Using the information provided from the Cntrl+F1 option printout in the lower lefthand corner, I wanted to compare VR versus nonVR to see if the Devs think that these results are expected. I picked an area north of Santa Barbara with the Learjet at 6700 feet or so just doing a gentle turn sitting in the right hand FO seat. Not completely Ultra settings but almost and a Pixel density override (render scaling) of 1.5. Vsync was OFF for the nonVR with a monitor resolution of 1920x1080.
NonVR VR
tsync = .2 tsync = .26
tcompute = .23 tcompute = .57
trender= 2.96 trender = 12.51 <<< seems high but maybe that is what you get when Vsynced to 75fps
fps = 254 fps = 74.92
I also had the Performance HUD on/off from the Oculus Debug Tool and saw that the performance headroom when looking sideways (away from the instrument panel) was about 6% but when looking at the panel was 0% or maybe -2% - therefore there is a very small margin for keeping my DK2@75fps happy with no stutter. (Is ATW implemented? )
I realize that I can turn down some of the settings to get more margin - that isn't my question. I just wonder if the Devs think that the render time which I would expect to be double for stereo plus a little more is correct - but maybe trender is just not "correct" when you sync to the Rift framerate - but when I synced the nonVR to 120fps (8.33 msec period), the trender is still the same (2.8+-)
My concern is simply wanting to have the same "outrageously" fast performance in VR as I get with nonVR - except for the factor of 2 eyes plus some other processing due to "supersampling" and whatever. I tried setting supersampling to 1.0 and trender still stayed at 12.5 or so. BTW, where is MSAA, FXAA, Aniso set? Also, with SteamVR, are the VRcompositor process and VRDashboard process part of the performance issue? I don't think there is a way to run in VR mode without Steam.
Thanks for a tremendous flight sim - especially for VR!!
Dave