does it have to be a dy 12 title in a api review.. maybe there are some thing that the api improves in general?
There's nothing that I can see different from 8.1 to 10 game performance wise. If WDDM 2.0 was supposed to improve things it either hasn't yet or won't til DX12. WDDM 2.0 was supposed to bring optimization in memory management etc, but that's not translated into FPS.
From what I see they're running 15.7 on all 3 OS, all have the CPU overhead optimizations. I wonder how they benchmark most of the games. Games like GTA 5 have the built in benchmark. If they used that then games like that are the one's to pay attention to and that shows nothing improved.
Speaking of CPU overhead, with drivers >= 15.7 they're fairly equal. It has to be something else. Either the benchmark itself was flawed or WDDM 2.0 did improve certain things.
For what i have seen, thoses are given result in cpu / memory / compute tasks ... ie, Winrar benchmark who provide a nice increase or some benchmark like pcmark ... But im really doubtfull, seeing how the games DX11 are coded, that it can bring any improvement outside the CPU API. ( and this will surely only be the case with games coded for it, outside some rare case in DX11 titles ). The actual CPU API in games have not been coded for WDMM 2.0.. As for the gain for AMD GPU's, i will not expect anything substantial before DX12 games are there.. simply, what is GCN and are in DX12 who can substantially increase the performance are not in actualDX11 games ( asynchronous compute, DMA engines ( finally used ), typed UAV load etc etc . And nothing is saying that Nvidia gpu's will not benefit too of them.