First tape-out with TSMC’s 16nm FinFET and ARM’s 64-bit big.LITTLE Processors

[DSC]

http://community.arm.com/groups/soc...-finfet-and-arm-s-64-bit-biglittle-processors

 

[Wynix]

SMC’s 16FinFET process offers significant improvement over 28HPM for high end mobile computing and networking. Since designs could gain >40% faster speed at the same total power, or alternatively reduce >55% in total power at the same speed over 28HPM

Things are looking good at TSMC, i wonder how Samsung and Global Floundries are coming along.

 

[xpea]

Things are looking good at TSMC, i wonder how Samsung and Global Floundries are coming along.

if 16FF SoCs will really be available early 2015, then it will be very interesting times in mobile market because Intel will not have anymore a huge process advantage...

 

[Turbotab]

if 16FF SoCs will really be available early 2015, then it will be very interesting times in mobile market because Intel will not have anymore a huge process advantage...

Intel's 14nm should still have a significant density advantage, given that TSMC's 16nm process uses the backend of their 20nm node.

 

[Grall]

I'm sort of suspicious TMSC is being (deliberately) overly optimistic (again), has there been one single node transition these past ten years they've not struggled hugely with...? Maybe chips will be available at the claimed date, only with terrible yields and poor overall quality.

 

[xpea]

Intel's 14nm should still have a significant density advantage, given that TSMC's 16nm process uses the backend of their 20nm node.

I'm don't know about density but I think performance is more relevant in these expensive 14/16FF chips. So do we know if Intel 14nm has better electric parameters than TSMC 16FF ?

 

[Entropy]

if 16FF SoCs will really be available early 2015, then it will be very interesting times in mobile market because Intel will not have anymore a huge process advantage...

Actually, their CEO very recently claimed that they had pulled 16FF to 4Q14. However, that may be the start of the production of some Xilinx gate array just like with 20nm, so what that means for mobile SoC volume production is still an open question. (They claimed the 16th of January this year to be in volume production of mobile SoCs on 20nm, I wonder when we will see these SoCs introduced, and in products respectively.)

ImgTech and TSMC released a blurb some time ago about 16FF and TSV interconnects that boded well. I think that it will be an interesting process node, if pricey to get into for the customers of the foundry.

 

[xpea]

Thanks Entropy, that's good info.
Unfortunately, with TSMC, we never know what to think about their claims and roadmap on new process. 16FF seems a big step over 28/20nm with some bold expectations.
My small finger tells me that Apple will jump early on this process for their new SoC and even, why not, targeted at desktop/laptop and not only mobile...

 

[Exophase]

I guess TSMC could be scrambling to get 16nm out ASAP since no one wants to use 20nm :/ Not that that means they'll actually successfully bring in the release date..

 

[3dilettante]

Intel's 14nm should still have a significant density advantage, given that TSMC's 16nm process uses the backend of their 20nm node.

It's going to be interesting to see how much of a difference there is, since the back end for the foundries has been significantly denser than Intel's for a number of nodes prior.
Intel made the choice of not scaling metal aggressively earlier, while the density-focused foundries allocated more effort to scaling metal. This was a contributing factor to the process schedule difficulties as of late.

I'm don't know about density but I think performance is more relevant in these expensive 14/16FF chips. So do we know if Intel 14nm has better electric parameters than TSMC 16FF ?

I think the safe bet is that Intel's process will have better performance per transistor, and better odds that its transistors will perform closer to the promised number than TSMC's.