Benchmarks of Python interpreters and compilers.

Benchmarks are only tentative. Feel free to contribute if you know how to improve the test programs.

 
 x64 Arch Linux™ Intel® i5-4210U® dual-core 
 
Python 2.7.13 (with NumPy and gmpy)
PyPy 5.9.0 (with gmpy_cffi)
Python 3.6.2 (with NumPy and gmpy2)
Python 3.7.0a0
PyPy3 5.9.0 (with gmpy_cffi)
IronPython 2.7.8a1 (with Mono 5.0.0.100)
Jython 2.7.1 (with JDK 8.u131)
Cython 0.27.1 (with NumPy and gmpy)
Nuitka 0.5.27.0
Shedskin 0.9.4
Numba 0.34.0 (with NumPy and Anaconda 4.4.0)
Pyston 0.6.1
MicroPython 1.9.2
Grumpy r123.08f9c2e
 
 x64 Arch Linux™ Intel® i5-4210U® dual-core 
 
Perform an N-body simulation of the Jovian planets
Repeatedly access a tiny integer-sequence
Search for solutions to shape packing puzzle
Generate and write random DNA sequences
Calculate Fibonacci number
Calculate 50th Fibonacci number
Calculate an eigenvalue using the power method
Read DNA sequences and write their reverse-complement
Generate a Mandelbrot set and write a portable bitmap
Repeatedly update hashtables and k-nucleotide strings
Match DNA 8-mers and substitute nucleotides for IUB code
Calculate the digits of Pi with streaming arbitrary-precision arithmetic
Repeatedly perform symmetrical thread rendezvous requests
Repeatedly switch from thread to thread passing one token
Render Jinja2 template
Allocate and deallocate many many binary trees
PyStone benchmark
Richards benchmark
Input/output benchmark
JSON encoding/decoding benchmark
 
 x64 Arch Linux™ Intel® i5-4210U® dual-core 
 
Which programs are fastest?

  Conclusions   License   Play   Source code