pyspeeder

A repository for testing different approach to runtime optimization of Python programs

Testing different approaches to speeding up Python programs, e.g. Numba, c-optimization and so on

Numba

Things to keep in mind when working with numba

When working with numba and functions that work with lists, it is best to work with numpy arrays instead of Python lists
When working with numba instead of working with initially empty lists to which elements are appended, it is better to create an initial numpy array of defined type and length; this will help numba
alternatively, typed list type provided by numba can be applied
Tosee the impact of this compare example 2 with example 3. Example 3 uses numpy arrays of predefined lengths, while example 2 facilitates the same overall logic with python lists that are initially empty and then are appended to

Things to keep in mind when applying ThreadPoolExecuter from concurrent.futures