@jonrob was the implicit double promotion wanted to reduce the error on 3/4 pi? if so, please let me know. if not, @cattanem can unwip as he pleases.