For many computationally hard problems six extra orders of magnitude isn't really going to buy you much.