switched out normal parallel check in simplification with a dirtier cross product based one to remove a couple of normalize calls - 75% perf improvement