Replies to post #136448 on Intel Corp (INTC)

[Tenchu]

Beamer,

I think the rumors of a 2GHz Cyclone core pointed to what Apple was targeting.

Where were these rumors coming from? That's ridiculous on its face.

No way Apple can bump up the speed of the core from 1.3 GHz (iPhone 5S) or 1.4 GHz (iPad Air) to 2.0 GHz without completely rearchitecting the core with "hyperpipeling" in the style of Pentium 4.

Process isn't going to do it alone. Obviously not. Even if the foundries have processes that match up to Intel's in terms of yield and quality, there was no way Apple would have reached 2 GHz.

I don't even expect the next iPad (iPad Air 2? iPad Dare?) to reach 2 GHz.

Tenchu

[Unkwn]

Of course, I think it's likely that we'll see better frequencies over time, and if Apple waits long enough, they'll see enough 2GHz parts to launch an iPad successor in first half 2015.

My theory is that the 20nm process is still not mature enough to deliver good yields and sufficient frequencies, but we'll see that improve over the next 6 months.

I'd say it's rather a power issue than a clock issue. Especially leakage will be a big issue with planar 20nm. Sure you can say it's related. To reduce leakage, you can reduce the voltage, which on the other hand makes the transistors switch slower. Besides: I think nobody really knows about how A7 and A8 are actually turboing. Apple just doesn't talk about such things and I don't think anyone really checked the clock spectrum of the chip when running at higher speeds - I may be wrong though. I still doubt that Apple can get this kind of performance without turbo clocking the core. 1.4 GHz could be just the base clock.

Anyway - things seem to get much harder for the ARM guys right now. Low hanging fruits have been picked - now it's the hard work which is left where Intel has a big lead - as Core M impressively demonstrates.

No way Apple can bump up the speed of the core from 1.3 GHz (iPhone 5S) or 1.4 GHz (iPad Air) to 2.0 GHz without completely rearchitecting the core with "hyperpipeling" in the style of Pentium 4.

Well, there are ARM cores clocked at 2.5 GHZ but those are not as wide as Apple's. It shouldn't be an issue to simply raise the voltage and reach 2 GHz, even with such a wide core as Apple's. That certainly would increase power consumption to Ultrabook levels - where Intel's cores deliver much higher performance at the same power level.

As I pointed out already many times: Intel already provides the best performance/watt in the industry. They just couldn't bring it down to tablet levels. Core M does just that and if Intel provides a fully integrated SoC with similar specifications at competitive prices with Broxton, I don't see any of the ARM guys compete with that within a year. Let's hope Intel delivers this time.