Function safe_arch::load_unaligned_m128

#[must_use]pub fn load_unaligned_m128(a: &[f32; 4]) -> m128

This is supported with target feature sse only.

Loads the reference into a register.

This generally has no speed penalty if the reference happens to be 16-byte aligned, but there is a slight speed penalty if the reference is only 4-byte aligned.

let a = [10.0, 12.0, 13.0, 14.0];
let b = load_unaligned_m128(&a);
assert_eq!(m128::from_array(a).to_bits(), b.to_bits());