Carzy Zarx’s PC Gaming Emporium - Catch Up on All the Latest PC Gaming Related News

[JEMC]

fatslob-:O said:

JEMC said:

It's true that 14/16nm is based on 20nm, a process that wasn't suited for high power chips like GPUs, but with FinFet tech to solve that problem. There will always be some tweaks to be made, and some time will be required until yields reach acceptable numbers and the production ramps up.

So the question is, do they go for the big chips right from the start despite the yields and that those chips aren't the ones that sell the most, or do they concentrate on what the market buys the most, the sub-300 $/€ one? With Nvidia we don't know yet, but AMD has openly decided to go the second route.

As for the second part, I wish I understood what you wrote (I know nothing about the different instructions and extensions). Thing is, Intel is more concerned about ARM and the low power market than launching faster products, and has put all its efforts towards efficiency, launching chips that are roughly 5-10% faster than the ones they are replacing.

20nm is not suited for high powered chips ? The SPARC M7 begs to differ ...

They should release the big chips cause people buy high end GPUs for PERFORMANCE and that is important but it's also why Nvidia has a 70%+ marketshare each quarter ... 

I too am disappointed that Intel has decided to focus on efficency rather than pure performance but they do have high targets and aspirations even if they have become a little unrealistic as of late ...

Of course they should release high end GPUs, but the mainstream cards sell more. And until the yield numbers are acceptable and TSMC & GloFo/Samsung increase their 16/14nm production, focusing on the high end will actually mean less people jumping on to the newer products.

I don't know about the SPARC M7, but I've read a few articles stating that the 20nm process wasn't suited for high performance chips like those in GPUs.