• Your Language is FAST
• Python is SLOW

• Is a widely used general-purpose, high-level programming language
• Its design philosophy emphasizes code readability
• Its syntax allows programmers to express concepts in fewer lines of code than would be possible in languages such as C++ or Java
• More from Wikipedia

import this

• Beautiful is better than ugly
• Explicit is better than implicit
• Simple is better than complex
• Complex is better than complicated
• Readability counts

• Python Is Simple
• Python Is Easy
• You write somehow on Python
• You assume Python is SLOW

1. Profile and optimize your existing code
2. Use a C module (or write your own)
3. Try a JIT-enabled interpreter like Jython or PyPy
4. Parallelize your workload

• IPython

• line_profiler
• Hunter
• Interactive Profiler

def calc(operation, a, b):
    mapping = {
        'add': lambda a, b: a + b,
        'subtract': lambda a, b: a - b,
        'multiply': lambda a, b: a * b,
        'divide': lambda a, b: a / b,
    }
    return mapping[operation](a, b)

In [16]: %timeit calc('divide', 1, 2)
1000000 loops, best of 3: 699 ns per loop

In [17]: %prun calc('divide', 1, 2)
         5 function calls in 0.000 seconds

   Ordered by: internal time

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    0.000    0.000 {built-in method exec}
        1    0.000    0.000    0.000    0.000 <string>:1(<module>)
        1    0.000    0.000    0.000    0.000 <ipython-input-14-cf58e5c65b2c>:1(calc)
        1    0.000    0.000    0.000    0.000 <ipython-input-14-cf58e5c65b2c>:6(<lambda>)
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}

def calc(operator, a, b):
    if operator == 'add':
        return a + b
    elif operator == 'subtract':
        return a - b
    elif operator == 'multiply':
        return a * b
    elif operator == 'divide':
        if b == 0:
            raise ValueError('Unable divide to zero.')
        return a / b
    raise NotImplementedError

In [18]: %timeit calc('divide', 1, 2)
1000000 loops, best of 3: 258 ns per loop

In [19]: %prun calc('divide', 1, 2)
         4 function calls in 0.000 seconds

   Ordered by: internal time

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    0.000    0.000 {built-in method exec}
        1    0.000    0.000    0.000    0.000 <string>:1(<module>)
        1    0.000    0.000    0.000    0.000 <ipython-input-17-966ad73e599d>:1(calc)
        1    0.000    0.000    0.000    0.000 {method 'disable' of '_lsprof.Profiler' objects}

• Do not assign constants inside the loops
• Use operator library in map/filter/sorted/...

Python 2 optimizations

• xrange instead of range
• '{0}'.format(...) instead of '{}'.format(...)

Django optimizations

• Use .iterator() for big collections
• Use foreign_id instead of foreign__id

Slow Python

In [1]: import datetime

In [2]: %timeit datetime.datetime.strptime('2015-06-20', '%Y-%m-%d')
100000 loops, best of 3: 9.16 µs per loop

Fast Python

In [3]: from ciso8601 import parse_datetime

In [4]: %timeit parse_datetime('2015-06-20')
10000000 loops, best of 3: 144 ns per loop

Result: 63x Faster

Slow Python

In [9]: import json
In [10]: %timeit json.dumps({'key': 'value', 'date': datetime.date.today().isoformat()})
100000 loops, best of 3: 8.36 µs per loop

Fast Python

In [11]: import ujson

In [12]: %timeit ujson.dumps({'key': 'value', 'date': datetime.date.today().isoformat()})
100000 loops, best of 3: 4.6 µs per loop

Result: 2x faster

Slow Python. Python 2 only

In [2]: import pickle

In [3]: %timeit pickle.dumps({'key': 'value', 'date': datetime.date.today()})
10000 loops, best of 3: 36.1 µs per loop

Fast Python. Python 2 only

In [4]: import cPickle

In [5]: %timeit cPickle.dumps({'key': 'value', 'date': datetime.date.today()})
100000 loops, best of 3: 8.32 µs per loop

Result: 4.3x faster

• Library: slow / fast
• Memcached: python-memcache / pylibmc
• HTML Parsing: BeautifulSoup / lxml
• Decimals: decimal / cdecimal Python 2 only

• Use Cython!

Cython allows,

• Write Python code that calls back and forth from and to C or C++ code natively at any point
• Easily tune readable Python code into plain C performance by adding static type declarations
• Use combined source code level debugging to find bugs in your Python, Cython and C code

PyPy is a fast, compliant alternative implementation of the Python language

• It Fast. Thanks to its Just-in-Time compiler, Python programs often run faster on PyPy.
• “If you want your code to run faster, you should probably just use PyPy.” — Guido van Rossum
• It eats less Memory. Memory-hungry Python programs (several hundreds of MBs or more) might end up taking less space than they do in CPython.
• It Stackless. PyPy comes by default with support for stackless mode, providing micro-threads for massive concurrency.

• Supports Django, Twisted and other frameworks
• All you need to run pypy yourfile.py instead of python yourfile.py
• But designed to use Python 2.7.9

• Hadoop
• Elastic MapReduce
• Disco
• And many-many others

• concurrent
• multiprocessing
• eventlet / gevent
• asyncio
• Celery / rq