Well, your comment led with "Linked lists have more consistent performance than std::vector." But the tests just don't support that conclusion, and the truth is the opposite in a real-world environment: lists will kill your performance.
Before you'll see any downsides of list, you'll have to introduce memory fragmentation into your test and ensure that your list's allocations end up all over your address space, like they would be in a real engine. At that point you'll see a huge slowdown compared to vector due to CPU cache thrashing.
Before you'll see any downsides of list, you'll have to introduce memory fragmentation into your test and ensure that your list's allocations end up all over your address space, like they would be in a real engine. At that point you'll see a huge slowdown compared to vector due to CPU cache thrashing.