True, but fdev stated we could use our current hardware for odyssey, and I’m not even including the FPS part of the game here. It seems reasonable to me to expect the same performance as horizons while in space at least.
Comparison of Frame Rates between VR and Flat Screen
- Thread starter CaptainPugwash
- Start date
This is the sort of info I was after, so thanks.As others have stated, rendering multiple view ports, from different perspectives, is more demanding than rendering one, even if the total pixel count is similar. Mirroring one eye to another screen is very low load, but doing the views of the two separate eyes in the first place is not.
CPU load of the render threads has almost nothing to do with number of pixels being drawn, total number of draw calls and/or the number of vertices/primitives/fragments is usually the overriding factor. Even the GPU side of things can be geometry limited, or be subject to severe cache contention, even if raw fill rate never becomes an issue.
Fast Stereo Rendering for VR
High performance stereo rendering for VR Timothy Wilson San Diego Virtual Reality Meetup January 20, 2015docs.google.com
Figuring out what approach the game is taking to VR rendering is the first step in deciding what to do to improve performance. This could be inferred, with some effort, from observing performance and resource utilization as various settings are manipulated. Ideally, you'd only be limited by draw calls and number of verticies, and could increase CPU speed and/or selectively reduce geometry load (turning down the view distance scale and terrain quality) to improve VR performance while sacrificing as little else as possible.
Also, looking at aggregate CPU load doesn't say much. If you want a clear picture of CPU load, per-core utilization, preferably with a short enough polling interval to catch relevant spikes, is required. Very near maximum GPU load is usually a strong contraindication to a CPU limitation, but there can be exceptions if GPU is not pegged at 99%+.
When I can I’ll post a trace of CPU and GPU load here, but I did check it, and the CPU was oscillatiing between 30 and 65%, and GPU didn’t go below 97% unless I limited the FPS. I still need to check sampling rate, but so far I don’t appear to be CPU limited.
As far as you know, who does the lions share of the rastering per eye workload? The oculus SW or the game client??
How much of that workload is deferred to the GPU normally?
thanks again.
I play in VR and the way to measure cpu/gpu performance is not by their % loads. The reason for this is that once reprojection kicks in, their loads likely drop anyway as they stop having to work quite so hard. The key factor is the time taken to render a frame, which needs to be sub 11ms to avoid reprojection/aws.
I’ll check the frame times as well, sure.
I was really just doing a straight dumb comparison between odyssey and horizons, and odyssey in particular seems to hit both CPU and GPU harder in VR compared to flat screen than horizons does.
What’s the reprojection about by the way?
Reprojection is where the VR software adds frames in when it can't maintain a certain frame rate: https://uploadvr.com/reprojection-explained/
It's not the frame rate you need to be looking at, it is the frame render time. If you are on a CV1, you can download the Oculus Tray Tool, which will give you performance overlays in VR. The Application Render Timing in the visual HUD options will show you frame render times. Whenever this goes above 11 or 12ms, reprojection will kick in, locking the headset at 45fps, which then reduces the load on the cpu/gpu.
Best way to understand it is if the headsets target fps is 90 and your cpu can do 45fps in the headset at 50% load, but can only manage 75fps at 100% load, the reprojection will drop the fps back to 45fps, so you will only see your cpu at 50% load as the reprojection is on, capping the frame rate. The only way to move beyond this position would be to upgrade the CPU/GPU combination so it can deliver a sub 12ms frame render time.
Oddyssey is so badly optimised in places, even the reprojection can't always save it and that is when the fps in the headset drops below 45fps and everything starts to get really ugly.
![[VR]Nieblaash](https://forums-cdn.zaonce.net/data/avatars/m/279/279677.jpg?1622050169){kind=avatar}
Something is really messed up in Odyssey in VR. No matter how low I set the resolution, even in the black with only stars around, the movement of the skybox when you tilt or rotate the ship fast is never smooth for me. I can always see the stars like a drag or vibrating, I dont know how to describe it. But it is not smooth. Meanwhile, Horizons have always run smoothly. So I don't think it is only a matter of optimization. There is something else broken with VR uin Odyssey but I am sure it is not high on their priorities to fix it...
Yep VR is definitely less performant in Odyssey
My CPU (3900x @~4.2Ghz) gets absolutely demolished by Odyssey on a Reverb G1. Quickly racks up 15-25ms frametimes doing anything other than sitting in empty space. My 6900xt is usually sub 10ms, but then once the CPU starts struggling it’ll start lagging out as well.
I can take the same settings and run Horizons smooth as glass, with only occasional drops to the mid 50s after two hours of hazres combat with another player + 2 fighters in the instance. I’m happy to get 60 fps outside my fleet carrier. I constantly break down to sub 35 fps with frametimes so high FPSVr just shows a solid orange line for the GPU and a red one for the CPU.
Very frustrating when horizons runs so well with no real visual difference between it and Odyssey besides the obvious lighting changes. Even DCS — which is notorious for being hard to run in VR — runs better than this. A shame, they hooked me in with promises not to ditch VR entirely, and even after a card upgrade I’m left running the game worse than I was on a 5700xt.
I can take the same settings and run Horizons smooth as glass, with only occasional drops to the mid 50s after two hours of hazres combat with another player + 2 fighters in the instance. I’m happy to get 60 fps outside my fleet carrier. I constantly break down to sub 35 fps with frametimes so high FPSVr just shows a solid orange line for the GPU and a red one for the CPU.
Very frustrating when horizons runs so well with no real visual difference between it and Odyssey besides the obvious lighting changes. Even DCS — which is notorious for being hard to run in VR — runs better than this. A shame, they hooked me in with promises not to ditch VR entirely, and even after a card upgrade I’m left running the game worse than I was on a 5700xt.
At least there are known frame rate issues that involve the CPU, so the performance fixes planned for the end of the year may (should?) make Odyssey feasible again, at least in space. By the way, have you done a trace of the CPU load, and if so did you see any spikes? My CPU didn't seem overstretched, but there may have been spikes in CPU load that I couldn't see with the sample rate of the tool I was using.
That'll be FSR causing the drag/vibration, I noticed it too.
Recommended spec is 8th gen 6 core processor.
Yours is only 4th gen quad core.
So CPU might be the bottleneck. Depends on how the engine uses multi core processors. I'm assuming Frontier recommended a 6 core processor for a reason.
Yours is only 4th gen quad core.
So CPU might be the bottleneck. Depends on how the engine uses multi core processors. I'm assuming Frontier recommended a 6 core processor for a reason.
If you bothered to read even the last 3 or 4 comments, you'd know there are people who have Zen2 CPUs (@Clumsywordsmith for example has a 3900x) who are still running into CPU issues. Unless Odyssey is packing some serious industry-changing future-tech, I'd imagine that at least a 3900x should be able to strongarm Odyssey just fine; the fact it doesn't says this isn't a simple case of "your hardware is potato."