pub fn _mm512_abs_ph(v2: __m512h) -> __m512h
Finds the absolute value of each packed half-precision (16-bit) floating-point element in v2, storing the result in dst.
Intel’s documentation