https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89670
--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> --- (In reply to Jörn Engel from comment #4) > Fair enough. That means the only way to get tzcnt without a conditional is > by using inline asm. Of course not. Either you can use _tzcnt_u32, or you can use x ? __builtin_ctz (x) : 32, both with with -mbmi expand to tzcnt when optimizing.