Python, C/C++ (переплетается): многоязычный, Процедурный

Четыре реализации:

• [переплетаются]
• [Python]
• [психо]
• [список]

Код:

#!/usr/bin/env python
""" weave_factorial.py

"""
# [weave] factorial() as extension module in C++
from scipy.weave import ext_tools

def build_factorial_ext():
    func = ext_tools.ext_function(
        'factorial', 
        r"""
        unsigned long long i = 1;
        for ( ; n > 1; --n)
          i *= n;

        PyObject *o = PyLong_FromUnsignedLongLong(i);
        return_val = o;
        Py_XDECREF(o); 
        """,  
        ['n'], 
        {'n': 1}, # effective type declaration
        {})
    mod = ext_tools.ext_module('factorial_ext')
    mod.add_function(func)
    mod.compile()

try: from factorial_ext import factorial as factorial_weave
except ImportError:
    build_factorial_ext()
    from factorial_ext import factorial as factorial_weave


# [python] pure python procedural factorial()
def factorial_python(n):
    i = 1
    while n > 1:
        i *= n
        n -= 1
    return i


# [psyco] factorial() psyco-optimized
try:
    import psyco
    factorial_psyco = psyco.proxy(factorial_python)
except ImportError:
    pass


# [list] list-lookup factorial()
factorials = map(factorial_python, range(21))   
factorial_list = lambda n: factorials[n]

Измеряют относительный уровень:

$ python -mtimeit \
         -s "from weave_factorial import factorial_$label as f" "f($n)"

1. n = 12

   • [ткут] 0,70 µ секунда ( 2 )
   • [Python] 3,8 µ секунда ( 9 )
   • [психо] 1,2 µ секунда ( 3 )
   • [список] 0,43 µ секунды ( 1 )

2. n = 20

   • [ткут] 0,85 µ секунда ( 2 )
   • [Python] 9,2 µ секунда ( 21 )
   • [психо] 4,3 µ секунда ( 10 )
   • [список] 0,43 µ секунда ( 1 )

µ секунда стоит в течение многих микросекунд.