That's pretty bizarre optimization, how does this not get throttled by the thread calls?