Python Interpreters Benchmarks

This project aims to compare different Python interpreters using wide set of benchmarks.

Structure of the project

• bencher/programs - source codes of benchmarks
• bencher/bin - python scripts to make repeated measurements of benchmarks
• bencher/makefiles - configuration for running benchmarks
• website - PHP scripts and configuration used to generate the pybenchmarks.org website

Currently only Linux is supported for running benchmarks.

How to run benchmarks

Prerequisities (ArchLinux)

yay -S dpkg python libgtop python-gobject\
	python2 pypy pypy3 pyston python-numba python-git ironpython \
	jython cython micropython-git graalpython-bin rustpython-git nuitka \
	python-gmpy2 python-numpy python-jinja
pip-pyston3 install numpy gmpy2 jinja2
pip-jython install jinja2
pip-pypy install jinja2
pip-pypy3 install jinja2

Run one benchmark for all interpreters

1. do some changes in source code of some benchmark
2. run this benchmark:

make run

Run all benchmarks for one interpreter

./clear_lang.sh <name_of_interpreter>  # e.g. cython
make run