pub unsafe fn _mm_sub_ps(a: __m128, b: __m128) -> __m128
Subtracts packed single-precision (32-bit) floating-point elements in a and b.
a
b
Intel’s documentation