Let's play a bit. I'd estimate a transistor density of around 11.5Mio/sqmm.
That's definitely not "barely more" is it?
my 2 cents: 3.4 blnLet's play a bit. I'd estimate a transistor density of around 11.5Mio/sqmm.
Must be those magical PhysX instructions!AnarchX said:- GTX 680 is 10% faster than 7970 in BF3, slower in 3DMark11
my 2 cents: 3.4 bln
Turbo sure sounds like a good way for it to become a benchmark hero. This could get interesting.
LOL, too many hintsCould be; now all you need is to calculate with speculative math the die area
Being 10% faster in BF3 shouldn't be too hard, since the 580 is already right up there with the 79xx cards.
edit:
Notably, the 7870 is right up with the 7950.
AMD's ROP capabilities did not grow very well in terms of BF3, so as long as Nvidia doesn't regress, it pretty much has a win there.
Being 10% faster in BF3 shouldn't be too hard, since the 580 is already right up there with the 79xx cards.
edit:
Notably, the 7870 is right up with the 7950.
AMD's ROP capabilities did not grow very well in terms of BF3, so as long as Nvidia doesn't regress, it pretty much has a win there.
I suppose it depends a bit on the scene one's testing, but at least at TPU's bench 7970 is ~20% (1920) to ~25% (2560) faster than GTX580, just like it's on average faster than GTX580, so I'm not sure where you get the "right up there" if you're not saying 580 is always "right up there"?
MSAA settings (versus just FXAA) seem to make a big difference from what I noticed. (Does MSAA even work correctly with BF3?)
Last year's Epic Games' Unreal Engine 3 Samritan demo ran on 3 GTX 580 Nvidia cards. Today, Epic is showing it on 1 new Kepler Nvidia card