Function _mm256_mask_popcnt_epi16

#[target_feature(enable = "avx512bitalg", enable = "avx512vl")]
pub const fn _mm256_mask_popcnt_epi16(src: __m256i, k: __mmask16, a: __m256i) -> __m256i

For each packed 16-bit integer maps the value to the number of logical 1 bits.

Uses the writemask in k - elements are copied from src if the corresponding mask bit is not set. Otherwise the computation result is written into the result.

Intel's documentation