Gunnars Driver

Just tried to fly on the most populated open server i found, aerobatics online kaukasus, was 54 players online (out of 75). Flew about 20 min on that server still, mostly, the CPU frame time was slightly lower than the GPU frame time. Sometime the frame times where about the same( this wasnt the case before with the 2080ti when the GPU always produced higher numbers).

This is probably quite common, to have the BIOS settings for power duration so that the CPU either doesnt reach the target clock or that it clocks down after a while on high power, just as Intel intended them to do. Also, buying high speed RAM memory and not setting the correct profile to get them run as intended it probably qiute common but it wouldnt have that high hit on performance. The above is the first thing I tell about what to check when people have problems with low fps or high frame times. Thats also the first thing people ignore when reading the answers and tips&tricks from others.

A lot of people think that this is the case for everyone. For me it isnt. I dont know how other people configure their computers, or how they do their setups. (I don't even know if some of these actually run a Celeron 300mhz from the late millenium :-) just joking...) I know people run on servers that loads the CPU more than it can take. But this doesnt mean that DCS is always CPOU bound. For me DCs is more or less GPU bound 99.99% of the time, at least it was with my overclocked 2080ti OC, but one could always hope to get CPU bound with the 3090 because this would mean that my GPU can eat all the power my 9900KS can deliver, and then I know I have one of the best DCS rigs there is, for now.

I look for minumum fps when tuning the system. This was a short test to just follow the benchmark from another thread(posted there as well). Fpsvr doesnt give the minumum fps but it gives maxmimum frame time. Old 2080Ti OC: 99.9th percentile: >30 ms ( = less than 33.3fps) New 3090 OC: 99.9th percentile: 18.1 ms ( = 55.25fps) This means the 3090 have a 76.5% higher minimum FPS with my settings in DCS for the reverb. Nope, I wasnt CPU bound at all, at least not before the change. When gaming I never see the CPU frame time go above the GPU frame time. In the pictures below we have the frame times for CPU, GPU and also the usage for CPU and GPU during the actual benchmarks. You can also see the max CPU thread use( the load of the most stressed core). Its clear that the CPU frame times stays well below the GPU frame times, and its also clear that during the most of the time the CPU its was stressed less than the GPU lowest stress. CPU mainly about 43% ont the most stressed core and the GPU 75% for 2080ti. We can also see that while the CPU got more stressed with the 3090 it still is below the GPU. With the 3090 the CPU 99 percentile is 13.6ms(= 73.5fps) the CPU is 17.7ms( = 56.5fps). In this benchmark my computer is not CPU bottlenecked at all, not even with the 3090. (For that part, when doing multiplayer Im often the server and with 5-6 other players and 5-10 AI aircrafts and 5-10 ground AI my CPU isnt still bottlenecking)

Yes, only refering to the setup to get the system running, not the specific settings. I think it would take a lot of unnessesary time to write up the setup, so if OP just does as the giude say and then tune as we say.

I just updated my rig from Asus ROG Strix 2080ti OC to a Asus ROG Strix 3090 OC. No settings changed, just dropped the 3090 in. didnt even clean up och change the Nvidia drivers. I did run this benchmark: https://www.digitalcombatsimulator.com/en/files/3309507/ Before: 45 fps average with my DCS gaming settings. After : 72 fps avarage with the same settings. Thats a 58% increase in FPS with my current settings together with the Reverb Pro. Curious to see how it can speed up the fps with my new Pimax 8KX.

Just put my 3090 Strix OC in place: 45.5 to 72 Fps, thats a 58% incease in frames per second compared to my las post above. Same settings, at least I tried to make sure nothing is changed( I got the Pimax 8KX one week ago, so I have changed some settings but they are back). Yes, it was worth the cost. I also did run a test before the change, to be sure that no driver or dcs updates did change things:

1. The main setting for 90hz is in WMR/windows mixed reality. Set it there. I think it is possible to set in steamvr also, but for now. 2. PD is the same as supersampling, but PD 1.5 is the same as 122.5% Supersampling. There is a slider for automatic resolution setting, set that to off. If settings are per vr4dcs the slider now will make default 100% SS = about 2206*2160. Do not use more than 100% initially, because it is hard on the GPU. The goal is to get the system produce fps enough to get it enjoyable when ut comes to smooth gaming. It will have low settings and not look the best fort starters, but it should be smooth. sett MSAA to off initially. there is multiple posts about the settings, heres one thread, read my post about specific settings: Help Me With My Settings - ED Forums

Next part is the nvidia settings, vr4dcs has a guide for this(just found it and it seems in line with my findings) Link below. After following or checking these guides we can help with the rest in this thread. One thing that will kill fps is using supersampling in DCS pixeldensity and SteamVR. SteamVR also have dual supersampling settings that multiply( ask me how I know). Only one of these three choises should have another setting than 1.0 / 100% when the setup is done. During initial tests and settings, none of these should be avove 1.0 / 100%. The initial goal is to get good frame rates, most probably with too low settings and then step by step increase the settings for picture quality and whatching the frame rate take a drop. When the mix of frame rate vs quality passes your personal preference its tine to stop increasing settings and maybe back of a little. 30hz seems like you did set the headset framerate to 60fps in windows mixed reality. suggest to start with 90hz and after a while maybe trst 60fps if needed. Many gamers do not like 60hz, and it may also increase the nausea probability. It did a short test and I dont like the 60hz flimmer. https://vr4dcs.com/wp-content/uploads/2020/10/Nvidia_CP_VR4DCS-c-1.pdf

Your hardware should be good enough for a descent DCS experience. Most probably your setup lack something or settings are too high. For example, Pixel Density should be set to 1.0 in DCS. (Theres multiple ways to do supersampling, and PD is the most demandibg for the CPU. Its a lot to write about to describe a total setup. Therefore, I recommend you to follow this guide, or at least check that the most important parts are done/set as per the vr4dcs reverb guide(link below). The guide might be slightly dated but is still valid. The fpsvr app/software (4 usd on Steam) is very helpfull when tuning the settings for a good experience. could post two links in the same mess so I will split it. Reverb: Initial Settings for DCS World VR Thud's VR4DCS

Yes, we have a long way until the VR systems reach the retina resolution. just got the 8KX the other day so I cannot really say for sure but it looks like 8KX has more fill density in each pixel making it more clear than Reverb G1(have it too). Yes, first eye contact can be really tricky and there I think a lot of users domt really understand the real situation. Also, training helps a lot with this ( in real life, not 100% transfereble to DCS/VR) and I guess the most DCS dont have that training. QUOTE=Snake122;n7117632] Sometime you should come fly one of those bug smashers, their more fun than you think :lol:. I did fly fixed wing and I also have some civilian time, among these also 90kt Bug smashers :-) Yes, it can be really fun and in many cases its more about real flying than with bigger aircrafts. (no pun intended, I just find the expression so fun I had to use it. I know it sometimes used to bully others, but please dont take it that way. I have the greatest respect for everyone flying, and also for the ones doing DCS only).

Yes, not within The usual ”14 days” :-) From the thread about Vulkan API written by NineLine(community manager): So we need to take actions to be able to enjoy DCS during the time. I dont think it is too bad though. I guess it is usual for people to use too high settings, getting stutter and then be unhappy.

No, the visual acuity isnt on long ways as full 20/20 Eyes, not yet and not for long I guess. But that isnt the same as seeing other aircrafts on shorter distance, it is depending on how DCS draws a aircraft. Also, in real life you have (should have) 20/20 eyes and then you can see more of the aircrafts shape and possibly ident the type further than in DCS. With the Reverb at least its mostly possible to detect another aircraft furter away than in real life said from my experience as airforce pilot, 30 years/6K hours. [Edit] I have an example from todays maintanence test flight I did. After leaving the control zone we got flight i formation from the tower about a small civilian aircraft( Cessna or other 90knots bug smasher) same altitude and we got it related to two places on ground it was training around. Distance on the moving map was 5Nm to the closest place and 10Nm to the furthest place. we vouldnt see this aircraft, despite two pilots with 20/20 vision and a sparkling sun with very clear winter weather. We tried to descend to low level to get it in sight but no joy. We didnt need to see it but had the opportunity to try to, for the purpose of this issue.[End edit]. I know, with first gennVR(oculus rift for example) it was hard to detect but due to low resolution. With the Reverbs, its way better. One problem I see in this VR section of the forum is that people believe you can see another fighter at 20 or 30 miles in real life and then the complaints is about VR being bad. I think maybe the reference is to 2D DCS and maybe DCS overdo the visual range on a flat screen(havent really played it in 2D). Another thing is people dont know how hard it is to detect other aircrafts far away.

Got it, but I couldnt join due to missing mandatory modules and maybe a mandatory mod. Instead I went to the most populated open server I could find, the I didnt see cpu frame time numbers even close to yours.

I use no labels at all and think its quite ok.

Theres articles on the net about the 5950x being clocked to 6ghz. That would be something !

No pm received...?

Ive never used it. Another player in our group usually do the tacview stuff and show us afterwards. A good idea to make sure that software doesnt load the dcs thread/core.

Yes, Ive noticed it. Regulary seing close to 11 out of 11. In fact I have reduced settings to lower the VRAM use/allocation and I think it made me see stutter less often. In missions I sometimes see RAM up to 27-28 Gb. And of course, I see low usage in FS2020 but its not 3D double frames(yet).

Does it put the load on the same Core/thread as the DCS main thread? (If it’s not it shpuld cause that much effects...?) I dont know, havent used that software

And no quality/readability degadation from lowering this value at all ? If so, and if it gives gain in gaming I have to try.

Im not with you there. Two reasons: - First of all we had the discussion about getting a new CPU or not. In this aspect I /we are saying that if the current CPU can be clocked to about 5Ghz, it will perform about the same as a new CPU@5Ghz-ish. Therefore, there is no real gain in DCS with a new CPU if its clocked to 5.0. -Second, with VR we know that the load on the CPU is hard, and if we have a too slow CPU-clock it will sure be a bottleneck. Clocking from 4.8 to 5.0 might only be give 4% extra CPU-power but it also about 1ms shorter CPU time when close to 22ms. The difference between keeping the CPU frametime below 22ms(for me with Rift at that time, Reverb today) is crucial. If it stays below = smooth gaming, if it goes above= stutter. Its not only a fps question, but a question of less stutter. For me, when I got the first VR-set 2016, I had a i7 3770K@4.5Ghz (+GTX1080) and after increasing the clock to 4.7-4.8 I got much smother gaiming( this was before fpsvr for me so I couldnt see the frame times. One year ago I tested 5.0ghz on my i7, it could do DCS but didnt make other stress tests. I found out that I had much less stutter but I couldnt rely in that CPU@5Ghz so I decided to upgrade. OP will most probably upgrade the GPU (3080?) and that will make a ”weak” CPU limit the graphic settings to keep the stutter down. I say that having a CPU that probably is capable of 5.0Ghz is denying yourself from a better gaming experience, at a very low cost.

Which server is it, can I test for reference?

I think SS does about the same(for the jaggies). I have used oit to and from but I dont think I really saw any diff with SS on 150% in the reverbs.

What about frame times? I did a quick run, instant action the uppermost( free flight?) and not changed the map from default so I guess its the default kaukasus map. Also set SS150(shows 149%.) I have The Reverb, motionreporjection set to auto so its always 45fps when it cant do 90. I think you should do a check on this for frames times etc to se if something differs. Adfter that I could do the same mission as you use that make things slow.