pub fn _mm_cvtsh_i64(a: __m128h) -> i64
Convert the lower half-precision (16-bit) floating-point element in a to a 64-bit integer, and store the result in dst.
Intel’s documentation