https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92335
--- Comment #2 from joseph at codesourcery dot com <joseph at codesourcery dot com> --- On Mon, 4 Nov 2019, rguenth at gcc dot gnu.org wrote: > The issue is probably some FP constraints that say we cannot elide ret += 0.0, Yes, with signed 0 you can't elide + 0.0 as -0.0 + 0.0 is +0.0 except in FE_DOWNWARD mode. (The other obstruction to eliding it is signaling NaNs. There's also the matter of quantum exponents in DFP formats, though I suspect GCC optimizations don't generally consider the matter of ensuring results have the correct quantum exponent.)