I think we can at least double the speed of the near prime searching. * Use `-O3`. * Use two (or more) threads. But first we need a benchmark or two so we know how much faster we are actually getting if we make these changes.