Vitralette

I will just repost what I already wrote regarding resolution. x1.7 will indeed result in native resolution for the sweetspot of the quest 2. Everything lower is undersampling. No other program shows any problems with these settings except DCS. But, just to answer the question, the problem persists for any other resolution tested (down to 1.0 @ 72hz). The reference given here is not some guy (Volga Aksoy) on twitter but one of the lead devs on the quest 2.

Well could be that things changed through the updates. At the time of testing these were my findings for my system. Due to the fact that the problem described in this thread was never addressed by the devs nor was anybody here apperarently able to solve it, I mostly quit the game, so I never followed up on this config. It's a shame really, but you can put all the budget into dynamics and systems as you want, if the engine sucks, it sucks.

Bump. Issue still persists. This Issue is still the reason why do not spend any additonal money on the product.

+1

Don't use Oculus Tray Tool (OTT) because its not being supported any longer by the dev. If you have to then leave all settings to default. Meaning PD = 1.0. Oculus Debug Tool (ODT): Same as OTT. Leave everything at Default. "Pixels Per Display Override" = 0 (same as PD =1.0). DCS - PD = 1.0 The only slider you crank up until your PC melts is the render resolution slider within oculus software. If you can set it to max (5408x2736) your Q2 will run at native resolution without any up or downsampling.

I have extensive experience with the Varjo XR-1 used for flight simulation due to my work. So I can't give you a comparison against the G2 but against the Ferrari of VR/MR (at least it was until the XR-3 came out about a month ago). It still has way higher sweet spot resolution than the G2 (XR-1 60 ppd vs. G2 22ppd). Compared to the eye-wateringly expensive "human eye resolution in sweet-spot" of the XR-1: Has the XR-1 higher resolution? Yes. Much better? No. Is there a "oh my god this is amazing compared to Q2" in the XR-1? No. Price to Performance? Q2 blows everything else on the market out of the water. So to say I am impressed with the Q2 would be an understatement. Ontop of Oculus Link providing a brilliant PCVR experience due to ASW support, you also can use it in wireless mode via Virtual Desktop which works flawlessly (e.g. Half Life Alyx). Though this may not be of any interrest to you if you are only looking for a device for seated simulation. But to me this is an immense feature. I am certain that wireless VR is the short term future, once Virtual Desktop is able to add ASW or a similar technology into the mix (ATW is already supported).

Thanks @Waltan, but it didn't do the trick for me. Tried that setting before messing with the ingame settings. @speed-of-heat yeah but many people seem to tick that one on at one point or another. I've seen people mentioning this in a similar context before.

Slider in Oculus Software to max, if your PC can handle it. I would not recommend using OTT anymore since the developer stopped the support. https://www.reddit.com/r/oculus/comments/5okoju/oculus_tray_tool/ Pixel Density in OTT or ODT (Oculus Debug Tool) to 1.0 (in ODT that is just a 0 for default settings). I would also recommend to leave every other setting in OTT or ODT at default because since Link v.23 that stuff is all tuned automatically based on your slider settings, except for forcing ASW of course. Ansonsten auch gerne auf deutsch

Oculus App lets you set the render resolution not pixel density. To achieve native resolution (3664x1920) in the sweetspot you will need to set render resolution to 5408x2736 (due to lense barrel distortion). This is the most confusing part about the Q2 setup because most people think setting the slider to 1.7x at 72Hz or 1.5x at 90Hz (=5408x2736) is a supersampling multiplier, which it isnt. This Tweet by Volga Aksoy explaines this.

I am baffled by how little the impact of the DCS graphics settings are for high resolution HMDs. So here is a list of things that don't make much difference and only use resources for the Q2 set at native resolution (5408x2736 render resolution): Textures - For the Hornet this doesn't make any difference for the cockpit, your aircrafts may vary. Airframe textures, runways etc. are lower res but I don't care about that. Set it on low, saves VRAM. Terrain Texture - Doesn't make a difference on Caucasus. Set it on low, saves VRAM. Visible Range - Distance at which buildings (not aircarfts or ground targets) are rendered, also the distance at which the terrain is getting rendered based on your terrain texture setting. If I wanna go sight seeing im gonna play MSFS2020. I put it on low. Res. of Cockpit Display - Only effects the resolution of rear view mirrors. Since rear view mirrors are bugged and half your frame rate gets killed when activated you never wanna have them on, so put the resolution to 256 in case you accidentially activate them MSAA - no difference ingame (shockingly), doesn't improve shimmer OFF (in case it wasn't already) SSAA - no difference ingame (shockingly), doesn't improve shimmer OFF (in case it wasn't already) Reload Radius - no difference 100 Anisotropic Filtering - no difference (shockingly), maybe less impact because Nvidia "texture filtering anisotropic sample optimization" was ON during testing OFF Most of these settings don't come which a significant performance impact on hardware (except MSAA) thus I was just putting them on higher settings. I mean if it comes free but I can't really tell the difference I rather take it right ? Apparently wrong. Now that I am using the opposit strategy by only activating settings that I can actually measure with my eyes ingame the problem seem to have solved itself. Maybe because less VRAM is used (although all tools still show that it's maxed of course ...) or maybe any of these settings is bugged (just like res. of cockpit display).

@ppokit I might have good news for you. I tried a new set of settings which seem to have removed the gpu firewall of death. Now, I am currently still evaluating, but I did about 5 sessions now over half an hour with no problems (which wasn't possible before). Doesn't mean it can't come back and this was just luck but it looks promising. I wanted to wait for a write up till I figured out which setting exactly change the behavior but I would appreciate if you could test these and post our results. I lowered both textures and clutter in order to conserve VRAM. I dont see any difference in the Hornet cockpit, but I am using this mod Clean cockpit for the F-18 (digitalcombatsimulator.com), so this may explaine why the setting has next to no optical impact. I also turned of preload, also no difference ingame nor are the loading times longer. Dunno what it really does. Nvidia Vsync on fast is really important I already tested that. Render resolution in Oculus App is your choice. I realized that at 90Hz the difference between 1.5x and 1.3x is marginal (at least in DCS for some reason) so you can save a big chunk of performance there. But with these settings and 1.5x I still have about 30% headroom with ASW45fps enabled so I just crank it up. The rest needs more testing regardings its importance. I'm gonna follow up soon.

Well yeah I seen some of these posts and people keep bringing up VRAM, but I've jet to see real evidence that VRAM is actually a problem. I've someone got a link to a thread where this gets actually shown I'd really appreciate it. That DCS is using ungodly amounts of VRAM is definitely true (MSFS2020 VR on ultra textures only reserves 7.2GB for example), but that this in turn leads to instant fps break downs once a mysterious threshold is reached isn't obvious to me. Continous performance degredation? Sure.

Could you be so kind and explain how a full VRAM is actually effecting the system? Fom what I "understand" is that once the VRAM is full it starts to copy data into RAM which can lead to FPS drops, ok makes sense. What I do not get tho is why I can fly for 30mins with a full VRAM right from the start and THEN all of a sudden it is a problem (as can be seen in the data provided 3 posts up)? I checked if maybe the shared memory is getting filled up (would then start to copy to pagefile/ssd) but that is not the case, at least it's not getting reported by either TaskManager / GPU-Z / Afterburner. Maybe this whole thing boils down to me not understanding VRAM usage at all Thanks in advance!

May I ask what happened to your G2? If this can even be determined at this point Anyway sorry to hear you had a bad experience with it. I also have this feeling that I can provoke it to happen by opening F10, going external etc. But then again, as shown by the example runs in the prior post it will also happen without using F10 at all. After the GPU spike I had another look at background processes and found desktop windows manager (dwm.exe) was using almost 20% of GPU 3D rendering resources! Now this was very variable, switching between 0% and 20% - maybe the reason I didnt register this the first time looking at the processes. Now I do not know if this is normal behavior. This could be just because I need to look at TaskManager thus triggering dwm.exe to do something. But I checked how much its using in idle and it stays around the 0% mark. During ingame GPU spike - dwm.exe GPU usage reaches 20%! Idle - dwm.exe GPU usage hovers around 0% Edit: I reactivated the integrated GPU and then told Windows to use the iGPU for dwm.exe. Now the dwm.exe doesn't go above 10% and is running on the iGPU. The issue remains tho, the only thing changed is that dcs.exe is now using the full 100% of the dedicated GPU . But at least I can rule this one out.

Today I was able to fly for 30 minutes on a MP server with about 10 players. Everything went smooth. After landing I restarted DCS in order to see if I could reproduce the problem or if I would get another clean run. Unfortunately the problem was right back the second time around. Same server same amout of players. After 10min I hit the GPU wall as described in the inital post. Good thing about that is that I can now provide some data. First Run - Solid 45fps for 30min (on MP server with ~10 players) Second Run - Hitting the wall after about 10min (same server as Run 1)