Производительность в Python. Легкий путь

cimport cython
import operator
import math

cdef extern from "math.h":
    double sqrt(double x)
    double atan2(double y, double x)
    double cos(double x)
    double sin(double x)
    double floor(double x)

cdef class v2d(object):
    cdef public double x, y
    def __init__(self, x, y=None):
        if x is not None and y is not None:
            self.x = float(x)
            self.y = float(y)
        elif type(x) == v2d:
            self.x = x.x
            self.y = x.y
        elif type(x) == list or type(x) == tuple:
            self.x = x[0]
            self.y = x[1]
        else:
            raise TypeError()
    
    # Position
    def get_pos(self):
        return (self.x, self.y)
    def set_pos(self,pos):
        self.x = float(pos.x)
        self.y = float(pos.y)
    pos = property(get_pos, set_pos)

    def get_int_pos(self):
        cdef int x, y
        x = int(floor(self.x))
        y = int(floor(self.y))
        return(x,y)
    ipos = property(get_int_pos, set_pos)

# ...

$ for power in 2 4 6 ; do for module in timing_{native,ctypes,cython,cython_range} ; do echo -n "$module: $power "; python2 -m timeit --setup "from $module import test" "test(10 ** $power)" ; done ; echo "---------" ; done
timing_native: 2 100000 loops, best of 3: 4.72 usec per loop
timing_ctypes: 2 100000 loops, best of 3: 2.27 usec per loop
timing_cython: 2 1000000 loops, best of 3: 0.21 usec per loop
timing_cython_range: 2 1000000 loops, best of 3: 0.211 usec per loop
---------
timing_native: 4 1000 loops, best of 3: 448 usec per loop
timing_ctypes: 4 10000 loops, best of 3: 30.1 usec per loop
timing_cython: 4 100000 loops, best of 3: 5.57 usec per loop
timing_cython_range: 4 100000 loops, best of 3: 6 usec per loop
---------
timing_native: 6 10 loops, best of 3: 248 msec per loop
timing_ctypes: 6 100 loops, best of 3: 2.59 msec per loop
timing_cython: 6 1000 loops, best of 3: 541 usec per loop
timing_cython_range: 6 1000 loops, best of 3: 586 usec per loop
---------