I spent some time testing the HT on/off in both DX11 and DX12 the other day. While the peak fps was marginally different, what I was focusing on was the 99% fps numbers as displayed by the Nvidia Performance Overlay, in an effort to see which combination gave the highest consistent frame pacing for the lows.
While I did not take time to record the results or take screenshots, my findings were that while using DX11, Hyperthreading Disabled gave the best 99% fps. Using DX12, Hyperthreading Enabled gave the best results.
Overall, out of the 4 combinations tested, DX12 with hyperthreading enabled gave the best 99% results, while DX11 with hyperthreading enabled was the worst. The split between the best and worst combination of hyperthreading and DX version was 4-6 fps average, on the 99% fps scale, on the ground at JFK. Again, this was focusing on the lowest fps recorded while spinning the camera in the cockpit like a possessed child during an exorcism.
Again, the peak fps number was not a focal point of my test, but the 99% fps scale should be a fairly decent indicator of CPU performance in regards to checking whether HT on or off is right for your system, since the dev mode counter, while thorough, doesnât do a particularly great job of showing trends for the lows.
Tl;dr, from best to worst on my system:
DX12 HT enabled
DX12 HT disabled
DX11 HT disabled
DX11 HT enabled
Based on these findings, and the fact that I am having a great time with DX12 on my rig, I have re-enabled hyperthreading. If, for some reason, I need to go back to DX11, I will be disabling hyperthreading once again.
YMMV, but if youâre bored, itâs an easy thing to test and might help smooth things out in dense areas.