Rustc fails to perform autovectorization for a loop when targeting aarch64-apple-darwin

const N: usize = 4096;

#[inline(never)]
pub fn example_fn(xs: &[u64; N], ys: &[u64; N], res: &mut [u64; N]) {
    for i in 0..N {
        res[i] = xs[i] + ys[i];
    }
}

I expected to see VADD in resulting assembly but unfortunately instead it uses just adds:

        add     z0.d, z4.d, z0.d
        add     z1.d, z5.d, z1.d
        add     z2.d, z6.d, z2.d
        add     z3.d, z7.d, z3.d

Link to godbolt: https://godbolt.org/z/YWhT5GqxW

Just for reference I would expect it to mirror x86 SSE: https://godbolt.org/z/z5Pf3G17e

I think rustc should be able to perform autovectorization without additional hints/features enabled and use VADD in the loop.

Start with the Rust reproducer and inspect the aarch64-apple-darwin output in the linked Godbolt example, comparing it with the x86 SSE example. Trace the rustc compilation path responsible for loop autovectorization and verify the generated assembly. Done means this loop is autovectorized for the reported target without additional hints or features.