inside the loop are many math calculations ... you could say that it is simple math calcs used to implement complex numerical approximations. Also remember that this isnt a small block of code ... its not uncommon for a Pentium 4 3+ GHz or a Pentium M to spend an hour inside just one execution of the loop. The values operated on all reside in class properties. Is that enough info to give you an idea of a preferred approach? Any ideas on how I can make this really lean and fast?