chapter five

5 Re-implementing critical code with Cython

This chapter covers:

An introduction to Cython from a data processing perspective
Using Cython to implement performant NumPy functions
Writing Cython code for NumPy that can be used with threaded NumPy parallelism
Write general Cython code that can release the GIL and hence implement true threaded parallelism

Python is slow. The standard implementation is slow and the language dynamic features pay a performance toll. Many Python libraries are performant precisely because they are partially implemented in lower level languages making available efficient data processing algorithms. But sometimes we will need to implement our own high-performance algorithms in something faster than Python. In this chapter we are going to consider Cython, a superset of Python that is converted to C and is substantially more performant than Python.

There are plenty of alternatives—other than Cython—that can be integrated with Python for performance, we will start by a brief overview of the options that are available. After that we will delve into Cython proper.

5.1 An overview of alternatives for efficient code re-implementation

5.2 A Whirlwind introduction to Cython

5.2.1 A naïve implementation in Cython

5.2.2 Using Cython annotations

5 Re-implementing critical code with Cython

This chapter covers:

5.1 An overview of alternatives for efficient code re-implementation

5.2 A Whirlwind introduction to Cython

5.2.1 A naïve implementation in Cython

5.2.2 Using Cython annotations

5.3 Profiling Cython code

5.3.1 Using Python built-in profiling infrastructure

5.4 Optimizing array access with Cython memoryviews

5.4.1 Cleaning up all internal interactions with Python

5.5 NumPy generalized universal functions in Cython

5.6 Advanced array access in Cython

5.6.1 A Cython implementation of QuadLife

5.7 Parallelism with Cython

5.8 Summary