http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50724
--- Comment #6 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-10-14 15:07:40 UTC --- I was also thinking, maybe from the user point of view, a good way to deal with this kind of problem today is splitting the computation in parts via the new optimization attribute and pragma, keep the is_nan & co classifications outside the -fast-math cores.
