Hi. I am lucky enough to have a G2 and a 4K monitor. Knowing from tuning the 4K settings that I can acheive a constant 60fps at 4K with a mix of medium and a few high and fewer ultra settings in dense areas, I did some quick math and the native resolution for the G2 is slightly more than 4K BUT close enough that if I can get 60FPS for 4K (8,294,400 pixels) then I should be able to get at least half of the native 90hz (45fps) for the G2 (9,331,200 pixels or 4.5K) because the G2 is only 12.5% more pixels than 4K.
So first thing to say is by fiddling with OXR and render resolution scale in MSFS, I kept the resultant resolution approx 4.5K but below 5K in VR at all times. I took off from Birmingham, UK airport and flew over Birmingham City centre at < 1000ft each test as it is Photgrammetry drawn and densely populated with buildings.
As many before me have stated and in simple terms, how many frames per second are delivered is derived from how fast the CPU can determine where every object is in relation to the the view point (Main Thread) and then how fast the GPU can receive that information and render that information to the display.
Long story - of endless tweaking sessions, loss of hair and learning a new language that I can’t speak in front of anybody I respect - cut short, I noticed that the 60-80fps I got in 4K did not translate at all with very similar settings to VR.
I have a 10900k, 3090 Ultra at stock and 32GB RAM DDR@3800Mhz, M.2 drive for OS and Fast SSD for MSFS so I was suprised to see how the frame times differed between 4K and VR.
In both VR and 4K I am seeing about 9-13ms frametime on the GPU
In 4K I am seeing 5-8ms frametime on the CPU (MainThread)
In VR I am seeing 19-28ms frametime in CPU (MainThread)
This I consider this to be an anomaly, I seem to be limited by my CPU.
I understand that each scene (left/right eye) has to be calculated for Stereo 3D but as each viewport is 5/8 the size of 4K would expect that the CPU calculations would only take (5/8 x 2) 25% longer, not over 3x longer.
Anyhoo, I have settled on Medium settings mostly for VR with rendering at 100 SS and TAA in MSFS and 50% in OXR to get about 4.7K overall resolution with super fluid motion, but will probably settle somewhere around 80/80 if I can get the same fluidity to get better clarity at distance and compromise on the guages, until such a time that Asobo work out that they don’t need to calculate the whole scene in 4.5K twice (which is a feasible assumption IMHO)
Happy to be humbled, but feel this may help others in some little way without having to spend 4 months fiddling to acheive what is impossible when assuming that 4.5K VR is similar to 4K pancake.