Function _mm256_fmsubadd_ph

#[target_feature(enable = "avx512fp16", enable = "avx512vl")]
pub const fn _mm256_fmsubadd_ph(a: __m256h, b: __m256h, c: __m256h) -> __m256h

Multiply packed half-precision (16-bit) floating-point elements in a and b, alternatively subtract and add packed elements in c to/from the intermediate result, and store the results in dst.

Intel's documentation