https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126946
--- Comment #7 from Drea Pinski <pinskia at gcc dot gnu.org> --- Note LLVM produces the exactly the same code generation ... Also this seems similar to PR 118505 . It is going to be hard to always to know if the data is truely random vs having a large value early and then not needing to take the branch. So I think this is more of just accidental choices.
