[Bug tree-optimization/124180] [16 Regression] No longer vectorized due to missed if-conversion

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124180

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
There is range info from a !(v_33 > 6.5535e4) && !(v_33 < 0.0) predicate, but
if v_33 is NaN then this is true and (short unsigned int) v_33 should raise
IEEE invalid, so as given we do have to predicate the conversion w/
-ftrapping-math.

So I fear this is INVALID and the GCC 15 behavior is wrong-code still.