pub unsafe fn _mm_addsub_ps(a: __m128, b: __m128) -> __m128
Alternatively add and subtract packed single-precision (32-bit) floating-point elements in a to/from packed elements in b.
a
b
Intel’s documentation