Performance penalty for using classes?

[Hans Nowak]

Miki Tebeka wrote:

> I need my M.Sc. to run fast but I don't want to recode it in C.
> I've noticed that using classes can cause my code to run about 30%
> slower.
> (See the below tests)
> 
> --- class_test.py ---
> #!/usr/bin/env python
> import random
> 
> class Dot:
>     def __init__(self, x, y):
>         self._x = x
>         self._y = y
> 
>     def x(self):
>         return self._x
> 
>     def set_x(self, x):
>         self._x = x
> 
>     def y(self):
>         return self._y
> 
>     def set_y(self, y):
>         self._y = y

These four methods serve no purpose and only incur overhead. Replace this with:

class Dot:
     def __init__(self, x, y):
         self.x = x
         self.y = y

Now, replace calls like "d.x()" with "d.x", and "d.set_x(z)" with "d.x = z".

> def test_class(iterations):
>     while iterations > 0:
>         d = Dot(random.random(), random.random())
>         x, y = d.x(), d.y()
>         d.set_x(y)
>         d.set_y(x)

It looks like this can be replaced with:

   d.x, d.y = d.y, d.x

No need for temporary variables and redundant method calls. :)

Now try running this code again. On my box, they both time at around 2.8s.

Of course, after this rewrite, you can still use Psyco, Python 2.3a1, and a 
faster computer. ;-)

HTH,

-- 
Hans (base64.decodestring('d3VybXlAZWFydGhsaW5rLm5ldA=='))
# decode for email address ;-)
The Pythonic Quarter:: http://www.awaretek.com/nowak/
Kaa:: http://www.awaretek.com/nowak/kaa.html