It'll take years before Intel's Haswell-level designs are low power enough to even be a consideration for mainstream tablets. There's a reason Intel is still aggressively pushing Atom here.
But I have no reason to believe Atom is going to leapfrog ahead of ARM in performance. Rumor is that Silvermont will offer 20-30% higher IPC. They'll need to clock higher than Cortex-A15 to compete with it, let alone beat it. I really can't fathom them exacting a huge lead, and I'm sure ARM and other ARM core designers aren't sitting still in the mean time.
There will be more than Haswell and Atom as we know them, soon enough. Intel's announcement of moving to an accelerated tick-tock model means they're working on something completely new. Out-of-order execution is pretty much confirmed. So there's no way Silvermont will only be 20-30% faster per clock. And yes, ARM designers aren't sitting still either, but meanwhile their cores are getting relatively big, expensive, and power hungry. So they're stuck between a rock and a hard place. They have to push forward but that makes them more vulnerable to Intel's design experience and process advantage.
Anyway, if I'm reading what you're saying correctly, it's that Apple could outsell every other tablets in spite of having much less CPU power (sometimes a good 50% less MHz for the same Cortex-A9) because their software was faster. I don't see a reason to believe this will suddenly change.
No, what I said is that Apple brought the first successful tablet to the market, and it wasn't the processor's ISA that was the differentiating factor. Sure, Intel failed to deliver a suitable SoC at that time, but that's only because they couldn't risk putting time and effort into something highly custom and with no guarantees of becoming profitable for them. The situation is radically different now. We know tablets are successful and we know their computing needs. Intel can safely invest into designing something that will be suitable for 90% of the market in a couple years from now. Looking at the accelerated tick-tock announcement, that's exactly what they're doing.
And you're still ignoring the differentiating factor that GPUs can provide in SoCs.
No I'm not. But you seem to be ignoring the fact that making the GPU faster than necessary is generally a waste. Just like in the desktop market, Intel will aim for adequate graphics performance for the masses. No more, no less.
Uh, you said they would change their business model if they had to. What did you think I was talking about when I said business model?
Design a suitable SoC for the majority of the mobile market. They weren't able to do that before, soon they will. This erases ARM's biggest advantage, and they'll have to fight on technological grounds next...
If you don't understand that Apple is interested in providing much more than what everyone else is when it comes to GPU then I don't really have anything else to say to you.
I understand it perfectly. The mistake you're making is that this would allow Apple to keep its position. If Apple wants to become the NVIDIA of the mobile market, that's fine for Intel, they'll still take the lion's share of the market by following in Apple's wake and providing adequate solutions to everyone else. That said, Apple isn't exactly known for being gaming minded. And the primary reason to substantially increase the GPU performance now is for games. So I doubt they'll be that aggressive about it for years on end.
Did I ever actually say ARM won't? Nope. You think that just because I'm questioning your reasoning it means I'm supporting an opposite view. The reality is neither of us really know.
With all due respect, you're grasping at straws here. You don't agree with me, but you don't disagree with me? You're clueless about the outcome, so everyone else's opinion is a coin toss as well?
Facebook
Twitter