In a general sense I'm not defending Kepler performance personally (in some cases it has been left behind in newer games sometimes woefully so) - I just seriously question the 780 results that are commonly used as they don't at all match with my every day usage i.e. take the Battlefront performance posted near the start of the thread - I've got a 970 and 780 I can test with that and in situations where the 970 is getting 76 fps like in the benchmark if I restricted my 780 to boosting like a revision 1 780 I still get 71-72 fps nothing like 65 and with my out the box boost (I have a GHZ edition revision 2 780 that is quite a bit quicker out the box than normal cards) I actually exceed the 970 by ~1fps - with both overclocked to the max it is still pretty close. Strangely enough I do get exactly the performance they get if I gimp my 780 to prevent it boosting above the on paper specs - which makes me question if they didn't just fudge the results based on the specs rather than using an actual card.
EDIT: Obviously in comparison to that one benchmark it could be down to using a different area to test, etc. but I've seen a similar story with several high profile sites where their 780 performance just doesn't match up with what I get compared to a 970.
The issue is not specifically to do with the 780 though is it? The whole range of kepler gpu's from the TITAN downwards are significantly underperforming compared to their AMD counterparts in Fallout 4 1.3 beta according to those benchmarks. If is a fact that the TITAN was neck and neck with a 290X yet the benchmark shows the 290X beating the TITAN by over 50%.
It's a question that needs to be raised and challenged. Mind boggling to see people defend that kind of blatant gimping in performance.
Last edited: