http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54075
--- Comment #22 from Dennis Lubert <plasmahh at gmx dot net> 2012-07-27 19:20:54 UTC --- Am on vacation so I don't have the testcase at hand, but it is the same as likan posted in the original bugreport, minus the reserve. The main difference is that without reserve I see 21 rehashes in gcc 4.6 and 155 rehashes. I have added some code that after each insert tests if the bucket count has changed, I think that should be trivial to add for yourselves without me providing it.