Numba is a Just-In-Time (JIT) compiler that translates python code into machine code at runtime via the LLVM compiler. The performance of Numba-optimized algorithms comes very close to the speeds of native C execution. Numba also has support for parallelization as well as GPU acceleration. 

Numba Module

Cython is an optimizing compiler that allows computationally expensive Python code to be translated into C and compiled for use within a python program. It contains a Python-like syntax as well as static typing. The main performance gain comes from Cython being able to bypass the CPython API (Where python code is interpreted from). 

Cython Module

Pypy is an alternative implementation of Python that uses a Just-In-Time compiler instead of an interpreter. Pypy consumes less memory and and is compatible with many python libraries. On average, Pypy is 4.2x faster than the standard CPython implementation.

Pypy

Python code can usually be optimized in few ways:

* Re-factoring code with optimizations that are specific to python.
* Using a Just-In-Time compiler (JIT) such as Numba or Pypy.
* Using an optimizing compiler such as Cython.

California State University, San Bernardino

It's important to measure the performance impact of your Python code and doing so will help you write faster, high performance code in the future. One way you can accurately evaluate the performance of Python code is with Python's built-in timeit module.

Evaluating the Performance of Python Code

Python's built-in timeit module allows you to measure the amount of time a code snippet takes to run 1,000,000 times (or any number of times you specify). The module can be used from the command line, the python interpreter, or inside of a .py file. 

Learn Before

Related

Learn After