https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88074
--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> --- OK, so it seems to be about large imagparts only and only if the realpart is non-zero. Thus in GCC we could stop folding ctan for imagpart (z) > 1000 for example. Like sin and cos tan is _usually_ invoked on normalized numbers(?).