I was trying to figure out the correlation between Open XR rendering and MSFS rendering. First, is there a correlation? Second, what is the correlation? but thirdly, does it matter? I do notice with my current setup that when I raise rendering in Open XR from 70 to 90, I get a massive fps hit and turning your head takes like 10 seconds in the sim with those black edges the whole time turning. By comparison, raising the MSFS rendering from 70 to 90, while reducing fps, I get no where near the degradation as Open XR. I suspect the percentage values in Open XR are not linear but exponential (or near exponential).
@Ianrivsmith finally finished my little experiment. My son wanted me to watch the Heavy Lift launch and then we chatted for awhile.
However, while flying the MB339 at sunset along the Big Sur Coast (Bixby Creek Bridge) I had essentially the same response with OXR enabled and disabled. The only issue I was aware of was a slight sense of nausea with it disabled. However the frame rates were comparable (within 5 fps) and the clarity of the image was the same.
I will keep the Custom Render Scale checked and set @100.
@Beulah6126 I will try your suggestion. FYI, my rig is capable of overclocking but I have no idea if it is or not - I think it is supposed to do that automatically but I have not taken the time yet to learn about it. Don’t know what benefits overclocking might provide…
1 Like
@Beulah6126 Ok, the flight is recorded. Cessna 152 flying out of Heathrow at 9 am to downtown London landing at the City airport. Weather sunny and calm. FPS at Heathrow and approaching London: 25-28. At the Shard in downtown London: 18-20. The image was clear and sharp but somewhat jittery. No nausea. The jitteriness surprised me cause when I first flew over London after the upgrade, I don’t recall experiencing that effect.
Typically, I fly over countryside and small towns - Monterey, CA for example. I’m guessing the vast expanse of smooth ocean plays a large part in the computer not having to work nearly as hard.
I will say, that as a newcomer to MSFS2020 and this forum, I have been surprised by the problems many people have with their rigs working well with this sim. My experience has been relatively trouble free. I have tried to educate myself regarding how this black magic actually works so I try various things people suggest to make improvements. However, I tend to come back to the settings I described because they work well for me.
One other observation I will make (I have no idea if it has any pertinence or not) , I bought my rig from Alienware (a way to circumnavigate the scalper prices plus I have no experience building computers). Alienware made their name specializing in building gaming rigs (which is why I chose them). Could their experience in selecting and balancing components have a bearing on performance?
Please let me know what you learn from my test results.
2 Likes
I just bought my new rig from alienware (R11 Aurora) so anticipating, perhaps unrealistically, minimum issues. I’ve owned several cyberware products and have had good success with them as well. I just finished a flight from Bar Harbor Maine to Plattsburg AFB, NY on my 2080 Super laptop with rendering at 70 in both Open XR and MSFS in the TBM and had a great (non-ctd) flight. I run a super cooler underneath the laptop which blows cold air to the laptop from underneath and keeps CPU temp under 85C.
Copy thanks…that’s what most 3090 guys are saying.
Thanks for doing this test. I think the numbers are much more realistic. 
I am getting about 23-25fps out of London and about 20 near the Shard. I agree my favorite place to fly is the west coast void of skyscrapers and house during the sunset hours. It’s also much easier on the machine getting 35 to 40fps.
If you have not tried it, most of the high-end cards have about 10% headroom for performance by overclocking. Right now even with these top end cards, every frame counts. I see very little difference between 100 and 80 render scale, and lowering some settings such as volumetric clouds and ambient occlusion will add additional 20-30%. I find 35 fps an ideal minimum for a totally immersive experience in VR at the moment.
Try to lower these settings and see what kind of performance you will get. I’d also be interested in seeing the performance from the 3090 users. Although the sim will use whatever VRAM the GPU reserves, having higher VRAM will afford more textures and details without losing performance.
Could you do the same test that onaryox did with your 3090 set up? I’d love to see how much difference the extra vram will make pushing the limit of graphics settings.
Use his sim settings, using c152 out of Heathrow and see what sort of fps you will get heading towards the city center. This will give us a good idea if it’s worth the upgrade to 3090 for the VR experience. Thanks in advance.
1 Like
@Beulah6126 , I re-flew the ‘mission’: Heathrow to London City @ ~ 2000 ft, Cessna 152, 9 am in the morning, clear and calm. I set Volumetric Clouds = Medium (down from Ultra); and set Ambient Occlusion = Off (down from High).
Frame rates improved to 31-34 flying into London and 23-25 at the Shard. The quality of the imagery while inside the cockpit was worse than yesterday (regardless of my planes location) - more stuttery / choppy. 3 or 4 times the frame rate stalled for a second then resumed. The imagery quality outside the cockpit was substantially better than inside: sharper, way less stutter, and no stalled FPS. I don’t know how to think about the difference between inside and outside image quality; nor how to think about the first test compared to the second. I believe the second should have been better because of the benefit of turning the two variables down (Volumetric Clouds and Ambient Occlusion).
I hear a lot of discussion about Vsync. Under the Graphics Tab, In the PC mode I have Vsync disabled cause I have a Gsync monitor. However, I hadn’t considered whether this variable would have any impact on VR performance, too. Do you have any insight into this specific issue or if PC settings influence VR settings and performance?
Lastly, I don’t know if the following info would have any impact on the performance issue we are discussing, but I am wondering if it could. I have an i9 10900KF CPU. The ‘KF’ designation indicates that there is no built-in GPU on the chip. I chose this option over the ‘K’ series cause various websites I had searched indicated that not having the secondary GPU might provide better overclocking performance and less CPU heat generation. Could the lack of the secondary GPU reduce the processing demand on the CPU and thereby improve overall performance? I have read discussions about MSFS2020 not utilizing multi-threading resulting in high CPU processing demand. It would seem logical to me that any reduction in processing demand (secondary GPU) would be beneficial to those applications waiting in line for processing power…(MSFS2020).
Looking forward to DX12 due this summer
1 Like
Thanks for the update. You may want to try different drivers. I am hearing that certain NVIDA drivers cause some stutter. Your lack of iGPU should have no bearing on the sim performance. I have the 10850K and disabling the HT had no impact either. My 9900K had about 5fps gain simply by disabling the HT but that may have also come from the sim optimization.
Reverb G2 full resolution is equivalent to about 6K, so I am currently running mine at 80OXR/100TAA in sim. That really seems to be the sweetspot for performance/visual. Your stutters may be coming from the AI traffic/Online Bing feature. Unstable network connection/speed/latency also causes stutter so you may want to lower your traffic or disable online settings to see if that is the cause.
Choosing your render/process res like this forces disabling motion smoothing invariably. Those wanting motion smoothing though most likely can’t at OXR100. For this reason, it is not just fiddling with OXR Dev Tools, it is setting the tool to one’s expectations instead.
1 Like
To get motion smoothing working you gotta drop down to unacceptable levels of low settings, using high end hardware. And then it’s a glitch mess.
@Beulah6126 I kept all settings the same but changed my driver from 457.51 to 457.30. I flew the Cessna 152 on the standard route: Heathrow to London City @ 9 am, clear and calm.
Much smoother image quality inside the cockpit as well as outside. No stutter, no frame stalls. FPS ranged from 34 - 27 (Heathrow to the Shard).
I think I will keep this driver.
Next I am going to try Overclocking. The specs on my Alienware R11 indicate that the CPU runs from 3.7 GHz to 5.3 GHz. Is that the operating range accounting for various overclocking settings? The computer has an on-board app where I just make a choice.
I will fly the standard route and see what happens…
1 Like
If you enable the “Ultimate Performance Power” scheme like I’m describing in “My VR Settings” topics, it will appear in Alienware Command Center (ACC) in the list of power profiles. You can then assign this profile either globally or per-game (but maybe not to FS2020 which doesn’t show in ACC).
NB: SteamVR automatically sets the power profile to “high performance” when starting, and reverts the power profile to previous setting when exiting. There is no override and it can’t be disabled (not that I know of), which means with SteamVR, the “Ultimate Performance Profile” scheme might not be active anyway.
Although your 10900KF can boost to 5.3Ghz, you will benefit from having a sustained all core or single core for gaming situation. You should have no problem heating 5.0Ghz all core or even 5.1Ghz stable. If you won the lottery, you can push 5.2Ghz but at that point, you have to ask yourself if it’s worth the power consumption and the heat it will generate. What kind of cooler do you run? I can push my lowly 10850K to 5.1, I saw no perceptible gain and settled on a more reasonable 5.0Ghz. The temp never reaches 70c in hours of flight. Running the G2, even stock frequency will be sufficient and the real performance will come from your 3080. I’d avoid using any built in overclocking (motherboard or app based). Disable your Multicore Enhancement in the bios and set everything manually. It comes down to tow main settings. Your VCORE + Loadline Calibration. Once you find the stable VCORE and LC value, up your cache ring ratio. Then play with lowering VCCSA and VCCIO. One change at time, run a stress test, if pass, then +/- your values, test again. That is if you enjoy tweaking and perfecting your setup. Have fun.
I’m not certain you have this much low level bios settings in the Alienware, even the R11. They are wrapping this up under the ACC app. You do have some bios settings, but I’m really not sure about VCCSA and VCCIO. Interesting!
Ahh, Alienware! I have no experience in how much control you have over bios settings. If they have done the work creating OC profiles, that may work better. Most MB manufactures use insane VCORE and VCC voltages that serve as the lowest common denominator for even the worst binned chips. My Gigabyte board came with 1.37 VCORE and 1.32 VCC voltage that push the chip near 90c at stock. Does AW finetune their machine individually? That’ll make overclocking much simpler.
1 Like
I can’t say and I wish I knew! But they come with OC presets you can select directly in the bios and/or in the ACC app, in addition to lower level bios access. I’ll try to grab something to illustrate here.
1 Like
Wow it took me ages to find some photos!
Here is one typical of the R9 (and maybe R11 overclocking bios options):
https://www.dell.com/community/Alienware-Desktops/Aurora-R7-Problems-using-AWCC/td-p/7328496
Yeah, you are right you do compromise visual clarity and detail using reprojection. However I ‘only’ have a 2080Ti and I think even with a 3090 it’s difficult to get 90fps without reprojection. I need that 90fps for a good flying experience.
For a 3090 card I would consider using reprojection with OXR scaling @ 65% and to disable the runtime preview - you lose the third extra frame but it’s a little bit less glitchy, a bit more clear, and maybe a good compromise for some with a fast card like a 3090 - then wack up render scaling in the sim to 100% possibly even 120%.