While this is not good representation of real-world production malloc behavior, it is representative of length (instruction-wise and well as cycle-wise) of fast-path. So this is better than nothing.
