Comparing speeds of different particle systems is all fine and good but at least do so on equivalent machines. Simply stating that I can get x-thousand/hundred particles running on a system with 'no slowdown' is meaningless without knowing what hardware is being used.