NVIDIA Maxwell 28nm vs. 20nm performance differences?

[Criticalacc]

For the upcoming NVIDIA Maxwell cards, would it be better to wait for the 20nm process coming out in Q4? How does performance benefit from smaller processes versus larger ones? Is it worth waiting until Q4 instead of just getting 28nm Maxwell in Q1? Also, has anyone heard any rumors on pricing for the cards?

 

[CraigN]

There hasn't been any information whatsoever on the Maxwell performance released yet to my knowledge, the last thing I heard was we won't be seeing any of them till the end of Q1. Pricing will probably be very similar to how the 700 series went - Expensive at first until AMD finally has a response, and then a price drop to stay competitive.

The only other thing I've heard on the rumor mill is that the 28nm Maxwells launching in Q1 will be upgrades to the mid-tier Kepler refresh cards, and the big kahunas (780Ti replacement and the like) will launch on the 20nm process. This is all speculation though. Nvidia talked mostly about their Tegra K1 at CES 2014 and GSync, no official announcements so far :\

 

[bystander]

As mentioned above, they'll be using 28nm on their lower end and medium cards most likely, then use 20nm on their high end cards. The advantage to 20nm is they can cram in more transistors/CUDA cores, allowing for a more powerful part. The same GPU's will not be made at both sizes.