Intel Corp (INTC)

[This Causes an Error]

mas

I appreciate your reply/viewpoint. Thank you.

Android software vendors would just stop developing/porting their code to x86. That means no future chance of Atom or even Core being in mainstream phones ever again just on software grounds.

That is true, but Intel would be fielding ARM chips in this case, making this a bit of a moot point. I don't see Intel selling much in the way of Core into Android devices anyway unless it starts building Core m into a fully-fledged, single-die SoC for mobile devices.

This, IMO, would be a very interesting development for x86 in Android if true and would probably solve my complaints about Atom since the latter would then be relegated to low cost SoFIA chips while Core-based SoCs would go after premium tablets/phones.

Really, is that what you are advocating for the very short-term gains promised in that article especially after all the heavy lifting Intel did to get Android developers interested in x86 ?

I am advocating that Intel do whatever it must in order to develop the best product that it can in order to generate revenue and profit for its stockholders.

I don't care if Intel uses ARM, MIPS, X86, or IA-64 -- I just would like to see competitive processors from Intel that OEMs are willing to buy and put into critical mobile devices.

There is not much in it and anyway the cores are not usually the biggest part of the SoC, the gpu is which is also other Intel IP that would be discarded in your suggestion. Once the 14nm Sofias are here Intel's mobile strategy will make more sense i.e. mostly Intel IP made on Intel's own FABs to give unbeatable and profitable performance/price. In your way Intel's modem would be left to market forces as well as the gpu as a byproduct of ditching Intel IP and just picking the perceived, to you, best.

Intel is using ARM IP in the first SoFIAs although I do know that they are planning a transition to Gen for future SoFIAs. Not sure what the intercept point will be though; potentially 14nm but maybe not until 10nm.

I agree with you that Intel has the raw assets to develop and sell mobile SoCs that deliver great performance/watt and performance/dollar, but I simply think that they are not advancing Atom quickly enough.

Part of this I suspect is to "protect" Core, but another part of it is just that Intel isn't putting as much into Atom as they are into Core. For example, Intel has two very highly skilled teams working on Core (Hillsboro and Haifa teams) but it has just the one team in TX working on Atom.

With far fewer resources, I don't think we will see the kinds of big generational leaps that the ARM folks who do have multiple parallel teams do. Intel, IMO, needs to bulk up its Atom teams significantly and it should not restrict their efforts based on any attempt to "protect" Core.

A57 because of throttling issues hardly seems any better than A15/Silvermont at least at 20nm. It is not a quantum jump over either in mobility environments which is why A72 was developed

Only Qualcomm seems to have had issues with A57. Samsung did just fine with both their 20nm A57 implementation as well as their 14FF implementation. I think the Qualcomm A57 implementation was basically quick and dirty as a hasty response to the 64-bit Cyclone.