benchmark_results.txt

Results of Example 2:

# Author: Zak Fallows (zakf@mit.edu)
# Copyright 2013
# Released for free use under the Creative Commons - Attribution license (CC-BY)
# If you reuse this material, please remember to give me credit.


#============================= Example 2, ex2.py ==============================#

# The following is what printed on my Mac when I ran ex2.py:

Interpreted Python:
    Time: 1.38237094879 seconds

Untyped Cython:
    Time: 0.438313007355 seconds
    Ratio: 3.154

Typed Cython, Python Function:
    Time: 0.208416938782 seconds
    Ratio: 6.633

Typed Cython, C Function:
    Time: 9.53674316406e-07 seconds
    Ratio: 1449521.000

# That is an implausibly large speed boost.
# By inspecting the C code output by Cython, I can see that the 
# C code really does call the function 10,000,000 times. So the 
# speed boost probably comes from either the compiler inlining 
# the function, or more likely, the compiler (in my case GCC) 
# noticed that the result is unused so it completely skipped 
# this piece of code, it generated no corresponding machine code 
# at all. GCC is too smart for us!

Typed Cython, C Function, Fudged:
    Time: 0.0279951095581 seconds
    Ratio: 49.379


#============================= Example 3, ex3.py ==============================#

# The following is what printed on my Mac when I ran ex3.py:

Interpreted Python:
    Result = 11611
    Time: 3.67643713951 seconds

Untyped Cython:
    Result = 11611
    Time: 1.72477602959 seconds
    Ratio: 2.132

Typed Cython, Python Functions:
    Result = 11611
    Time: 1.40393590927 seconds
    Ratio: 2.619

Typed Cython, Python Functions II:
    Result = 11611
    Time: 1.02636504173 seconds
    Ratio: 3.582

Typed Cython, C Functions:
    Result = 11611
    Time: 0.0388560295105 seconds
    Ratio: 94.617