Mastering Object-oriented Python

Improving performance

We'll look at two performance tweaks for the Power3 class.

First, a better algorithm. Then, a better algorithm combined with memoization, which involves a cache; therefore, the function becomes stateful. This is where callable objects shine.

The first modification is to use a Divide and Conquer design strategy. The previous version chopped into n steps; the loop carried out n individual multiplication operations. If we can find a way to split the problem into two equal portions, the problem decomposes into steps. Given pow1(2,1024) ...