Function _mm_fnmadd_sh

#[target_feature(enable = "avx512fp16")]
pub const fn _mm_fnmadd_sh(a: __m128h, b: __m128h, c: __m128h) -> __m128h

Multiply the lower half-precision (16-bit) floating-point elements in a and b, and subtract the intermediate result from the lower element in c. Store the result in the lower element of dst, and copy the upper 7 packed elements from a to the upper elements of dst.

Intel's documentation