There are many benchmarks performed by super computing centers and they usually test these tags, I can't find an example right now, but it might give you a place to look.
I think you should try it yourself for known systems. (E.g. H2 molecule, Cu crystal) I also wanted to increase the speed of my calculations before and tried using LREAL=.A instead of the default LREAL=.FALSE. The result is that I do not get the correct binding energy of H2 molecule.
You can also read the attached link which is related on speeding up VASP calculations.