Compare packed single-precision (32-bit) floating-point elements in a and b to see if either is NaN.
See Implementation
Compare packed single-precision (32-bit) floating-point elements in a and b to see if either is NaN.