CODEWITHSUNDEEP
python
amuttsch
amuttsch

Reputation: 1244

Speed dict mapping vs if elif

I have a function implemented as a lambda as well as a if elif ... and compared the execution time :

light = lambda c, v: {'turn on': 1,
                      'turn off': 0,
                      'toggle': (v+1)%2}[c]

def l(c, v):
    if c == 'turn on':
        return 1
    elif c == 'turn off':
        return 0
    elif c == 'toggle':
        return (v+1)%2
    return v

t = Timer(lambda: light('turn on', 0))
print('light: ' + str(t.timeit(number=23163958)))

t = Timer(lambda: l('turn on', 0))
print('l: ' + str(t.timeit(number=23163958)))

The output is:

light: 8.976719211001182
l: 3.9458757909887936

Why is the if statement almost twice as fast? Is it possible to increase the performance even more? I have to execute the function over 23 million times.

According to this I thought the dict lookup would be faster: https://stackoverflow.com/a/15925086/2014080

Upvotes: 0

Views: 1056

Answers (1)

chris
chris

Reputation: 5006

The problem is that each call of the lambda you instantiate the dictionary again, this takes time. If the dictionary is just being referenced, not created, it runs quicker than the conditionals.

Some simplified code to demonstrate the point.

from timeit import Timer

d = {'turn on': 1, 'turn off': 0}

light = lambda c : d[c]

t = Timer(lambda: light('turn on'))
print('light: ' + str(t.timeit(number=23163958)))

# light: 3.66314601898

Upvotes: 1

Related Questions