Supmua

https://videocardz.com/newz/zotac-rtx-4090s-pcie-gen5-power-cable-adapter-service-life-is-up-to-30-connection-cycles ^This was the concern. Now we don’t know if this also applies to other AIBs or this is just a precautionary recommendation from Zotac. In any case, I already have a new ATX 3.0 PSU coming from Newegg so I don’t have to deal with this potential issue.

The USD pricings seem much more reasonable than elsewhere. Of course there’s also a possibility of US resellers readjusting (i.e. jacking up) the prices above MSRP, they definitely did that with the last gen cards.

You might be right. 2-3ms improvement in single core performance per year in CPU frametime isn’t cutting it when you need at least 10-15 to jump from 45 to 90 fps tier in VR with the current DCS engine, and this is just average not counting 5th or 1 percentile numbers which also contribute to smoothness.

I believe flightsimmers (not necessarily DCS) will gobble up these new GPUs and a lot of them use VR so we’ll definitely see some VR benchmarks soon. I’m also planning to do some DCS benchmarks on my own as I’m really hoping to see 90 fps in some scenarios (after years of dreaming). Would’ve been nice if we have CPUs with equally large performance leap, maybe the 13900KS or 7950X3D will come close.

I don’t have to uninstall it disable Reshade when using OpenXR, but I’m using Varjo Aero so behavior with other headsets may not apply. I’ve been pretty much flipping back and forth between Steam OpenVR and OpenXR modes by just swapping the openvr_api.dll file. Reshade only works in OpenVR though.

First impression on the new Varjo Base update 3.7.0.5. Improved chromatic aberration but still noticeable on texts. Sharpness is good I have no complaint, same goes for geometric distortion which is minimal. The 60% default brightness is neat but too dim for me, I like it at 70% and will stick with this for DCS. The Workspace gamma level is still off perhaps slightly better but you can still see significant black crush and white out on black and white level charts especially when compared against other desktop apps such as Steam overlay or Virtual Desktop. Overall I think it's a good update and would recommend it over older versions. Also, motion prediction is coming in 3.8 beta according to discord announcement, so that's something to look forward to. Some interesting observations on SteamVR vs OpenXR mode. With the latest DCS beta, NVIDIA driver, OXRTK, etc. I see 2ms frametime reduction with OpenXR, at the expense of 1-1.5ms increase in CPU frametime (Harrier Caucasus Freeflight). Have to do more testings over time to check reproducibility. Necksafer XR beta4 is still not working for me, beta2 is the last working version for my setup. Edit: In more taxing scenario (Apache Syria Hotramp) where I tuned my settings so CPU and GPU FT stay below 21ms on ramp, GPU FT advantage reduced to 1.0-1.5ms while CPU FT was exactly the same between OpenXR (with OXRTK) vs SteamVR (with Reshade 5 and fpsVR). This is roughly similar to past findings with older software.

The size, lol. This the the Asus ROG Strix which is bigger than average cards. For me size isn't that important as long as cooling is sufficient. The only issue I have with these cards though is that they still feature DP 1.4a rather than 2.0 ports which are already present in the new Intel GPU cards. This would've been great for next gen VR headsets as they would allow extra bandwidth support, and this means that high-res headsets such has Varjo may not need to rely on NVIDIA's DSC tech to operate. Of course the headsets themselves have to support the new protocol also.

Been watching unboxing videos and these cards are truly monstrous in size. My 3090 is already huge next to the 2080Ti and these 4090 cards easily puts it to shame. The Asus ROG Strix one in particular is larger than other cards. Too bad no official performance or benchmark reports allowed until Oct 11th. Also, a very cool video on the RTX 4090FE from Gamers Nexus with an NVIDIA's thermal engineer showing the design process, cooling tech used, etc.

I did some simple calculations, and if the new GPU yields 36% performance uplift in frametime then it will get me to 90 fps in Harrier Caucasus freeflight mission without me touching any current settings other than GPU swap. The expected performance uplift from the 4090 is supposedly even twice of that so 90 fps is going to be doable at least in some simple missions. My current CPU frametime on this mission is ~9ms and GPU FT ~15ms, and you can see that even with the 9900K CPU which is going to be 4 generations behind soon this CPU is not a bottleneck at all in this mission. In general, my GPU FT is almost always higher than the CPU counterpart even in complex maps such as Mariana and this is why I've been holding off on upgrading CPU for my DCS rig even though I also have an 11th gen i7 in another PC. This might change soon with the RTX 4090 as the CPU will likely become the bottleneck instead.

These cards including 3rd party’s are already listed on Newegg, Bestbuy, Bhphotovideo, etc. You just can’t preorder them until Oct 12th. I have no idea if AIBs will be ready to ship right away. If the leaked data is true then GPU will likely no longer be a bottleneck. Which means we could see up to 90 fps in VR in some light flying sessions (if CPU’s frametime stays below 11ms), 45-60 in most situations except when CPU is truly taxed (supercarrier, highly populated areas, etc.). So yeah I’ll take the 78% performance gain over 35% from my last upgrade.

According to recently leaked Geekbench data, the 4090’s rasterization performance (CUDA cores only) is 60% over the 3090Ti and whopping 78% over 3090. That is quite a significant jump, considering the rasterization gain from upgrading my 2080Ti to 3090 is only around 35%. Now add the Tensor cores to the equation in supported games and the performance gap widens further. This is a better performance upgrade compared to previous gens. AMD’s upcoming GPUs will likely offer similar rasterization performance as well.

To each his own but for the last few GPU generations, the upgrades have been worthwhile for me allowing either extra frame rate, higher rendering res (which also means minimizing shimmering), or stability in MP as well as congested areas. There are some situations in my play that are also CPU bound (though not a whole lot, and nothing that would drag the overall fps below 45), so I’m planning to upgrade the CPU also. This isn’t just for DCS of course as NVIDIA’s DLSS 2 mode in the other flightsim yields significant performance uplift in VR, and once the new DLSS 3 is VR compatible it will be even better. There’s no more EVGA card now Don, they’ve stopped making GPU.

Chromostereopsis - 3 Layers of depths "Rivers, Valleys & Mountains" - Chromostereopsis - Wikipedia When I view this (link) on my Aero, there are varying depths that I can see. Blue being the most recessed and red appears popped out, while green is somewhat in-between. And this is a partial explanation as to why not everybody will have the exact same experience when it comes to depth perception.

There’s no clear concensus. I suspect this may be related to either driver issue or chromostereopsis. If it is the latter then not everybody will see the effect. First check out some chromostereopsis pictures (Wikipedia etc.) while wearing your VR headset and see if you can experience varying 3D depths. If you can’t see varying depths then I wouldn’t bother.

VB software version 3.6.2.19 is out. First impression is pretty good. The sharpness is similar to 3.5.1 and much better than 3.6.1. I see moderate amount of chromatic aberration on white texts/black background, not very noticeable in games but very apparent on web browser. I haven't played in a while so my brain isn't adjusted to the amount of CA this headset has so it might be more noticeable than usual. It's relatively distortion free compared to 3.5.1. So this version might be a keeper. Apache cockpit looks great. Tested with OpenXR mode only with brightness setting at 75%, will try OpenVR mode at a later date.

This happens with flat screen also. It's the pixel size that helps with spotting in DCS. I still prefer to play in higher res, though.

As far as I know, SteamVR motion smoothing works by taking into account of 3 things: headset rotation, headset position, and other object positions (it’s also not depth aware unless they added that in recently). When this option is off and fps is consistently below 90 it goes into old reprojection mode might be called interleaved or something which has no motion/position support to make the game playable without inducing nausea or eye strain (missing or dropped frames don’t always occur at consistent intervals). Not sure how OpenXR works but apparently better, with depth support as well but I don’t think DCS has depth output.

BTW, MSFS depth reprojection option helps eliminate cockpit tearing during horizontal movement, and also proves that motion smoothing can also be done from the game setting itself although probably not as good as headset-specific setting provided by HMD company.

I use the latest version. IMO 0.5ms frametime difference isn’t much if your initial fps is not high, 1-2 fps at most (1 if your fps is closer to 45) and even that is still within margin of error.

No motion smoothing/compensation stuff but it’s still gotta fill in the missing frames to hit 90 in one sec. Yeah try SP first to make sure it’s not the problem with your setup.

You should be able to get better framerate than mine with that specs. Make sure to minimize or get rid of unneeded background processes (LED light control, etc.), and optimize your windows settings for gaming. Avoid going crazy on overclocking, you want stability with sustained performance rather than peak performance.

Try to remove all the mods you might have and run repair DCS. Then drag the openvr_api.dll in the attached zip from the very first post into the \bin folder, if that still doesn't work drag the other 2 files in as well. I switched to OpenXR this weekend to do some comparison, no issue. The GPU frametime number is actually the exact same as what I got from OpenVR mode (using OXRTK and fpsVR, respectively), the only difference is that OpenXR feels smoother. This was Apache hot start Syria low flying, so it's fairly taxing to the system and I was able to maintain >45 fps most of the time in both modes. I think OpenXR inherently has better reprojection algorithm vs OpenVR in general (especially seen with MSFS where low fps still results in minimal stuttering). However, after doing fps lock in OpenVR it's pretty much the same smoothness (not 90 fps smooth, but 45 fps smooth). For gameplay that expects <45 fps though, OpenXR may be a better way to play.

Haven’t switched recently so no idea. In the past it’s been around 1ms frametime which isn’t much for me considering the cons of not being to use Steam overlay, Reshade, desktop in windows, fpsVR, etc. My StemVR performance is quite good right now, free flight Caucasus with the Harrier I get up to 60-68 fps depending on the area on the map (and I just need 45). We don’t have the exact same setup and tuning so you may or may not see similar results.

Did anyone notice performance improvement after the last update? Apache Caucasus runway start, my fps is mid 50s on the runway. Even Syria hotstart the frametime is 21ms on the runway which mean fps 47-48, it's a noticeable improvement albeit a minor one--it dips below 45 during hovering takeoff but pretty much 50s once I'm fully up in the air. I never had fps above 45 on the runway in Syria in the past. Haven't changed any settings in a long time, this is Steam OpenVR btw.

As of now DCS uses quad mode if you run it straight from the .exe as stated in my very 1st post. Very easy to check by either take off the headset and look at your PC monitor or bring up Steam overlay and check the VR mirror that way. You can forget about foveation for now as the only app that truly supports this is the Varjo Workspace. Remember that DCS software itself is quite old and most people also don’t play in VR, so don’t expect any cutting edge VR tech unless it’s hacked in.