I've done this with a true-RMS power measurement-set. Not the best one, not
very accurate, but at least there shouldn't be any conceptual errors in the
measurment process. Simply calculating from measured voltage and current
will not work, because one have to take the current waveform and power
factor (phase shift) into account. (Well, I'm pretty sure you are aware of
this ;-)
The only problem with this measurement setup is the unknown and most likely
non-linear efficiency ratio of the PSU. But it should be sufficient to get a
rough idea of which power saving strategies really saves power ;-)