https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102943 --- Comment #32 from Richard Biener <rguenth at gcc dot gnu.org> --- But they are in reasonable territory now, no longer slowest of all but in the same ballpark as others.