I posted this in a bug report already, but want to bring it up here, to discuss potential workarounds, if the idea is correct. I've been having same issue as a lot of people with DK2 after patch, in my case with AMD 290X (latest beta driver).
I was previously using 0.6 runtime in Win10, not SteamVR, and played at 1.5x SS, full HMD quality slider, and MSAAx2 mostly smoothly.
Updated OVR runtime to 0.8 and set up SteamVR yesterday, and suddenly 1.5x SS is essentially the same as 1x, but with a moderate hit to frame rate. 2x doesn't even seem to be as clear as 1.5 used to be. I think I've seen mention of the same thing happening for people with access to Oculus 1.3 runtime, so I'm guessing the problem is with Elite and the way it's handling DK2s in 2.0.7.
I'm wondering if maybe the SS multipliers were changed to expect the CV1 resolution, causing a weird coefficient with the DK2? Possibly the multiplier associated with the HMD Quality slider could have been changed too/instead. In the actual settings file, it is called "HMD pixels per display pixel", or something similar, so perhaps the resolution at 1.0 used to be hardcoded to match the DK2 screen, but was increased to match the CV1.
I don't know if this is actually what's going on, but if so it seems like we could be able to tweak one or both of those coefficients to get back to a round multiple of the DK2 resolution.
I'll update this if I can figure it out myself....
I was previously using 0.6 runtime in Win10, not SteamVR, and played at 1.5x SS, full HMD quality slider, and MSAAx2 mostly smoothly.
Updated OVR runtime to 0.8 and set up SteamVR yesterday, and suddenly 1.5x SS is essentially the same as 1x, but with a moderate hit to frame rate. 2x doesn't even seem to be as clear as 1.5 used to be. I think I've seen mention of the same thing happening for people with access to Oculus 1.3 runtime, so I'm guessing the problem is with Elite and the way it's handling DK2s in 2.0.7.
I'm wondering if maybe the SS multipliers were changed to expect the CV1 resolution, causing a weird coefficient with the DK2? Possibly the multiplier associated with the HMD Quality slider could have been changed too/instead. In the actual settings file, it is called "HMD pixels per display pixel", or something similar, so perhaps the resolution at 1.0 used to be hardcoded to match the DK2 screen, but was increased to match the CV1.
I don't know if this is actually what's going on, but if so it seems like we could be able to tweak one or both of those coefficients to get back to a round multiple of the DK2 resolution.
I'll update this if I can figure it out myself....