Any further update on your settings and results with this @BuggyMasters?
Also, as you have a similar system to mine, except you have the 4080 and I have the 3080 and I am considering such an upgrate, I was wondering if you could load in a complex aircraft (eg. PMDG 737) at parking spot Gate E 10 at KLAX, live weather, MP and AI traffic, high preset, set TLOD to 100 and TAA 100% and OXR at 80% and tell me roughly what FPS you are getting in VR looking forward in the cockpit. I just want to know if your CPU is still holding you back or if the GPU upgrade improves such a scenario regardless. Thanks.
I’m getting 30fps on TLOD @ 220 in the 737 at KLAX live weather with ai traffic. That’s with most settings on ultra. I suspect if I dropped it to 100 on TLOD I’d get 35. I’d say over a 3080 you’d be pretty happy.
With TLOD @ 220 and all settings Ultra I get 23 FPS. VRAM usage is showing 110%, which explains why it’s very stuttery if I look around at those settings as the GPU is likely paging from system RAM. Interestingly, reducing TLOD to 100 didn’t improve my FPS, again likely because of overallocated VRAM. I’d likely benefit from a GPU upgrade for increased VRAM capacity alone.
Thanks for the comparative data, which has been helpful in my own deliberations.
For increased levels of detail the gpu memory is crucial. That is why I want to wait and see what AMD will be offering. DLSS3 frame generation does not work for VR, that would have been a reason for me to upgrade to a 4080.
Since that is off the table, next in line is memory and AMD seems to be offering a lot of that.
I am currently running a 3080 and my case will not accomodate a 4090 due to power constraints.
I am also interested in how the new AMD cards will go. I don’t care for the blur that DLSS produces no matter how much you over render first, so raw GPU power and large memory is what I really want. Only a few weeks to go!
I think you would see that DLSS3 + frame generation ( especially that ) is something different than what we are used to.
The problem in VR is latency, frame generation delays your scene output with two frames. In other words, what you are seeing is the state two frames ago. That creates a weird sensation in VR.
win10, hp reverb g1, 13700k, 5600T ddr5, 4080, openxr 100% Samscene3d Tokyo all ultra highest , ultra 400 lod.
DLSS 2 balanced, mr depth+ motion 27-29 fps. Dlss 3 of course is not working on VR. I set the cpu affinity only to the Pcores which reduces cpu bound lag
Thanks for sharing your results. Can you share how you setup CPU Affinity to Pcores?
I have been testing the waters with OXR > 100% and interesting results. Testing higher OXR vs higher LOD in non MR mode.
Might have to try high Lod and 100 oxr.
My current setup:
Win 10, 12700K, 32 GB, G2, Oxr 200%, MR off, Dlss (Qual/bal/Perf) HAGS ON, Resizable Bar On, HT enabled, DX12
Nvidia setting: low latency- ultra, power- Pref Max performance, anisotropic Off, Negative lod Clamp, Threaded Opti on,
MSFS: Most Ultra, few high and medium LOD 125.
I am getting 30 - 43 smooth over NYC. DLLS Performance in mid 40s, quality brings it down to mid 30s.
Openxrtoolkit turbo mode game me me CTD. If I have it enabled and switch to VR, screen freezes for like 5 seconds then CTD. Both in DX12 and DX11. I have tried the latest version 1.2.2, went back to 1.2.1 still freeze CTD. whats weird is I had toolkit running couple weeks ago with DX12 and turbo on with mind blowing performance. Not sure which update/tweak broke it. I even stopped the hollowgraphic shell.
update: ran toolkit in safe mode, reset the setting and was able to run it. as soon as I enabled Turbo on, freeze and CTD. hopefully there will be an update soon from Matt, I enjoyed turbo.