5x app performance improvement via caching intermediate results and some clever bit vector manipulation. My work here is done.