Even with perfect cache locality, a linked list will be significantly slower to traverse than an array because of worse pipelining and parallelization. Partiall

Even with perfect cache locality, a linked list will be significantly slower to traverse than an array because of worse pipelining and parallelization. Partially unrolled lists can help, but at the cost of additional complexity. Of course not all container accesses require traversal.