A simple way to benchmark your code in Python — First published in fullweb.io issue #50

python-timeit.md

A simple way to benchmark your code in Python

You’re working on your script and want to see how fast a function would be. Here’s how to do that with the timeit module, included in the standard library.

Let’s say you have the following script:

import random

def sort_rand():
  random.seed('pouet')
  pool = [random.random() for i in range(1000)]
  sorted_pool = sorted(pool)
  return sorted_pool

if __name__ == "__main__":
  sorted_nb = sort_rand() 

And you want to check how fast sort_rand() could be. Just add these 3 lines:

import random

def sort_rand():
  random.seed('pouet')
  pool = [random.random() for i in range(1000)]
  sorted_pool = sorted(pool)
  return sorted_pool

if __name__ == "__main__":
  sorted_nb = sort_rand()

  from timeit import Timer
  print(Timer("sort_rand()", setup="from __main__ import sort_rand")
              .repeat(repeat=3, number=1000))

This will print the time, in seconds, to run sort_rand() 1000 times, and repeat that 3 times. The default number of calls if not provided is 1,000,000.

The output is a simple Python list with the elapsed time as a float, so you can call min(), max() and the likes to get the number the most meaningful to you. Be sure to see the official doc for more info and examples.