[Bug target/126933] Inlined SSE2 float -> bf16 conversions should be improved

[liuhongt at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126933

--- Comment #5 from Hongtao Liu <liuhongt at gcc dot gnu.org> ---

> 
> But it doesn't implement
> 
> (fromi + 0x7fff + ((fromi >> 16) & 1)) >> 16
> 
> when TARGET_AVXNECONVERT and TARGET_AVX512BF16 aren't enabled.

That's why we have flag_unsafe_math_optimizations