VR - experimental support

User avatar
KingOfTheRoad77
Posts: 228
Joined: 07 May 2021 13:27
Contact:

Re: Oculus Rift - experimental support

#1961 Post by KingOfTheRoad77 » 18 May 2022 17:37

Krists wrote: 16 May 2022 12:20Kinda hard to justify 1000 EUR upgrade for mostly one game and still not being able to run at stable 90fps with antialiased game (ATS), and this is exluding the cost of the remaining parts :D
Exactly this.


I just did two further tests:
1. Activated "upscaling" in PiTool for the 8K-X, increased PiTool-resolution to 1.25 and tried playing a bit - but even if FPS increased significantly, I had to abort after a couple of minutes since everything just got so damn blurry I simply refused to play like this :D

2. Re-activated "native"; then decreased PiTool-resolution to 0.75 and increased SteamVR-SS to 130% which gave me 3860 x 3376 and the following results:
GPU: NVIDIA GeForce RTX 3080 (30.0.15.1179, Tavg 64.5, Tmax 69) CPU: AMD Ryzen 5 5600X 6-Core Processor (Tavg 54.4, Tmax 58)
Delivered fps: 67.82 Duration: 70.6min. Headset was active: 100%

GPU Frametimes:
Median: 13 ms
frametime <13.3ms(vsync): 58.9%

CPU frametime:
Median: 11.6 ms
frametime <13.3ms(vsync): 70.6%

Max. SteamVR SS: 130%
Render resolution per eye: 3860x3376(by SteamVR settings, Max.) (HMD driver recommended: 3386x2962)
Routes: Boise > Yakima; Yakima > Spokane

=> FPS still drop in cities to an avg. of 50 (so ~ +10 compared to before I'd estimate), but go up to 60+ as soon as leaving settlements and most of the time even 70-75!
Of course, the image gets a bit blurry, here too, but... okay... not as much as the first "PiTool upscaling" trial, and it might be a quite good compromise considering the much higher FPS (~ +17 in avg. !!) which make the whole driving feeling much smoother, even and especially in curves.

Just leaves the question open whether CPU frametimes now would benefit even more of the 5800X 3D... 🤔
I am a bit undecisive if I should give this a trial, too... especially as I've already claimed it for refund and should return it by Friday latest. I don't really think it would provide me stable 75 FPS in every situation, even with this setting now.

Krists wrote: 16 May 2022 12:20Nevertheless, I still love playing these games, if possible, I will probably buy next gen graphics cards when they come out, and if I would have space in my home for motion rig, that would be the next purchase :D
Yes, I absolutely agree with you: Both truck sim's are among my personal Top 5 of the best games I've every played - and I am playing games on PC for more than 30 years now ;)
And yes as well: Next Gen graphics card will be mine, too. I am already thinking about whether to give the new AMD 6950 XT a trial if I can get one cheap enough... 🤔 Question here, too, is: How much performance gain will next gen GPU's provide over 3080/6950 XT, and at which price?
User avatar
KingOfTheRoad77
Posts: 228
Joined: 07 May 2021 13:27
Contact:

Re: Oculus Rift - experimental support

#1962 Post by KingOfTheRoad77 » 19 May 2022 17:12

Okay - I wasn't willing to give up - and it seems as if it was worth it :D

=> I reinstalled the 5800X-3D today and furthermore reset PiTool to 1.0 + SS to 100%, but installed fholger's openvr_fsr once again at default settings, and then played for an hour:
GPU: NVIDIA GeForce RTX 3080 (30.0.15.1179, Tavg 65.7, Tmax 70) CPU: AMD Ryzen 7 5800X3D 8-Core Processor (Tavg 69.1, Tmax 79)
Delivered fps: 71.0 Duration: 64min. Headset was active: 100%

GPU Frametimes:
Median: 13.0 ms
frametime <13.3ms(vsync): 70.9%

CPU frametime:
Median: 10.8 ms
frametime <13.3ms(vsync): 86.5%

Max. SteamVR SS: 100%
Render resolution per eye: 4512x3948(by SteamVR settings, Max.) (HMD driver recommended: 4514x3948)
🥳🥳

"Unfortunately", it started to rain on my last trip from Coeur d'Alene to Colville, so FPS significantly dropped from that moment onwards until a couple of miles before my destination. Nevertheless, I've never been so close to stable 75 FPS ever 🥳😎
Image quality is once again a little better than with my former trials adapting PiTool resolution and SteamVR-SS, especially as fholger's mod sharpens the image's center quite a lot compared to simply lowering the overall resolution with the other methods. Of course, it is not the same as with "100% native" without any mod - but I think, this is really very close to the best I can get right now and until NextGen GPU's will come out.

On the other side, remains the question:
How much comes from the 5800X-3D...
... and how much comes from fholger's openvr_fsr mod ? 😜😂

So, I think I'll have to do another test without fholger, but lowered resolution once again (even if it looks worse 😩), and perhaps also reinstall my 5600X in a couple of days one last time and see, which values it reaches with fholger.
This way, I won't be able to hold the latest return date for the 5800X-3D, unfortunately - but well, my fault... could have let it installed longer and done those tests earlier... As the supply of it is still short, I still might be able to sell it via ebay for 500€ if I'd go back the 5600X. Part of the hobby ;)
Krists
Posts: 35
Joined: 30 Nov 2021 02:14

Re: Oculus Rift - experimental support

#1963 Post by Krists » 20 May 2022 09:23

Good result!

What is pitool and how does it affect graphics and performance?
Can you also do some ETS2 tests and perhaps take some through the lens photo/video?

And why would you reduce graphics when comparing fpsvr effects?
User avatar
KingOfTheRoad77
Posts: 228
Joined: 07 May 2021 13:27
Contact:

Re: Oculus Rift - experimental support

#1964 Post by KingOfTheRoad77 » 20 May 2022 16:28

Krists wrote: 20 May 2022 09:23What is pitool and how does it affect graphics and performance?
PiTool is Pimax' driver tool that needs to be launched and run in background in order to start and use the headsets. It has a couple of options that can be set, i.e.
  • FOV (large (~170-180°), normal (~140-150°), small (~110-120°), very small (90-100°));
  • resolution/SS in 0.25 steps (1.0 = default, 100%);
  • "native" vs. "upscaled";
  • Hz the headset is running with (native => 60/75/90; upscaling => 110);
  • "Smart Smooting" (= Asynchronous Spacewarp/Motion Smoothing);
  • Parallel Projection (needed for many SteamVR games (i.e. Project Cars, Truck Sim's, Elite Dangerous, and others) due to the canted displays; may cost quite some performance - but some games can also be played in Oculus-clone-mode - unfortunately that doesn't help at all with the truck sim's and Elite Dangerous in terms of gaining FPS...);
  • colour tone (blue/green/red) & contrast;
  • and a couple of other little settings and tweaks.
Obviously, FOV and resolution setting/SS have the major impact on performance and graphics.
Krists wrote: 20 May 2022 09:23Can you also do some ETS2 tests and perhaps take some through the lens photo/video?
I will also do some ETS2 tests of course. But I have no idea how to take through-the-lens photos or a video while driving... and I just have my smartphone camera, no other camera or equipment 🤔
Krists wrote: 20 May 2022 09:23And why would you reduce graphics when comparing fpsvr effects?
Because I am impatient 😂

=> My main goal still and simply is to achieve close to 75 FPS / 70+. Which apparently I can only achieve with my hardware if I reduce the graphics in any way - and of course the way which costs the less of it ;) Or better said: Which I see and notice the less.
All former trials in the past years weren't really successful although I tried a lot.
But due to my impatience, none of my trials were structured enough; I already got frustrated after half an hour and then continued playing as before for a couple of weeks until I found the mood to try something else.

So, yesterday, after having installed the 5800X-3D, I - unfortunately - remembered fholgers openvr_fsr and decided - well - to try both at the same time: 5800X-3D and openvr_fsr... ... ... once again not analyzed well enough before ;) ...

... but with the best result so far! 😁

___

I swapped back the 5600X earlier this afternoon and did another session with openvr_fsr in order to get a feeling of 5800X-3D's influence on yesterday's result.
So everything as before, just the 5600X instead of the 5800X-3D
GPU: NVIDIA GeForce RTX 3080 (30.0.15.1179, Tavg 62.8, Tmax 68) CPU: AMD Ryzen 5 5600X 6-Core Processor (Tavg 52.2, Tmax 56)
Delivered fps: 65.8 (instead of 71.0 with 5800X-3D) Duration: 69.5min. Headset was active: 100%

GPU Frametimes:
Median: 13.4 ms
frametime <13.3ms(vsync): 49.3% (instead of 70.9% with 5800X-3D)

CPU frametime:
Median: 12.4 ms
frametime <13.3ms(vsync): 58.6% ((instead of 86.5% with 5800X-3D)

Max. SteamVR SS: 100%
Render resolution per eye: 4512x3948(by SteamVR settings, Max.) (HMD driver recommended: 4514x3948)
Area was the same than before, just in a different order depending on the contracts I got:
1. Colville > Coeur d'Alene (nighttime)
2. West of Coeur d'Alene to East of CdA
3. Coeur d'Alende > Sandpoint (with some rain en route)
4. Sandpoint > Spokane

So, it seems - with openvr_fsr reducing the load on the GPU significantly - , the 5600X really becomes the bottleneck and/or it puts a brake on the 3080 => in the end it decreases FPS by appr -5 = -7% (based on 71).
Or the other way round: The 5800X-3D might improve FPS by about +7,5% and improves frametimes both on itself and the GPU, too.

=> Two questions:
1. How realiable is this result? Although I once again drove for an hour in the same area, this game has a lot of influencing factors that change a bit from session to session. It probably would take at least 5 "statistical" sessions with both CPU's in order to confirm this first estimation.
2. And if ever the 5800X-3D would provide about +7,5% FPS in this setting = 5 FPS => are 5 FPS worth 500 € ? Wouldn't it be better, perhaps, to wait for RDNA 3 or Ada and pay 1000 €, but for 15 or 20 FPS...? Unless, the new GPU's can't be fully unleashed without a better CPU... ;)

Still quite undecisive 🤔
Krists
Posts: 35
Joined: 30 Nov 2021 02:14

Re: Oculus Rift - experimental support

#1965 Post by Krists » 21 May 2022 10:21

I did “through the lense” photo with phone directly capturing the picture of the lense. Not possible while driving, but possible while parked.

Regarding performance of 5800x3D - in my case I am keeping mine as even though 5fps is not much, I already went through this effort or buying and swapping and testing, and psychologically I dont like the idea of “downgrading” :D

Besides that, it compliments 3080ti more than 3080.
Kuski79
Posts: 7
Joined: 18 Mar 2020 10:46

Re: Oculus Rift - experimental support

#1966 Post by Kuski79 » 21 May 2022 17:51

I do enjoy watching this thread and all the testing you guys are doing. This game relies on single core performance and I think that needs to change more than anything. I think your best bet is to overclock your CPU as high as it can go for the single core (if that's possible). At least for the time, before SCS can optimize this game for multicore performance. I'm running the game on Linux (with Proton) on the Valve Index. In the newer countries like Austria the game is stuttering like crazy in the city area, but on the older parts of the map and on the roads outside the city it runs fine at 90Hz and 150% supersampling on SteamVR.

Even without the VR headset, playing on the monitor, I would see framerates drop to like 64fps with stuttering on city areas, because the single CPU core can't feed enough data to the GPU and I see the GPU usage dropping because of that. I'm running an AMD 5600X CPU with AMD Radeon 6800XT. The game's engine needs an overhaul, but I do understand that a small company can't do miracles, while they need to make money to keep them profitable.
User avatar
KingOfTheRoad77
Posts: 228
Joined: 07 May 2021 13:27
Contact:

Re: Oculus Rift - experimental support

#1967 Post by KingOfTheRoad77 » 21 May 2022 18:56

I've done further trips, both in ATS as well as some in ETS2:

In ATS, I can confirm now almost sure, that I'll get +/-65 FPS in average with my 5600X, RTX 3080 Gaming OC, 32 GB RAM and fholgers openvr_fsr mod. More in detail, FPS drop in larger settlements and towns down to 60 at least, and if I stay for longer there at one place, i.e. deliveries within one and the same city, FPS sometimes could drop down to 50-55. But that's the minimum - and far better than anything I experienced before in the past years (30-35 at worse; a couple of times even <30).
On the road and highways, FPS mainly stay at 70+ with many many parts @75 FPS stable 🥳 - and I finally even see things and details in the landscape around me I wasn't able to really notice before due to the stuttering 😁

In total, I'd say a benefit of +10 to +15 FPS just by using fholgers mod.
Yes - image quality is a little worse than with everything 100% native - but really very very little, so that even I myself - high-FOV-image-quality-sharpness fetishist - can take the deal here ;)

Concerning ETS2, things are a little bit difficult right now:
Since the latest little patch, I am experiencing random crashes-to-desktop, probably due to a mod. But I was not able yet to determine which one is causing it...

Nevertheless, I had one trip of about an hour before it crashed once again:
GPU: NVIDIA GeForce RTX 3080 (30.0.15.1179, Tavg 63.5, Tmax 70) CPU: AMD Ryzen 5 5600X 6-Core Processor (Tavg 51.0, Tmax 56)
Delivered fps: 66.43 Duration: 58.5min. Headset was active: 100%

GPU Frametimes:
Median: 13.3 ms
frametime <13.3ms(vsync): 51.8%

CPU frametime:
Median: 11.5 ms
frametime <13.3ms(vsync): 67.4%

Max. SteamVR SS: 100%
Render resolution per eye: 4512x3948(by SteamVR settings, Max.) (HMD driver recommended: 4514x3948)
This result seems to be not so good as expected.
But: I was driving from Lago di Garda to Klagenfurth and can confirm what @Kuski79 wrote:
Almost stable 72-75 FPS in Italy / significant drops as soon as I came to the revamped Austria areas.
In an attempt later, I was able to confirm this once again, driving from Klagenfurth to Bologna: ~55-65 FPS only in Austria, and back at 72-75 FPS in Italy. (And another ctd close to my destination.)
(Besides larger settlements always, of course.)

Krists wrote: 21 May 2022 10:21 I did “through the lense” photo with phone directly capturing the picture of the lense. Not possible while driving, but possible while parked.
Okay, I will try what I can do perhaps tomorrow. For the moment, I'm a bit unpatient to fix those ETS2-crashes first.
Krists wrote: 21 May 2022 10:21 Regarding performance of 5800x3D - in my case I am keeping mine as even though 5fps is not much, I already went through this effort or buying and swapping and testing, and psychologically I dont like the idea of “downgrading” :D
Yes, I can understand that 😂 Absolutely 🤣

I've brought mine to the post office today for return - at the eleventh hour, hopefully not tooooo late 😟

I'll first see and probably invest in a 4080 or an AMD if it is competitive enough. Perhaps, my 5600X can fuel it already sufficiently and I'm fine - or I'll do another CPU upgrade testing then, when they'll got a little cheaper hopefully.

(Besides: By using fholgers openvr_fsr, both truck sim's now have a maximum power draw of <300W !! Before, there have always been power spikes up to 320/330/even 350 Watt sometimes, although my 3080 is UV to 0.893V! Now: Up to 50 Watt less, and temperatures stay 3-5 degrees cooler as well 🤗 Could be a nice tweak even for a new 4080, perhaps not with default settings, a little higher ones for better image quality, just enough, to avoid longer high load peaks with higher temp's.)
User avatar
KingOfTheRoad77
Posts: 228
Joined: 07 May 2021 13:27
Contact:

Re: Oculus Rift - experimental support

#1968 Post by KingOfTheRoad77 » 21 May 2022 21:39

I hopefully got my crashes in ETS2 fixed and was able to drive from Venezia to Bratislava passing Klagenfurth and Vienna:
GPU: NVIDIA GeForce RTX 3080 (30.0.15.1179, Tavg 64.3, Tmax 69) CPU: AMD Ryzen 5 5600X 6-Core Processor (Tavg 51.0, Tmax 54)
Delivered fps: 69.45 Duration: 37.5min. Headset was active: 100%

GPU Frametimes:
Median: 12.9 ms
frametime <13.3ms(vsync): 65.5%

CPU frametime:
Median: 10.3 ms
frametime <13.3ms(vsync): 78.2%

Max. SteamVR SS: 100%
Render resolution per eye: 4512x3948(by SteamVR settings, Max.) (HMD driver recommended: 4514x3948)
Italy and Slovakia 73-75 FPS most of the time.
But even Austria surprised me: this time, ~60-65 FPS only behind the border, then again shortly when by-passing Klagenfurth, and last around Vienna. Else, on the autobahn, 70-75 FPS here as well.

I think, it has a large impact on how long you "stay" in one sector:
First time today, when going from Gardasee to Klagenfurth, I first had to take a rest by taking an exit from the autobahn, then I decided to explore the rest area "Wörthersee", which also took quite a couple of minutes driving around there. Enough time for the game to fill the sector with traffic.
Now, I just drove by everything, no stops except one on a rest area just before Vienna. So, not much time for the KI to spawn enough traffic for the FPS to drop significantly, I think.

On the whole, I am quite happy with the result :)
Just wished a bit I had discovered the use of fholgers openvr_fsr a bit earlier. But I was too much fixated on getting FFR work as I thought this would provide better image quality - but well: I can't get to work what doesn't want to work, neh? ;)
Just glad, that openvr_fsr finally is much better than I thought it would be :)
User avatar
KingOfTheRoad77
Posts: 228
Joined: 07 May 2021 13:27
Contact:

Re: Oculus Rift - experimental support

#1969 Post by KingOfTheRoad77 » 23 May 2022 16:38

So, I did a couple of further trips in ETS2 with openvr_fsr. To make it short: Here are the results:

Nurnberg to Milano
GPU: NVIDIA GeForce RTX 3080 (30.0.15.1179, Tavg 64.2, Tmax 70) CPU: AMD Ryzen 5 5600X 6-Core Processor (Tavg 51.1, Tmax 61)
Delivered fps: 66.51 Duration: 65.9min. Headset was active: 100%

GPU Frametimes:
Median: 13.2 ms
frametime <13.3ms(vsync): 53.8%

CPU frametime:
Median: 11.5 ms
frametime <13.3ms(vsync): 68.8%

Max. SteamVR SS: 100%
Render resolution per eye: 4512x3948(by SteamVR settings, Max.) (HMD driver recommended: 4514x3948)
Milano to Innsbruck
GPU: NVIDIA GeForce RTX 3080 (30.0.15.1179, Tavg 64.6, Tmax 70) CPU: AMD Ryzen 5 5600X 6-Core Processor (Tavg 55.0, Tmax 59)
Delivered fps: 67.49 Duration: 47.9min. Headset was active: 100%

GPU Frametimes:
Median: 12.8 ms
frametime <13.3ms(vsync): 64.1%

CPU frametime:
Median: 10.1 ms
frametime <13.3ms(vsync): 79.3%

Max. SteamVR SS: 100%
Render resolution per eye: 4512x3948(by SteamVR settings, Max.) (HMD driver recommended: 4514x3948)
So, could still be a little bit better - but has been far worse in the past before use of openvr_fsr.


Furthermore, @Krists , here are the requested photos. Unfortunately, taking the through-the-lens wasn't easy with the phone - I wish I've had a third hand ;)
[ external image ]

[ external image ]
I don't know whether you can see anything of interest there... :/

So I made four more ingame-screenshots with openvr_fsr vr-screenshot-shortcut - perhaps they are little more helpful:
[ external image ]

[ external image ]

[ external image ]

[ external image ]

I think it's not easy to judge about image quality based on these photos and screenshots. Of course, there is a noticeable loss compared to 100% native. But even for me, it is not as significant as I expected with those settings (openvr_fsr at "ultra quality") and almost negligible with regard to that massive FPS gain.
But of course, I still would prefer 100% native if it was possible with those FPS and frametimes ;)
CU
Krists
Posts: 35
Joined: 30 Nov 2021 02:14

Re: Oculus Rift - experimental support

#1970 Post by Krists » 23 May 2022 20:52

Thank you so much! Good results, I also used fpsvr in my tests. Your ETS2 results seem very good and even better than mine, so I think there is something wrong with my ETS2. Do you feel any difference between aliasing of ATS and ETS2?

Regarding through the lense - for G2 it was also a bitch to take, but I somehow managed. The screenshots though have very weird looking FOV and I think are too clear on the out of ceter image.
Post Reply

Return to “General discussion about the game”

Who is online

Users browsing this forum: CirclesArePointless and 5 guests