Function _mm_cvtsh_h

#[target_feature(enable = "avx512fp16")]
pub const fn _mm_cvtsh_h(a: __m128h) -> f16

Copy the lower half-precision (16-bit) floating-point element from a to dst.

Intel's documentation