A new arbitrary-precision mechanism for evaluating many numerical constants will be
explained, and it will be used to show how many of the elementary functions have been
implemented much more efficiently. Some optimizations to arithmetic and some potential
future directions will also be discussed.
