Originally Posted by zunx
What is needed is a true a full Mac with just 400 to 600 g. The Mac in your pocket. Always. And I mean a true Mac with Intel x86 inside; not an ARM-based iOS device.
Erm... But that Mac in your pocket won't be a Mac as we know it. It will be iOS. Apple is clear that below the 10" form factor it will never be a regular keyboard-screen OS X.
Originally Posted by Dick Applebaum
Maybe I should have phrased it like the OP -- because Apple controls the platform, OS and hardware, they can take advantage of ARM/SIMD in iDevices.
It is doubtful that MS or Google can control this to the extent that Apple can.
Precisely. And again the scope of scaling up ARM-SIMD-PowerVR is phenomenally easier than cramming down x86 into tablet/mobile land.
Originally Posted by SSquirrel
Yes but again, a current gen discrete desktop GPU is blisteringly fast and can pump way more data than a current gen top end ARM GPU. So even if the CPUs stacked up managed to even out, the desktop would still have the advantage with the GPU compute capabilities.
Yes and No. Those blisteringly fast GPUs burn at least 50W to in excess of 150W. On Windows and even Mac OS X, GPGPU just hasn't lived up to the promise. On Windows, even gaming with a good GPU is near impossible nowadays with endless patches and graphic driver updates. I feel so burned up sometimes that Nvidia and AMD/ATI threw away such fantastic GPU engineering by just eating up more and more watts, failing to fab at lower nodes in a timely fashion, and just not sorting out the driver issues with Windows and games AND GPGPU stuff in Windows.
Given Intel in-CPU optimisations like H.264 encoding and so on, a modern Core i7 with the right software trounces GPGPU... in addition to the advantage of *not needing* another GPU besides Intel Integrated.
Originally Posted by Macnewsjunkie
One of the really cool take aways from the Jobs biography was how secrecy allowed Jobs or the current CEO to make the decision for the next platform at the last minute with all the dominoes lined up for both decisions. It is a decision point made based only on what is best for the companies future rather than its present. Both PowerPc and Intel mac books were made by independent development teams. If one failed the other could take over. At the last minute after sleeping on the decision Steve chose the intel chip. Doing both insures against intellectual theft and empire making by the drones in different departments. The fact that it makes for awesome theater and marketing is just icing on the cake. No wonder Steve always seemed so excited by the moment. The decision was just made and the future is right now during the keynote.
So yea I would say that this is now SOP at Apple. Let the press and internet Guess. Apple doesn't even know yet, and any employee who thinks he can get away with telling reporters about what the next great thing is will be in for a surprise.
Yeah, this was awesome isn't it? In the best traditions of American "skunkworks". Apple sure gives Area51 a run for its money.
Originally Posted by Snowdog65
LOL. At first glance, I thought I missed the sarcasm tag.
But no you are serious. There is nothing quite as entertaining as long winded under-informed opinion "educating" the rest of us.
Impossible for x86 to be power efficient?
If you read some actual CPU designer opinion on the the x86 overhead it is only 5-10% overhead versus other ISAs. Intel has simply been slow to wake up and change course.
There is nothing inherently wrong with x86 for modern smartphones or tablets, except an installed base of ARM code. Similar to the issue ARM has on the desktop.
Perhaps theoretically x86 does not have major disadvantages. But Intel has a tough job ahead of it to come down to ARM-level power draw without the x86 CPU being worthless.
Originally Posted by tipoo
Exactly. He still hasn't responded to the article I posted above, Medfeild looks like it will be both power competitive and performance competitive. And this isn't even using their 22nm fabs yet. When it does, it will have quite a lead. There's Cortex A15 coming out of course, but Intel is one tough gorilla to compete with in the long term.
More food for thought. I know it doesn't mean success or failure, but Intel sure has a lot more to throw at R&D if things should go sour.http://www.wolframalpha.com/input/?i...M+Holdings+AMD
Fair enough, but I'm not feeling Intel on this, the weight behind ARM is signficant with Apple, Google and Microsoft all on the bandwagon, among others.
This is a battle Intel can "afford" to lose in the next five years, but as we know in tech, sometimes losing the battle means you have nothing much left to fight the war.
Again, the scenario of an iPad 4 which delivers Core i5 "experience" (due to optimisation etc) and DX10-quality 1680x1050 graphics, all playable on a HDTV, presents a tantalising possibility that you wouldn't want to touch your laptop unless for intense content creation. For business and content consumption, x86 in five years is going to be total overkill. Yeah, Windows 8, Office 2013 or whatever will drive the x86 forward but the road is getting narrower and the cliffs closer.
Originally Posted by afrodri
I'd agree, and also say that Intel _has_ realized it for a while now, its just that their attempts to break in to the mobile world have been flops. Still, you are right not to count them out. They are a _huge_ company which can take a brute force approach like none other. I remember the 90s when it was "obvious" that the x86 architecture was doomed and RISC architectures would crush them, and on a purely technical standpoint x86 didn't stand a chance. But, because Intel could shovel cash and engineering and manufacturing knowledge at the problem, x86 survived and even took over much of the high end. They still might be able to pull it off in the mobile world.
I wouldn't be surprised if Apple has both ARM and x86 versions of MacOS and iOS internally, just like they had x86 versions of MacOS years before the PPC->x86 changeover. If Intel does come up with a decent line of low-power cores, Apple would be able to take advantage of it.
It's true, PPC RISC was definitely awesome but Intel got its act together early last decade. However they were lucky in some sense that the Pentium3 line proved to be the way to go in amongst the kludge that was Pentium 1, 2 and the ill-fated Pentium 4. Also, Motorola and IBM both dropped the ball, somehow it seems very strange that the best they could do was a RISC processor eating well over 100W to compete with the stuff Intel (and don't forget AMD) was putting out with ease. In fact before the fiasco of the G5 Motorola was already sruggling badly with the G4.
So in some sense, PPC RISC and x86 were not that far apart, trading blows for a while until Moto and IBM screwed up, and Intel rallied the troops.
But if you're looking at ARM vs x86, the "gap" in thermals and no doubt performance is much bigger. Again the argument for ARM is that it has immense scope for scaling up performance and cores whereas x86 is facing the reverse challenge.
Intel x86 might be able to pull it off in the mobile world, or at least tablet world, but they have a bit of an uphill battle. ARM and PowerVR are on the straight and just have to keep on keeping on.