Cloudfire777
Golden Member
- Mar 24, 2013
- 1,787
- 95
- 91
How old are most of you guys? AMD(ATI) and Nvidia have been one-upping each other to this degree every other generation or 2 for over a decade.
Yeah but they have been relying mostly on new better nodes to shove in more transistors. Newer nodes have become more complicated over the years which is why we are still at 28nm and not 20nm. FinFETs are needed to make it worth the jump to a new node. 16nm FinFET isnt ready until late 2015, and its not 16nm but 20nm. Why do you think the 500mm2 GPU is coming from AMD? If it was 20nm it would have been substantial smaller.
Look at Intel and the huge problems they have with 14nm. They will be a year behind schedule when we get high end Broadwell. That is the company with the most advanced equipment and most experienced staff.
TSMC on the other hand regulary mess up, have customers like Qualcomm threatening to produce their chips at Samsung or Intel fabs instead. If Intel have huge problems with newer nodes, yoy can bet your money on that TSMC will too.
So Nvidia have been adapting to this. They now have managed to overcome this with a much more efficient architecture on the existing node. AMD have shown no signs of this, with the recent big launch of R9 285 thats nowhere near in efficiency to Maxwell. In fact Gibbo said AMD had the next generation cards ready in August this year. And he was reffering to Tonga...
So believe it or not, the 500mm2+ GPU with all signs pointing toward a hotter GCN card (no 20nm availabilty, Asetek making the cooling needed), is whats coming from AMD.
This is AMD playing the "more cores, massive TDP" they have been doing against Intel for several years now. This is very bad news