Function _mm256_mask_cvtps_epi64 

pub fn _mm256_mask_cvtps_epi64(src: __m256i, k: __mmask8, a: __m128) -> __m256i

Available on x86-64 and target feature avx512dq and target feature avx512vl only.

Convert packed single-precision (32-bit) floating-point elements in a to packed signed 64-bit integers, and store the results in dst using writemask k (elements are copied from src if the corresponding bit is not set).

Intel’s Documentation