Gunnars Driver

I guess motion reprojection / Smoothing should be of, if anyone should co pare the delivered fps? In this test reprojection = ON As is says, SteamVR SS set to 150%(=149%) PD in DCS = 1.0 Settings: All as i use when gaiming. ...not CPU bond...

Same here. Aurelius test (+41%) give some indication on what to expect, Im sure there will be a lot of openings to either get 90fps more often, or to use the extra power to increase settings that increase the overall gaming experience. As Im looking to get a Pimax 8KX, I think I can justify a 3090 only because of that. Reverb at 150% SS as I use now means 18% more pixels to produce with 100%SS /PD1.0.

Same here. We shouldnt really need to defendingly explain why we did get the best CPU’s there is. Ridicilous. The thread is taking a new dead end track, just as the *CPU bound* track. The statement *not the best choice for DCS* should be changed to *not the most economical choise* for DCS. Big difference. Just about any CPU with at least two or three cores on 5Ghz+ is a good choise, today. Any CPU only running 5Ghz on a few cores might be not good in the future. Either buy cheaper CPU’s more often or more expensive less often. There is no wrong with getting a cheaper CPU getting the same DCS performance as with the overall best CPU. Its also not wrong getting the best CPU. The thread is about 2080ti vs 3090, and I guess that settles sort of the frames. You have a 2080ti (hi end) and and thinking about getting a 3090(hi end). But how many of us in description above have sneaked out on the CPU choise? Im waiting for my 3090 and I sure hope it delivers good enough to make my CPU the limit, so next step will be a better performing CPU in the future.

I dont think I ever saw the main thread switch core ?

1. Once an overclocker, always an.... :-) 2. There is also other applications: videorendering, CAD/CAM, 3d scanner rendering and maybe also another game... For me both 1) and 2) alone would do it.

Absolutely! Didn’t wanna insult someone not wanting to do any type of OC, so I tried a soft approach. I have never had cores running at different speeds, so that was one of my questions: -Do we know for sure that the DCS thread will be run on the highest clocked core? I also do 5.2 on all cores(9900KS).

Why run a 9900K at 4,7Ghz? Doesnt make sense. For running DCS that mainly run on ome core you should go as high as possible in clock. At least make those two cores run at 5Ghz all the time during gaming.

No, you dont add the fram times. The highest frame time sets the possible FPS. For VR and a 90hz headset, if both cpu and gpu have frametimes below 11.1ms you get 90 fps, and if both is above 11.1 but below 22.2 you get 45 if using ASW/reprojection. For your example above, if cpu framtime is 15ms you will get 66.7 frames per scond, and if changing to a better gpu and lowering the fram time from 5 to 3ms you still get 66.7 fps. The numers in your example is not really representable to people having a 2080ti, as this thread is comparing 2080ti to 3090. Most people using a 2080ti would have a processor that not shoots the cpu fram time above the gpu frame time. I have mostly 50% hiher fram times on the GPU(2080ti overclocked) then on my cpu(9900ks). The representable numers in fram time for me is 8-10ms for CPU and 13-15 for the GPU

Its so dependent from what mission and settings etc, it cannot be compared lika that. Im mostly between 7 and 12ms. <— of no value for Harlikvins question, I think. If to compare, use a standarized quick mission in the same map, at the same position ? And also, post settings.

Already on order.

Exactly. The usage differs a lot between people and with that the needs. A 1080P gamer wouldnt need to change his 2080ti at all I think. In my case coming from a overclocked 2080ti I think the win with a 3080 isnt as big as test shows with a standard 2080ti. As it looks for me I would get around X% better fps with a 3080 and about 2X better with a 3090. In that perspective, it does not feel wise to get a 3080. In my case its more of either get a 3090 (I can afford it) or wait until next generation. I have almost always bought high quality stuff and can not rememeber even one time I have regetted buying the best. If I look at frame times, I need about 40% lower frame times to make the GPU break even with the CPU. 3080 probably wont do that, but 3090 might come close. Doing VR-only gaming also is a high resulotion gaming. It seems the higher the resolution the bigger the difference between the cards. At 8K the 3090 seems around 25% better than the 3080. For my current VR-settings I am somewhere around the middle between 4K and 8K. 4K is 8,3Mpixels VR / Reverb+150% SS Is 14Mpixels. 8K is 33 Mpixels

Thank you for the info, should be good for cross comparing.

Was the hardware: CPU incuding clock speed and RAM memory comparable to your own ?

I dont know about the export.lua, but some other stuff: I recently saw geforce experience totally bog down my friends new build( 18core i9, 128Gb RAm, 2080ti). He had bad performance(DCS not installed, but in a other game). Needed a uninstall/total clean from geforce expoerience and then install the geforce driver only. I have exactly the same 2080ti, with the same driver and geforce experience installed, so it doesnt nessesary kill all computers. Also, deleting all shaders can help if having problem with dcs performance. Never done it myself, not had any problems. Information how to should be found on this forum. First: make sure the hadware is running as per my former post.

There might be some things to check. Main settings might be done on a less demanding map so it isnt bottlenecking the CPU regardless of settings. The newer maps seems more demanding. First of all: Hardware setup. How have you set your computer up ? Does the CPU: - Run on 4.8Ghz all the time or only for shorter periods? - Does the CPU run 4.8 Ghz on all cores or is it only one core ? Default settings sometimes dont keep the cpu clock and if it goes down, the cpu frame times will go up. Also, if not all cores is 4.8 I think it is possible to have the DCS main thread on a "slow" core, which also will give you higher cpu frame times. You need to make sure of these things first because a 2.4ghz clock will give you doubled CPU frame time and no dcs settings can fix that problem. Edit: Dont know how much you know about computers? Might be kicking in open doors. Theres a lot of ways check the CPU real clock on all cores. For example, HWinfo64 can be used free: https://www.hwinfo.com/download/ You can use the small window to se all cores clocks in real time. Correct setup you will have 4.8Ghz all the time(preferableon all cores).

1799usd ( US price) plus 25% (Swedish) VAT is more or less om the dollar what I payed.

In sweden there was a few 3090's in stock. Not many, and not all versions. This on serious sites that actually could deliver when 3080 was released.

Wilco!

I selected a Asus Strix OC. I already on my fourth strix OC so I just have to continue. + That as I read they wont start ship Asus strix until next week = where I ordered, they wasnt in stock, but expexted in a couple of weeks = I thought that not that many would go for the most expensive not even in stock. I can live a couple of weeks on my 2080ti strix OC, but barely :) Had two (swedish) sites open, they was possible to order on both and had one card on each place in the shopping chaert for a short moment.

3090 on order, since 15:00:30 today.

For me, motion reprojection was disabled by default. It also reverts to disabled after each steamvr update, needing to change back to ”auto”. I understood motion smoothing is reprojection but as I remember it is SteamVR’s version with a different algoritm than reprojection which is the WMR version. I did not like motion smoothing at all, maybe I didn't get it to run properly. Yes, I deleted the row when I tried motion smoothing. Generally, my experience from Oculus is that ASW* (2016-2019) is not near as good as motion reprojection. Double pictures/ghosting is frequent. With motion reprojection its not that far from true 90 fps in my eyes. I might be a bit biased on the motion reprojection because its that good in my eyes. *) Seen info about oculus released a new version of ASW so it might be better now.

I dont really think ít work with WMR/Reverb alone. I use motionreprojection. It works fine, better than oculus ASW did. I tried motion smoothing a lot when I did the VR setup and I really dont think it worked for me alone, without motionreprojection. Motion smoothing appeared like motionreprojection = off. I actually have motion smooting left "on" anyway, because I felt that it may help when motionreprojection is on. The above is how I found it when I did my Reverb setup. I also just tried it again, both motionreprojection only, motionsmoothing only and the both together. I thought that maybe it has been changed to the better. But I got the exact same results today.

For this test I did (forgot to write that) set reprojection to ”none”. This to get the real frame per second that the computer delivered. Otherwise it wouldnt really give anything. Set to [none] for the test and back to [auto] asap after :lol:

Yes, Its also stated in the fpsvr information on Steam website. Still, there doesnt seem to be a noticeably difference. When using motionvector in auto the switch from 90 to 45 fps seems to come very close to 11ms and getting stutter above 22ms seems to time with going below 45fps. [Edit]Just had to test: Did test both different instant missions and start servers with missions I usually play. The difference between GPU frame time(always limiting for me) 0.2, 0.4, 0.5, 0.6, 0.7 and 1.4ms.Sometimes also fpsvr reports a slightly higher framtime than possible frame time. As fps is reported in hole numbers the mean difference should be expected as around 0.5 just because of the rounded fps number. It certainly looks like the FPS frame times is very close and that the difference in most cases can be ignored when it comes to tuning your computer and also for comparing between other forum users. Different graphic settings should make the numbers differ a lot more. Explanation if needed: 1000 divided by the reported FPS = the mean frame time(of the highest of CPU/GPU) in milliseconds actually used for that moment. For the first upper fpsvr in the picture above, 1000/80fps= 12.5ms and fvsvr reported 13.2ms, that acutally 0,7 higher than possible.

As Hansa and Supmua says. I am between about 6-12ms. CPU frametime would depend on CPU generation, clock speed and the graphic settings. I did all tuning with the help of fpsvr since upgrading hardware and getting Hp Reverb one year ago. In my case the CPU frametime is always lower than the GPU frametime. There might be some settings that load the CPU more. With my old hardware i7 3770K 4.7-4.8Ghz /GTX1080 /Rift CV1 ( didnt have fpsvr), CPU Load was also lower than GPU load most of the time, and always no stutter at 45fps( = All frame times below 22ms). I run my i9 9900KS at 5.2Ghz, and your 9700 is same generation. Should be about 90% of my CPU single core performance or frame times = 10% higher than mine. It seems logic that you load your CPU by high settings? Do you run the Reverb @ 90 or 60hz? If 90hz, and motionreprojection you should need to stay below 22ms at all times, to stay free of stutter. Our hardware doesnt differ that much( I have a i9@5.2, RTX2080ti slightly OC, 32Gb RAM@3200mhz) so performance should be close? Du you run your CPU at 4.6-4.8Ghz on all cores ”all the time” or do you have BIOS settings letting the clock slow down after a while? My frametimes, shot this weekend: https://forums.eagle.ru/showpost.php?p=4492481&postcount=310