pub unsafe fn _mm256_sub_ps(a: __m256, b: __m256) -> __m256
Subtracts packed single-precision (32-bit) floating-point elements in b from packed elements in a.
b
a
Intel’s documentation