Set each bit of mask result based on the most significant bit of the corresponding packed single-precision (32-bit) floating-point element in a.
See Implementation
Set each bit of mask result based on the most significant bit of the corresponding packed single-precision (32-bit) floating-point element in a.