By using this site, you agree to our Privacy Policy and our Terms of Use. Close
fatslob-:O said:
JEMC said:

I don't think that my expectations are too high.

When the first 28nm debuted, the 7950/7970 were <370mm2 chip and the 670/680 were <300mm2, and now both FuryX and 980Ti are twice as big, being 600mm2. With the move to 16nm I expect the first bunch of cards to be around 300mm2 again, so while the transistors will be more expensive there will also be less of them.

Besides, there's no way in hell that 480X/1070 won't outperform or at least come close to today's high end cards, and I also find it very hard to believe that neither AMD or Nvidia will launch them for more than $400.

With the Fury X and 980 Ti, both Nvidia and AMD are really cutting in their margins ... 

A 16nm 300mm^2 chip should have roughly the same amount of transistors as a 28nm 600mm^2 chip assuming ideal 2x die space scaling. I expect savings to come from microarchitectures, not from process nodes at this point ...

I don't even know the 480X/1070 details yet to make an educated guess ...

Well, of course the newer microarchitectures will have a big impact on performance. What's the point of making them if they don't improve things? .

I expect them to be more like 350mm2, and while some of that die space will be "wasted" with newer features (DX12 and Vulkan stuff), HBM controllers are smaller than the GDDR 5 ones.

I don't have 480X/1070 details either, but they are supposed to be the second best cards (I hope AMD stops the "X" and "non-X" nonsense and launches less cards), they can't be sloughs.



Please excuse my bad English.

Currently gaming on a PC with an i5-4670k@stock (for now), 16Gb RAM 1600 MHz and a GTX 1070

Steam / Live / NNID : jonxiquet    Add me if you want, but I'm a single player gamer.