Intel Corp (INTC)

[Unkwn]

.....don't see any in house designed 64 bit ARM cores here.......roadmap thru Q2 2015...........probably means no 14nm newly designed ARM cores until some time in 2016......at least my take.......maybe an opening for Intel Broxton if they can get it out before 2016

How do you link between "no custom ARM core design by Qualcomm until Q2 2015" to "no 14nm ARM designs until 2016"? Qualcomm may introduce updates of their 20nm lines with custom ARM cores - I don't see any connection to 14 or 16nm designs at all.

Regarding Broxton, we only have promises from Intel so far, no facts at all. Whether Broxton will be competitive depends on so many factors that it doesn't make sense to try to compare it to any planned products of Qualcomm.

Thanks for the report by the way. It shows just how important the whole packaging and integration issue is for mobile. Something Intel really needs to accelerate a lot. Remember: No integrated modem for Broxton. We'll see about the other parts and, in the end, as seen with BayTrail, what the final BOM will look like - that's what is going to make it a winner or a looser (together with its power consumption and performance).

What I find interesting about the roadmap is also the fact that Qualcomm produces their RF frontend chips in 28nm. Why is that? For analog circuits there shouldn't be reasons to continue the shrink since those circuits usually don't shrink much. Maybe there is still lots of digital circuitry integrated for some reason. Anyway, we'll see how all those ARM guys will do with the transition of their custom designs to Finfets.

The Big.little approach Qualcomm is going seems a bit desperate. Also the hint that there will be a quad A57 and a dual A57 high end version of the SoC tells me that they are about to have serious thermal issues running all cores at full speed. Maybe one is designed for high end phones where the other is for tablets. Big.little will perform well in benchmarks when all cores run at full speed (until they throttle due to thermals), but it will have issues for standard applications, since it requires a complex scheduler and tha switching between the cores will likely affect the user experience at bursty tasks, which are the most common actually. As I said many times before, beefing up two cores as much as you can, just like Apple did, makes much more sense than the stuff Qualcomm and the likes are trying. I bet in a few years, we wont see any Octacore Big.little designs anymore - it only helps ARM with higher royalty revenues.