https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90356
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jakub at gcc dot gnu.org
--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
The rules for + 0.0 say that:
for non-zero x,
x + (+/-0.0) = x
0.0 + 0.0 = 0.0
-0.0 + (-0.0) = -0.0
-0.0 + 0.0 = 0.0 unless rounding towards negative infinity, in that case it is
-0.0.
So, for -frounding-math we can't optimize this, but with just -fsigned-zeros I
think we could optimize (x + 0.0) + 0.0 to x + 0.0 and (x + (-0.0)) + (-0.0) to
x + (-0.0), because if x is non-zero, then the result will be still non-zero,
when not rounding towards negative infinity and if x is 0.0, the result will
still be 0.0 and when not rounding towards negative infinity and if x is -0.0
the result will still be 0.0 in the first case and -0.0 in the latter case.
