Comparison among PyPy, Cython and Numba

CPython is the standard Python implementation while there are alternative implementations, extensions and packages available to boost up the speed. However, some sacrifices are required to get the full throttle speed.

Here's the extract about the comparison of three popular approaches to make Python code running faster:

Name of technology	Python Package/Full implementation	Type of compiler	Dependency	Package supported	Python features supported	Coding style	Performance
PyPy	Full implementation in RPython	Just-in-time		Only pure Python package (Especially NOT SciPy, Matplotlib, and scikit-learn)	Full	Pure Python syntax	High, 10x times faster than CPython
Cython	Python package	Ahead-of-time			Partial	Cython syntax	Very high, 100x times faster than CPython
Numba	Python package	Just-in-time	LLVM		Partial	Only decorator syntax required ahead of desired function	Very high, 100x times faster than CPython

Benchmarks are collected from here.