It's not like VR requires a lot more cpu power, the vr task takes about 1-3% according to Taskmanager with a oculus rift cv1.
Additionally, the render pipeline is a little bit different afaik so you lose some fps from the game too.
Overall my experience is that it's about 5% fps loss.
The bigger issue is that you need to have consistent fps! On a monitor you are either not that sensitive and don't use vsync or you are very sensitive and use vsync or gsync/freesync.
But a few little stutters here and there aren't a big issue for most people. Sure, when it has a hiccup at your braking point it's not great but in VR, a stutter freezes your brain, lol.
This is the biggest difference though. On a monitor you are flexible or only need 60 fps... In VR you need consistent 90 fps for most headsets. That's 30% higher demand. Due to sim racing titles only using 2-4 cpu threads, getting these 30% more performance out of 2-4 cores is difficult when we had only 0-10% single thread performance gain from generation to generation...
About vr:
I only had a oculus rift for a few weeks. The ASW is very well done. When you drop below 90 fps, you'll drop to 45 fps and the headset software will render an additional frame to keep it at "90 fps".
You get some latency when moving your head though...
I was fine with constant ASW or without it but when hovering between 80-90 fps and switching between native 90 fps and ASW 90 fps, I got motion sick...
With most monitors you can run 60-80 fps and be fine. With vr, it's 90 fps or motion sickness if you're sensitive...
With different headsets, the "asw" is on different levels too. So with some headsets you really need these native fps.
With medium settings and 15 cars visible in acc, my fps dropped below 90 for race starts, pit stops etc.
---------
Now some facts about multi thread usage:
With process explorer (it's a little Programm to download), you can look into the properties of each exe running and you can see the threads and their cpu usage.
With my 12 cpu threads (6+ht), the maximum that one thread can use is 8.33%. 100/12
A cpu core can either do stuff or idle. That is for each clock cycle.
So when you see a nice 20% load on all cores in taskmanager but an application is running on 1 thread, it's actually:
Cycle 1: core 1 100%, core 2-6 caching
Cycle 2: core 2 100%, core 1+3-6 caching
And so on. At 5 ghz and a tick rate of 1s in taskmanager, you'll see 5.000.000.000 cycles averaged...
With process explorer however you can see how many threads are using how much cpu.
For ac it looks like this:
Thread 1 = 8.25%
Thread 2 = 8.12%
Thread 3 = 4.68%
And then a lot of threads using below 2%.
Sadly it's not scaling directly.. Rocket league for example looks similar at 40 fps (limited) but I can actually get 200 fps. I don't understand this but you can't see the limits of the cpu via process explorer, you can only see how many cores you need and if your cpu could actually be maxed out.
Anyway:
So the first thread will be limiting if you gpu is fast enough and you need enough cpu cores to always have one core doing that big thread.
While you also want a second core free to not make the second big thread limiting.
For ac with the loads above you want a 4 core cpu with 2 cores boosting as high as you can.
And the rest of the cores doing the other little threads.
Windows will automatically shuffle around all loads to have the big thread on the highest boosting core.
Due to managing this and some latency, overclocking all cores as high as possible gets more consistent and higher fps than boosting 2 cores 200 mhz higher and going up and down all the time.
At about 400 mhz single boost vs all core oc, it becomes the same, performance wise.
But I've never seen a cpu that could boost 1 or 2 cores 400 mhz higher than what's possible with all core oc...
I tested this with my 10600k and the result was either all cores 4.9 ghz or this:
5.1
5.0
4.8
4.6
4.5
4.5
Which meant most of the time all cores where between 4.5 or 4.6 ghz and occasionally one core hit 5.1 ghz.
All cores to 4.9 ghz gave 10-15% more fps.
Anyway, this is how it works.
For acc you want 5 cores as it uses one big thread more than ac.
Raceroom and rf2 use 2 big threads like AC.
Now when you put the vr rendering on top, you'll directly lose fps. Afaik it needs to run linked to the big rendering thread of the games...
-------
Sadly nobody can tell you exactly what your performance will be or if you'll bottleneck your gpu or not.
For gpu it's pretty simple: rendering vr or monitors isn't a big difference. Just calculate the amount of pixels you'll have with your vr headset and use dsr/superresolution to simulate the same amount of pixels.
For the cpu: put the resolution to the lowest possible to not get limited by the graphics card but keep the graphics settings you would want to use (shadows/reflections/amount of objects are independent of the resolution and increase draw calls the cpu has to calculate).
Then see how many fps you get and then take 5% off of the result.
That's a good guess for vr performance.
----------
Comparing different CPUs for sim racing:
You can select the amount of rendering threads in cinebench. For sim racing, 2 and 3 is the amount you want to bench to simulate the situation.
This is very important! At single thread and full multi thread, AMD 3000 CPUs are better than almost all Intels.
But as soon as you load 2 or 3 cores of these CPUs, the boost frequency will drop a lot!
While the easy to overclock Intel CPUs will be rock steady at 4.9-5.2 ghz (depending on lottery and cooling etc).
At full multi thread, amd simply has more threads so a better score. Not interesting for sim racing...
Sadly almost no one benches 2 or 3 threads in cinebench although it simulates a lot of programs and games!
If you can throw some cinebench at your workstation CPUs, we can compare the score if you want.
I'd recommend cinebench r15. The newer r20 takes ages when using 1-3 threads and also uses avx, which no sim really uses...
My score for cinebench r15 with 3 threads is 610. Fluctuating around that value.
I hope that helps. Probably a lot of mistakes in this post, wrote it on the phone while laying in bed...
Good night
