Gunnars Driver

In what game mission ? How is your 7700 set up ? Doest it run @5ghz on all cores and does it do it all the time ? The frame timings seems odd. I have a 9900ks at 5.2, but I also have run it in 5.0. You CPU frame times is much higher than mine. I also more or less dont see cpu frame times ablove 10 if not on a heavy populated server. My CPU/GPU frame time portions are the other way for me: GPU is more or less always higher. And, I can use decent high settings and have a very smooth ride. Of course, the aircraft type makes a difference as som moduel makers seem to produce much more hard driven modules. You should check your bios setup so you know you get the most of the system. Suspect there is more to get fron your system. HWinfo64 would be. Usefull. 15ms is very close to the 60fps limit for 60hz(16.67ms). Per definition fpsvr only measures the dcs thread frame time. There can be other processes taking time from that same cpu core making the real frame time higher. I have tested this in my computer and for me this is NOT the case. Frame timing correlates very good to actual fps for the limiting part so it mostly shouldnt be an issue. However, you might wanna get yourself a better marginl to the actual max frame time limit for the selected headset frequency/Hz. If you use 60hz the maximum frame time limit is 16.67ms, but to not get frames every now and then that shoots above 16.67ms you need to perhaps aim for 12 ms or less as the ”normal” you se. Most probably your cpu make frametimes that is above the limit which causes stutter. 60hz is flickery for most people, I think. I dont like it so 90 fps is the way to go for me. ​​​​​​​ Why not use 90hzx and reprojection that gives the headroom for 45fps/22.2ms max frame time ?

100% reported in steamVR is quite high resolution, dont remember but can check it up. With pitool supersampling set at 1.0 its not as sharp as my reverb at 150%. I earlier forgot that I was comparing to the reverb at 150% and I havent run the reverbs at 100% exept when testing almost one year ago so I need to check it up. I have a feeling that reverb still is slightly sharper but the downtuning to 100% will even out part of the difference I have seen. Have to get back on this. I think the 8KX has very much better resulotion and sharpness than the frist gen VR and for the cockpit most text is possible to see without leaning forward, but not everything.

What CPU and at wich clock speed? overclocked ? RAM memory? size and RAM-speed ?

I have the Reverb G1 (also have OR CV1 and Pimax 8KX). I have en 9900KS@5.2Ghz and Asus Strix 3090 OC. Had Strix 2080ti OC before and it did run really good with the Reverb. A lot of people start by using way to high settings. DCS in not easy driven for the hardware and very different from other games when it comes tohardware needs. I suggest start with settings low enough to be sure that it should keep above 45 fps at (almost) anytime. This should give smooth gaming when using reprojection. If it doesnt keep 45fps we need to find the reason. Maybe wrong BIOS settings for the CPU, not allowing it to run att full speed all the time, or problems with DCS( shaders etc). For starters, when testing do not use a complex mission that bogs the CPU and or GPU down. Beginning with too high settings will not be a good idea.(it doesnt matter you have the 3090). The goal is to get smooth gaming without stutter and then slowsly increas the settings step by step and when getting "non smooth gaming" or stutter that is too much for your own taste, back of until DCS is smooth enough again. The final settings depends on personal taste and acceptance to stutter or non smooth video. We are all different when it comes to this. You can see the frame times and the load on % on the most loaded core of the CPU and also the load on the GPU in fpsvr. use that information to tune.

PD should be set to 1.0 and all SS should be done in SteamVR. The setting in SteamVR isnt logic, the generic SS multiply with the per application SS, so you should not use both. To start, use only 100% in both generic and per application. The setup overall for WMR has some possibilities to get it wrong. I suggest to follow the reverb guide in vr4dcs. The DCS settings: Shadows should be flat or of. Terrain textures, start with low Heat blur = off MSAA start with OFF Depth of field = OFF Lens effect = none motion blur=OFF SSAA= off Clutter and grass= begin with minimum. Trees visibility = start with low setting. Chimney/smoke = 0 Anisotropic filtering 2x for start. Terrain shadows = flat Cocköit global illum = OFF Also, make 100% sure that you are able to activate and deactivate reprojection. Its really a key to success. Suggest reprojection "auto", which should give you 45fps for most of the time. If you see values between 45 and 90 fps more than very brief there isnt any reprojection and the ride will not be smooth. Get the fpsvr app, 4 usd at steam, tuning and understanding what is the limiting factor is very easy with fpsavr (also detecting other problems that causes stutter etc). Its easy to see what part: CPU, GPU, RAM or VRAM, that is the limiting factor and also easy to detect the effect of changed settings - tuning has never been as easy. The initial setup for Reverb: https://vr4dcs.com/2019/09/10/reverb-settings-for-dcs/ Plus the tunings that can be done in nvidia control panel: https://vr4dcs.com/2020/10/19/nvidia-control-panel-settings/

Is there aircrafts without a clock in DCS? IRL I selldom use the stopwatch, besides from system support the easiest way is to use mental arithmetic. (you dont have the app IRL either). There is rules of thumbs and simplifications you can use to come up with the answers quite easy. It still needs a little bit of training to be quick but I still think its a better way than a calculator. Maybe, for the ones that can not do mental arithmetic at all, an app could be it. Still, if being average on this you should be as good at the DCS computer as somebody thats really good on it on ground but wheres a flying helmet. ”The helmet”and IRL workload reduces that capacity quite much. 600 knots= 10 miles( Nautic) per minute 500 kt = 8 miles/min 400 kt = 7 300 kt 5 200 kt. 3 100 kt. 1.5 Knots valid at low altitude. Values start to deviate above 5-10K feet of course depending on how exact calc you need. Mach works ok on all altitudes. At low altitudes(warm air) calc as below is about 10% to low and at high altitude (cold air) its quite right. Mach values x10 = miles per minute Maxh 0.9 = 9 Miles Mach 0.8 = 8 Mach 0.7= 7 Mach 0.6 = 6 Mach 0.5 = 5 Mach 0.4 = 4 Mach 0.3 = 3 Mach 0.2 = 2 So 30miles high alt m 0.5= 6 minutes. 30 miles at low alt at mach 0.5 = 5 min 30 s ( 6 min - about 10%) [Edit] Had to walk the dog. You can also use Groundspeed from INS/IRS/GPS. Isnt affected of altitude. Hope this helps.

Id say the immersion would be a liltte killed by zoom in VR. when I used the CV1 I used high pixeldensity to sharpen the picture and leaned forward to se some instruments.

If you did read the microsoft statement i linked to, it stated that the GPU VRAM usage in the perfomance tab is the correct value. The performance tab gets the value from VidMm which is the operative system GPU Video memory handler. As I said before, theres a lot of misinformation on the net. I guess the use of the term [allocated] comes from people mixing dedicated VRAM with shared DRAM memory. I cant really be more clear about this. As for the OP I also enjoy no stutter when loading missions. Why, I dont know but its really freeing to avoid it.

How should they respond? VRAM usage will depend on which VR Headset (resolution), graphics settings and the played game( single or multiplayer, weather etc). I do not think ED follows all thread like this either so I do not think you could expect to get an official answer here. I would say checking in Win10/task manager/performance tab is the best answer you can get. Advaned users may be able to use GPUView but it wont be as easy to use as lokking in the task manager.

Yes, I understand the 3080 shortage doesnt help the ”which to get” issue. To be clear, I dont say anyone absolutely need 24Gb VRAM. Regardless of which GPU we buy we need to tune the settings in DCS. ( Other games might differ. I already had every setting in Asetto Corsa on absolute maximum with the 2080ti). So we get a new GPU and we tune the game and the system. When done we have smooth gaming. What might differ is the final setting, how high we can tune things. Articles say that the new Ampere GPU( 30X0-series) use VRAM on another way, more effective than Turing/RTX20X0 series. 10Gb in 30X0 series might be better than 11Gb in 20X0, as 3080 is sold with only 10Gb It probably is. For me, Im quite sure that I have been limited in DCS with 11Gb with the HP Reverb and I also had the Pimax 8KX on order. It was quite easy for me to do the choise, as I could afford it.

Probably to high settings for the map/aircraft. There are a few settings that isnt worth using for VR and also some make about the same effect: choosing the one that use less recources could be a win. One example is Supersampling vs MSAA. MSAA reduces jagged lines but Supersampling also does, plus intease the sharpness of the instrument panel. Ise too much of both= 5090ti needed. Havent gotten the Syria map (yet?). I have buttersmooth flights with F-18 in Kaukasus with Reverb, still with the same graphic settings as with the 2080ti. It was smooth before, tuned for 2080ti it of course isnt that hard for the 3090. I recently got the Pimax 8KX and I havent tuned anything yet( still wait on the mighthouses), so using 8KX with these settings is hard on the GPU. With normal FOV(150) it keeps up with the 37.5 fps needed for reprojection but it isnt that smooth if I look to the side at low level high speed. Tuning will be needed.

Here it is directly from microsoft: https://devblogs.microsoft.com/direc...-task-manager/ As I read the article, its clear that used memory is reported. I would also guess that people did read this article or similar and mixed up the text avout allocated memory that really is split DRAM allocated as video memory. The allocated part has nothing to do with the dedicated video RAM we have on hour gaming video cards. What I found after reading and checking in the task manager and fpsvr the reported memory in fpsvr during dcs gaming the amount reported is more or less the same. Fpsvr is clear that the reported valuers for fram times etc only coers the actual thread and that there could be other processes taking more time(ive checked this earlier and there was no deviation for me), so the reported delta between 19.7Gb in the task manager/performance tab and 15.3-15,7Gb is closae to the 2.2-2.4Gb used if I close DCS. My conclusion is that What I read in the fpsvr is at least quite close to the used VRAM.

Logging in to a certain server is starts with 15.6Gb wich after a few seconds increases to 15.6 and then 15.7Gb. When selecting aircraft the reported VRAM increases to 17.1Gb. This is the Fpsvr values and the performance tab in windows show a little bit more( fpsvr onfly shos the actual thread usage and not all processes of windows. The windows desktop probably accounts for the difference between these two. After a while I se 18.1Gb in fpsvr, increasing steady to 19.3Gb. You still think the GPU only uses 7Gb or so(this discussion was present when Nvidia increased the VRAM from 8 to 11Gb also, at that times "it never went pass 8Gb). If it uses a lower value, Why on earth would it progressively increase the allocation to 19.3GB ? During this flight, after landing it decreased the reported memory to 17.3-17.5Gb, as seen in the picture. When closing DCS the reported VRAM in the task manager is 2.3Gb, so that add up. Heres a statment refering to microsoft that the performance tab reports the used VRAM and not the allocated: https://www.extremetech.com/computin...reators-update This might be an interresting read: You probalby could use GPUView if you dont believe microsoft. [Edit]Bugger! I seem to have moved the fpsvr in the pic below so that the VRAM usage cannot be seen.

So your thought is that the allocated memory do not have anything to do with the used amount? I think, as we dont have a better source it is of interrest. If my GPU allocates 9Gb in one game/mission, and 6 or 15 in another game/mission, would it not use about the allocated amount? I know I could get stutter when I increased settings that showed close to 11Gb on my 2080ti and I know the stutter reduced or dissapeared when reducing the settings so I had a better margin. This despite the reports for GPU usage not hitting 100%. I didnt get a higher allocation with the lower adopted settings when changing to the 3090 but when I increase the settings I see usage over 11Gb. Why do we think it allocates only 8 or 9 Gb in some games/servers and 15Gb or so in others? Are we that sure it never uses more than 11Gb ? I think it does. After windows 10 Fall creators update a few years back Microsoft said that the reported VRAM in the performance tab is used VRAM. So you can check there.

You dont need Steam /SteamVR for non WMR( Windows Mixed Reality) Headsets like Oculus and Pimax. With the Reverb that is WMR you need SteamVR. I have seen some threads about using WMR without steam but not digged into it. There isnt any problems or troubles to use Steam so I dont see any idea to round Steam. fpsvr is a Steam app so I dont think you can use it without SteamVR.

What level of foveated rendering do you use? I havent tried it yet... Tried a public server( Aerobatic online Kaukasus) togheter with smartsmoothing/reprojectioopn and it actually worked quite nice. Almost completely smooth.

I made a simple missions, just Kaukasus and my F-18, to se the graphic perfomance without any other loads on cpu or gpu. Same settings, CPU is not the limit at all, and the GPU doesnt limit for the goal to keep 37.5fps. Even the GPU 99.9 percentile is 20% lower than needed to keep 37.5(75fps with motionsmoothing).

This is the fpsvr values for a short flight in kaukasus. F/18, my computer is the serverfor the mission, about 15 AI ground units, and about 5 other AI aircrafts. SteamVR set to 100%, resolution according to SteamVR = 3772x3160 per eye(3792x3160 according to fpsvr, se below . Pimax 8KX in Normal FOV(150 degree), 75Hz Smart smoothing in Pitool = OFF to be able to se what fps the computer is able of. 75Hz and using smarth smoothing( reprojection/ASW) gives headroom for 26.6ms frame time and still keeping 37.5fps (=75fps smart smoothed).

Ok, yes at least youre within the limits of the Reverb G2 now :-)

As I just got the 8KX and the 3090 I didnt change settings in DCS. I had the highest possible for buttersmooth gaming with the reverb, just changed the SS to 100%. It will be possible to lower the DCS settings if needed to not get into trouble. I still havent git the lighthouses so gaming is a bit ockward with only 3DOF. Ill be back with my settings, and maybe a movie. I wont have the time to fine tune yet( the idea with getting the 3090 is to be able to keep as high settings as possible so I dont need to use [LOW] on all settings. Further, its so little difference between large FOV and 150 degree Normal FOV that I think Ill use normal 150 degree.

According to those numbers your far ipd is 64. You should use far IPD for VR range. Whats very strange is that in the normal case near IPD should be a smaller than far IPD and I can not even see how a near IPD would be a grater value than far IPD ( Near IPD= eyes angled ”inwards the nose” moving the pupils closer thogether). So I would guess the optician made a misstake when the numbers was written down...unless your eyes are angled like Marty Feldman. Anyhow, these numbers clears you for the Reverb G2 but you really should get quality secured numbers. Call the optician and ask for clarification. You would like to know and not guess, for the best VR gaming experience.

I use 2.5.6 open beta. Couldnt find blueflag will look for alpenwolf

Any example of a server like this to test on ?

I have both the Reverb G1(sqame resolution, same FOV) and Pimax 8KX. The reverb is sharper. Reverb G1 has a lens thats only sharp in the centre, but its good damned sharp. If at 100% Supersampling its good and you can redan the most in the cockpit, with SS 150%(or more) its really sharp and everything you need to see in the cockpit is esay to se/read.(Edit) Hig SS is a bit costy on the GPU.... G2 isnt out but we now its the same resolution and same FOV so pixels per degree should be the same. G2 has a new lens that should increase the sweet spot where the picture is sharp. G2 fixes most of the issues G1 had(werent big issues, its really good to), but with increased sweet spot and mechanic IPD adjust it will be it. One could only wish for OLED screens or something that give better black levels and more accurate colors( not that importand for gaming for the average gamer I guess) :-)

Yes, all three latest posters is right. Two options, from which one isnt an option if you ask me: -Get an accurate IPD number to be able to make a safe decision and no risk of throwing alotta’ money in the lake. -Or save $13 and buy VR Headset on vague data, and risk to make a expensive faulty choise.