@ronvalencia: again that's one benchmark others show different results, but obviously 980ti doesn't strugle vs amd as bad as the 970 does which was my point to begin with? and those are reference cards, everyone knows 980ti has more oc headroom than fury x does

hum? do you think frame time variance that is roughly twice as big has no part in this?

@ferret-gamer: yes you can't completely put the blame on vsync, but due to how double buffer vsync works(most games use tripple buffer) you get this:

As you see the frame times jump between 33.33ms(30fps) and 16.67fps(60fps) this happens because once your fps drops from 60fps to 59fps on double buffer vsync it actually cuts your refresh to half -> 30fps then your fps gets back to 60fps, which causes this huge difference in frame times

well the poor frametimes are partly due to the double buffer vsync, which then again could be tripple buffer(as apparently tomb raider on UWA is able to do this)

No the frame time issues are less severe on higher HZ monitors so it's not "nothing", I can look up the benchmarks and I'll edit in a bit

edit: http://www.pcgameshardware.de/Quantum-Break-Spiel-15745/Specials/Technik-Test-1191358/ you dont really need to know germany to to look at the pics

@ronvalencia: that seems very odd, how the hell are they getting lower results on 1080p than others on 1440p? I have higher avg fps than that on 1440p

that being said literally every benchmark QB has shown different results so far

MSI afterburner

@StrongBlackVine: UC3 thread is one of the very best, TOAT contender no doubt

@R4gn4r0k: you also have a lot of oc headroom in 970, good chip can reach almost 980 levels and it's still way less power hungry than corresponding amd cards, but the future doesn't seem too bright for 970 sadly due to maxwell sucking on async compute and the vram, it'll still be a good 1080p card

Well the thing is Dx12 games that use async compute favor amd as nvidia cards just plain suck at it(maxwell cards are unable to do it purely on hardware), not exactly up to raw computal power