But these performance numbers are meaningless without some sort of standard comparison case. So if you measure that e.g. some string operation takes 100ns, how do you compare against the numbers given here? Any difference could be due to PC, python version or your implementation. So you have to do proper benchmarking anyway.