should deliver more insight as to how well they are able to reduce power consumption.

I'm amazed at how much traction this story has gotten.

Even Jon Peddie and Nathan Brookwood have chimed in

Quad-core mobile Sandy Bridge is still 45 W, although this time there will be an integrated GPU.


It's on-die.

Yes I was thinking about Nehalem's package.

Also for the pro Sumer that is heavy into music, logic, cubase, nuendo and editors, all who need strong FPU AND CPU coudnt Apple start building workstations with 12 cores for cheaper, say from $999 for 4 cores up to $1499? These would also sell we'l with gamers which last I read is bigger than music and movies combined and there are certainly numerous bands, composers, editors that dont need a server class desktop. The iMacs can remain as is, for the consumer, iWork, iLife, ms office crowd since they offer very little in ways of upgrading. I bet you would see a lot of desktops fly off the shelfs with the artist knowing he/she is Viking to get 24 tracks with fx plug INS samplers (orchestra), synced with Reason/Live, etc.

The latest rumors about Apple may acquire AMD. If that is the case it will all make MUCH more sense.

Since Apple taking control, all those problem i mentioned will instantly zeroed out. Not to mention it gets Global Foundries controlling Vote.

AMD only cost 7 - 8 billion to buy out.

Who's saying this? Do you have a link? Not that I'm doubting it but using AMD processors and GPU is one thing..buying them is totally another.

Though I wouldn't rule out Apple making a Billion dollar acquisition at some time during my life.

I don't think this is a rumor in the strict sense. To me it seems more like ideas/discussions making rounds in several internet sites. Here is an example.

How can you say that when so far, AMD has no multithreaded chips as far as I can recall?

I haven't seen a single AMD 4 core design, that when tested, had beaten an Intel 2 core design. This has been considered to be an embarrassment for AMD.

Forget price. Too many people here are talking price. Apple's machines cost more for various reasons. The cost of the cu is just one of them The most Apple could do would be to cut 10% off the price with a significantly cheaper chip. Would that be worth it? I don't think so.

The Tegra is a good chip, but no better than anything else in its class.

At any rate, we're not talking about phone ARM chips, but x86 chips.

Sorry, but I'm not always home, and when a thread reaches a certain size, it takes too much time to read all the posts made after I left, so I don't always get to many. I didn't see your response. I was debating whether to join this one in progress. I never leave a thread because I have no reply, though rarely, someone gets too frantic, and it doesn't pay to continue.

The reason why Intel and AMD chips aren't exactly the same is because they are different companies, and have their own designs. Often Intel does something that AMD does not, and AMD does something that Intel does not. Often (usually) because of Intel's position in the market, developers will support Intel's work, but not AMD's. Intel's SSE instructions vs AMD's attempts to push their own is a good case in point. I'm not going to get into a detailed argument with you here. You can find the performance results on many sites including Ars Technica, ad Anandtech, where they also point out the architectural differences between the two companies designs. If you're an "expert" in his area you should know of these differences. Its pretty obvious that both companies are on different trajectories, or we wouldn't be discussing it. Nor would anyone else. The chips would be exactly the same, and they would perform exactly the same.

What you seem to be saying is that because they are both using x86, they are the same. They are not.

One point I made in that post is well known. Though you (or someone) stated that Nvidia could make x86 chips, I pointed out that they have no license. You didn't mention that. Via is not known for hi performance chips. Whether they could make x86 chips now that would matter is debatable. I don;t know if they have a license for Intel's new designs as AMD does.

Larrabee is an experiment for Intel. It doesn't invalidate their cpu roadmap. Like ATv for Apple, it doesn't detract from their actual business. Perhaps Larrabee will be a success, perhaps not. It's not relevant.

Apple has shown that for their mobile computers, they have a workable, indeed, an elegant way around Intel's built-in graphics. Thats not a problem.

For their desktop machines, there's no problem at all. Except for the Mini (so far), they've abandoned mobile chips and their chipsets.

I don't care about "other manufacturers". I care about what Apple will be doing. Other manufacturers use Windows. Should Apple do that because "other manufacturers" are using it?