[Bug target/93112] Incorrect rounding for float to uint64 on x86 (32bit) with -fexcess-precision=standard

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93112

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
It is not ia32 vs. every other architecture, it is just that FLT_EVAL_METHOD 0
is the most common and you can get it even on ia32 with -msse2 -mfpmath=sse.
Some m68k configurations have FLT_EVAL_METHOD 2 too, I believe s390 has
FLT_EVAL_METHOD 1.