https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105480
--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Kewen Lin from comment #8) > (In reply to Richard Biener from comment #7) > > (In reply to Kewen Lin from comment #6) > > > Both fcmpu and fcmp would trap for sNaN, is it expected with the current > > > GCC > > > implementation? > > > > But the key is -fsignalling-nans (default off) > > > > Thanks for the hint, but the behaviors don't change with one more explicit > option -fsignalling-nans (before or after the option for trapping-math). I > saw the option description saying "This option is experimental and does not > currently guarantee to disable all GCC optimizations that affect signaling > NaN behavior.", does it mean target codes don't honor it much so far? That could be very well the case.
