I haven't done extensive testing, but I did do some testing with opening large assemblies with the "automatic" rep vs opening them with the master rep. These were assemblies previously saved with Creo 2.
In those cases the "automatic" rep was actually slower than the master rep by 1.5-2X. I have not had a chance to test out assemblies made in Creo 4.
We have keep the default value for our deployment and address it in training.