Originally Posted by Kuivamaa
...and the problem is that as I said, other popular 780Tis wouldn't reach that number. You are asking for MAX possible boost - Anandtech on a reference 780Ti merely hit 980Mhz. Yes,it was reference,not a WF but this test brings in general 780Ti up agaisnt 290X. All 290X models will hit 1000Mhz in that game (even reference in a very loud profile), that cannot be said for the Ti as anand showed. The 780Ti is capable for more, but for whatever reason, stock reference it won't do that. Get over it already.
Proof of all reference 290Xs keeping 1000MHz on BF4 at all times? As we all know the differences between retail reference 290X boards are pretty huge. Some cards do better than others. PCPer, Legitreviews etc. have shown that press samples =/= all retail samples.
I have not seen any BF4 frequency graphs for the 290X (only thing out there is the 1000MHz number over at anand but it's clear that he got a really good sample since others have shown bigger frequncy changes in uber mode) however it's clear that in a lot of games a 290X does droop under 1000MHz even at uber settings:
Even 290s fluctuate in BF4:
I'm sure that not all 290Xs do this however low boosting 780Tis are also very rare. Just like low boosting Titans. You had to have extremely bad luck back in the day to get a 960MHz Titan just like you have to have bad luck these days to get a 1006MHz 780Ti or a 290X that can't keep up the frequency even in uber mode.
However if you don't like the max boost number I gave you, how about pcgameshardware just using whatever max boost their card has? If they're really unlucky it might be 1006MHz however it's more likely something above 1040MHz.
Or how about just letting both cards run without locking the frequency. That way the 780Ti will stay at its max boost anyway.
And to top it all off this argument becomes irrelevant one OCs are taken into account since all GK110 cards clock considerably higher on average than Hawaii ones.
Originally Posted by felon
nonsense. Who cares what you have been reading, i have a 3770k with a 280x, i am telling you there is substantial difference. Someone that's not lazy really needs to get some numbers on here. Why is nobody posting their benchmark comparisons?
95% of the fps mantle gives comes from removing CPU bottlenecks.
Even AMD knows this, their own testing showed an increase of 2% with a 3960X and a R7 260X.Edited by Alatar - 2/20/14 at 6:06pm