I'm not saying they're slower, I'm saying the performance is more consistent.
99.9% of the time the std::vec implementation will out preform the linked list. However, that 0.1% seems to be wildly worse than the linked list. This is not a problem for most programs, however in gaming engines that hiccup shows up as a visible hitch in the framerate.
So to you, that 0.1% is specific to gaming engines, how ? I don't really believe it.
I think "that 0.1%" demonstrates that linked list are really irrelevant and that vectors should be used all the time.
The only useful advantage of linked lists is to be able to insert an element in the middle of the list. I don't understand why anyone would want to insert data in the middle of a list, you rarely want to do that in O(1), and even if you do, you still need to know where exactly you want to insert it.
If you want to have data sorted, use map<>, if you want a faster map, use a hash map.
Maybe lists are relevant for vertex geometry, because it might be interesting to quickly edit and add vertices, but it's such a specific case I don't think it matters. Nobody really edits vertex geometry. You could even just build an index vector instead.
I don't believe it's problem endemic only to game engines, but game engines are particularly vulnerable to such latency. Google reveals many examples of how hitching causes problems for players.
I'm not going to claim that memory is the primary, or even a remotely major cause of such hitching, but it's certainly not going to help when you have to spend one of the 16 milliseconds given to render a frame on memory allocation and re-ordering.
99.9% of the time the std::vec implementation will out preform the linked list. However, that 0.1% seems to be wildly worse than the linked list. This is not a problem for most programs, however in gaming engines that hiccup shows up as a visible hitch in the framerate.