Ah you have a 5900. Nice. I’m really glad these settings helped! They work equally well on FSX and P3D also.
With your 12 physical cores you could just lock FS out of logical cores 0 and 1 entirely and it will probably be even smoother. My ancient 6700k only has 4 cores so I have to let FS use one of the first logical cores. Really hoping the Alder Lake and Win 11 combo give good performance in the sim. Been wanting to upgrade for a very long time.
Any remaining microstutter can maybe be smoothed out with driver level fps locking, or with riva tuner statistics server. They lock fps in slightly different ways. On some systems driver level lock is smoother, and on others rivatuner wins. Worth experimenting with both.
A gain of around 10 fps and significantly higher sharpness in the cockpit and the surrounding area. Such a performance boost for the Reverb G2 with a setting of a few seconds is comparable to switching from a RTX 2080 Ti to an RTX 3090!
Nvidia 472.12 (latest version, max. Performance, standard settings)
OpenXR version 107.2109.100.10 (latest runtime preview deactivated, no motion reprojection)
Windows: HAGS off, Game mode off
CPU: Hyperthreading off
Setting for Windows Mixed Reality> Headset display: User experience option: Optimize for high performance (not quality)
Resolution: 4320 x 2160 best quality
i just tried steam SS at 200% and msfs render scale at 50%. it gave maybe 1 or 2 fps increase so not statistically significant, no increase in visual quality. i want my 10 minutes back!
tbh, all this messing with supersampling in one program (steam), then counteracting that with messing with that in another program (msfs), is just silly lol. 200 x 50 is the same as 100 x 100. theres no magic in messing with these settings. if it really worked, asobo would be telling us to change these settings. its just placebo.
of course if you really believe it works for you (i cant argue with fps but i can argue with subjective opinions of clarity) then good for you but i think its all a load of rubbish, my test just now has backed that up.
I have been telling this to people here for months and it seems no one is even willing to try. I have been running 200% OXR and 80% FSS for months now with significantly better results than “casual” oxr (60-100%) resolution. The reason i have 80-90% FSS res is because i have an older wmr headset (dell visor). The results speak for themselves though. The world clarity and performance are much better. Glad someone else finally noticed this as well.
This sounded interesting so I tested it. I setup a recurring 8 minute VFR flight over Philadelphia. I takeoff, ascend to 2000 ft, and land mostly using the autopilot. The top result is my normal settings, then I tested 200/50 as has been mentioned and also 100/100.
The 200/50 combination was more clear than my normal settings but ran a little worse although still acceptable. The 100/100 was equally as clear as the 200/50 but performance was unacceptable. So the 200/50 was the clear winner for me. Note that I tried other combinations such as 250/50 and 200/40 but they either had bad performance or bad visuals.
Finally, this chart shows how I prefer to fly. I cap my framerate at 31 fps (I tried others and 31 works best for me). This chart is a comparison of 200/50 uncapped versus capped. You will clearly see that the 1% lows and 0.1% lows got a significant bump in performance. When flying with the 31 fps cap the visuals were completely smooth with no stutters.
Ok - tried it. Not for me - too much visible aliasing for me at this render res of around 2200x2100 with OXR 200/TAA 50. But perhaps my eyes are spoilt from OXR 80/TAA 100.
200/50 working good for me, less rubberbanding too, not sure why…
Openxr res definitely improves exterior sharpness, far off mountains are way clearer.
Trees were shimmering more though, and looked like they were superimposed on the landscape, lowering them to medium has helped, as has turning preview mode off.
Keeping these settings though tomorrow might be a different story.
This seems the same strategy as the one suggested and discussed in the "TnT Quest 2 Sharp and Smooth " topic, but with a different headset of course. I have personally preferred to keep 100 render scale to keep glass cockpits as readable as possible
I’ll try those, thanks. What are you measuring and getting this graph with? I’d like to analyze my frames like that to make my own conclusions on my PC.
A while back it was reported and I saw first hand different rendering qualities depending on OpenXR and game rendering settings being used.
In general terms the EFIS displays and cockpit was more readible / improved with higher in game rendering and the external environment was improved with higher OpenXR rendering.
So yeah the two were slightly different. I’ve not played with this for a while so not sure if this still holds true.
I was sceptical but I tried it last night, together with setting some process affinity away from core 0, as suggested by Ramasurinen earlier in this thread. To my great surprise I gained 10fps and the view out the cockpit was as good as my previous settings of 90 OXR and 100 in game render scalings, with lods at 110.
With clear skies away from a city I was getting 45fps in the Stearman, dropping to 40 fps with scattered clouds, set at ultra. This on an all AMD system, 5600X and 6800XT, using a G2 headset.
Edit: As you can see in this frame, I was limited by main thread, in game it was constantly switching from main thread to gpu limiting and back, so it was nice balance.
Tested it with MR on (can’t stand it off) on my 3080 / 5900X. Lots of microstuttering, unfortunately. Otherwise smooth, but punctuated by frequent micro-freezes, much more often than OXR85/85 or OXR80/100 or even OXR100/TAA75. I didn’t check the FPS, it could be higher, but microstuttering ruins it. It does look sharp outside overall, but I can see some aliasing/shimmering in the trees etc. and glass cockpit seems to be less sharp. Installed new today’s Nvidia driver, and windows update as well. I tried some more Process Lasso tweaking. Too bad there are microfreezes. Otherwise would have worked.
