Real-world Python code 700 times slower than C

[Skip Montanaro]

    Tim> By cheating a bit I came up with a version using Numeric that runs
    Tim> more than 50x faster than the original Python version.

If Tim can cheat, so can I. ;-)

I modified the version I wrote that uses PyInline to use Py_List* macros to
set the elements of the array:

    #!/usr/bin/env python

    import PyInline

    m = PyInline.build(code="""
    void
    Ramp(PyObject* result, int size, double start, double end)
    {
        double step = (end-start)/(size-1);
        int i;

        for (i = 0; i < size; i++) {
            Py_DECREF(PyList_GET_ITEM(result, i));
            PyList_SET_ITEM(result, i, PyFloat_FromDouble(start + step*i));
        }
    }
    """, language="C")

    def Ramp(result, size, start, end):
        step = (end-start)/(size-1)
        for i in xrange(size):
            result[i] = start + step*i

    def main(ramp):
        array = [0.0]*10000
        for i in xrange(100):
            ramp(array, 10000, 0.0, 1.0)

    import time
    t1 = time.time()
    main(m.Ramp)
    t1 = time.time()-t1
    print "fast:", t1

    t2 = time.time()
    main(Ramp)
    t2 = time.time()-t2
    print "slow:", t2

    print "fast is", t2/t1, "times faster than slow"

This version runs about 17x faster than the pure Python version.  (The first
version I posted ran about 7x faster.)  I would have added a PyList_Check
test in there and done the usual set an exception and return NULL if the
test failed, but I don't couldn't (easily) figure out how PyInline handles
functions that return PyObject*.

-- 
Skip Montanaro (skip at pobox.com - http://www.mojam.com/)