https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68664
James Greenhalgh <jgreenhalgh at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Target|powerpc*-*-*, aarch64*-*-* |powerpc*-*-*, aarch64*-*-*, | |arm*-*-* --- Comment #6 from James Greenhalgh <jgreenhalgh at gcc dot gnu.org> --- While this could still be a series of target issues, the decision to make the square root operation unconditional is bad across a number of targets, so I wonder whether there is a common heuristic that can be tweaked. On AArch64 this is happening in sched1, and the branch looks like it is being incorrectly statically predicted as likely taken.