Due to the physical architecture of CPUs/etc., data structures that have "worse" asymptotic performance are often quite a bit faster than those with "better" performance. For example, iterating through a small array to find an item and test for existence is often quite a bit faster than using a hash set for the same operation.