The search chart is more troubling. How exactly are you searching? Both std::vector and std::deque have random access iterators, so you should be able to get log(n) search time – the best possible. std::list only has bi-directional iterators, so by definition it only offers O(n) search time. I suspect you have an unsorted container and are just doing std::find, which just does a linear walk. This is suboptimal, typically when searching a container is an important operation, you ensure the container is sorted and do binary searches.

I suspect the differences in your chart are due to the methodology you used, but since you haven’t posted the source code I can’t be of any help here. But generally:

* When inserting, you should be comparing std::vectors push_back to the equivalent in array_list
* When inserting, you should be comparing std::deque push_front to the equivalent
* When searching, you should be using the member function search/find etc on the stl containers if present, otherwise use std::binary_search (to test t/f) or std::lower/upper_bound (to actually get the element). And the container, obviously, should be presorted.

It’d also be helpful if you posted source code when you make such claims.

I’m having a hard time getting around those graphs. What does the insert graph mean ? I can’t really imagine that inserting one element into a list ( not at the end or beginning but somewhere in the middle ) should be that much slower than doing the same op on a vector.

Greets,
Mad

P.S.: Now that I read the previous post, I guess having named the “insert” graph “push_back” graph would quickly resolve those questions

It would be interesting to see what the cause for the speed difference is, i.e. to compare the implementations.