https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122101
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Severity|normal |enhancement
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
So the difference comes from doing the induction variable as a double rather
than conversion to double. and that is I think only allowed with
-fno-trapping-math where GCC and LLVM default differs too. (note GCC still does
not do it with -fno-trapping-math). I have not been able to find a simplier
testcase where LLVM will change the type of the induction variable to double.
