Hello,
I was looking into the cause of a GPU bottleneck and came across the “FEngineloop_Tick_PollGameDeviceState”. This call is taking up from 25ms up to 100ms. However I have no clue what is causing this or how to prevent it.
Can anyone help?
This was profiled through a standalone build using the Oculus Rift S.
I would like to know too - I came across this one on the a standalone build for the Quest
I came across this too. My frame time is spiking for a long time due to this, for about 20 seconds at a time. Could this be a VR specific thing?
I came across this too, is taking about 75% of each frame. Im using unreal insights connected to a deployed game on oculus quest 2.
Apparently this is due to your game running at a low fps, and then the engine locking the game at a lower fps to ensure proper tracking, hence waiting.
https://www.reddit.com/r/unrealengine/comments/ryy09u/comment/hzmkhwx/?utm_source=share&utm_medium=web2x&context=3
figured it out: vr headsets want a specific framerare (80 for oculus?) if your game is not able to hit this framerate, the game will run at 40 to let the hardware to smooth the hmd movemets autogenerating the missing framerate. so, if your game is running at 70 fps instead of 80 (14ms per frame) the framerate will be locked at 40 (25ms). so your game loop will have to wait for 25-14 = 11ms. you will find this time in Poll Device State.
so, if your game will reach the desired frame rate of your headset, you won’t see these ms “wasted”
