Client prediction input latency
I've been going through the documentation for server authorative client prediction and there is something weird.
We collect input in Update, but only actually use that locally for the motor a few frames later (in sync with Fixed update).
If my game is running at 144FPS that is an Update() call every 6.94MS
On a Fixed update of 60, that means we only use that input locally every 16.6MS.
What this means is that we are adding 2 or more frames of input latency.
Is there any way around this (i.e. immediately using the client input on the movement motor in Update) which still working with server authorative?