https://gcc.gnu.org/bugzilla/show_bug.cgi?id=54075
--- Comment #50 from François Dumont <fdumont at gcc dot gnu.org> --- This performance issue is a result of fixing: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=41975 It resulted in many more modulo operations and so expensive float divisions. I plan to commit an alternative hash policy using power of 2 number of buckets so that modulo is trivial. Bench are showing that performance are better but still not at the level of tr1 implementation on the operations you are interested in. So it is difficult to close this ticket cause performance regression is still there but it might stay this way for a long time.
