你如何使用timeit来计算一个函数运行多长时间？

Question

Trying to figure out how long it takes this piece of code to run:

import timeit as t

def fib_recursive(n):

    if n==0:
        return 0
    elif n == 1:
        return 1
    else:
        return fib_recursive(n-1) + fib_recursive(n-2)
print fib_recursive(29)
print t.Timer("fib_recursive(29)")

Output was the following:

514229 timeit.Timer instance at 0xda28c0

Answer 1

To expand on thefourtheye's comment (which is correct), you generally want to isolate the steps necessary to define a function into the setup parameter for timeit . Given your setup, I would do the following:

import timeit as t

def fib_recursive(n):
    if n==0:
        return 0
    elif n == 1:
        return 1
    else:
        return fib_recursive(n-1) + fib_recursive(n-2)

setup = 'from __main__ import fib_recursive'
t.timeit('fib_recursive(29)', setup=setup)

I'll assume that you're aware of the various techniques to improve this algorithm and are choosing to measure its speed to establish a baseline only. You can experiment with the number keyword parameter to timeit to control the number of repetitions.

Answer 2

When you use IPython, the easiest way for timing is using the magicfunction %timeit :

%timeit fib_recursive(10)
>>> 10000 loops, best of 3: 70.2 us per loop

Answer 3

Use timeit.Timer.timeit :

timer = t.Timer("fib_recursive(29)", setup='from __main__ import fib_recursive')
print timer.timeit()

or simply use timeit.timeit :

print t.timeit("fib_recursive(29)", setup='from __main__ import fib_recursive')

NOTE: You need to pass import statement to setup argument.