https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117163
--- Comment #3 from Paul Caprioli <paul at hpkfft dot com> --- Note that this is not for me; it's for everybody else. The idea is for this to be the standard macro for specifying the IEEE rounding to nearest attribute `roundTiesToEven`. My contacts at Intel tell me they can't change the intrinsics guide until there is compiler support. So, it's a chicken-and-egg problem; somebody has to go first. It makes sense for the compilers to go first and the documentation to reflect the improvement only after it is usable. I hope we can agree that is wrong/misleading (for a human code reviewer) to see "round to nearest int" when an instruction does not do that (because it returns a correctly rounded floating point value). See also: https://github.com/llvm/llvm-project/pull/99691
