InvestorsHub Logo
icon url

This Causes an Error

03/14/17 5:16 PM

#148223 RE: mas #148222

I thought the 28nm tsmc one was a complete waste of time as it negated Intel's in-built foundry advantage over fabless competitors but the 14nm airmont Sofia could still be and should be done now as it would be a compelling product at the right price.



Here's your 14nm Airmont SoFIA

http://www.spreadtrum.com/en/show_news.html?id=fe766282-e9cc-4ffd-b211-e3d19675d3f7
icon url

This Causes an Error

03/14/17 5:36 PM

#148224 RE: mas #148222

Tie that up with a high end modern Medfield like a single core Skylake with smt and intel would have competitive mobile products again but this would require a CEO with a clue and the attention span of an animal higher than a goldfish !

https://www.cnet.com/uk/news/meet-intels-sofia-the-new-low-cost-smartphone-processor-created-in-singapore/



I disagree. Using the Skylake/Kabylake/etc. lineage cores in a phone power envelope would be suboptimal, esp. in a single core configuration.

No, if I were tasked with making Intel a mobile player given the current assets, I wouldn't even use x86.

I'd license ARM CPUs, ARM or ImgTec GPUs, and go all-in on building competitive "uncore" blocks, the stuff that matters to the phone SoC. If Intel's manufacturing technology lead in terms of perf/power efficiency is real, then they should be able to build ARM-based SoCs that perform better/consume less power than the competition. Integrated with Intel modem and the rest of Intel's IPs, that could be interesting.
icon url

fastpathguru

03/14/17 10:24 PM

#148228 RE: mas #148222

14nm airmont Sofia could still be and should be done now as it would be a compelling product at the right price.


I'm not sure a shrink of a several-year-old chip is going to be competitive* with the best of what has happened in ARM licensees plus the best of the what has happened in the foundries since then.

fpg

*Unless of course it's sold "at the right price."
icon url

Andy Grave

03/14/17 11:47 PM

#148232 RE: mas #148222

I thought the 28nm tsmc one was a complete waste of time as it negated Intel's in-built foundry advantage over fabless competitors but the 14nm airmont Sofia could still be and should be done now as it would be a compelling product at the right price.

.....what in built advantage?.......you don't know Intel's fab cost at equivalent yields and you don't know what Intel's yields are in comparison to the foundries.......I suspect they are worse in both respects....you suspect otherwise...........and that's all that can said.........but your track record has been positively awful..... while I've been pretty much right on.

....also,....when are you and Eassa going to comment on Intel's 14nm process density in comparison to AMD Ryzen on Samsung 14nmLPP?........if either of you want to maintain any credibility at all, you MUST engage on this subject.........so let's hear your rationalization for this apparent Intel obfuscation.