Benchmarking serialization/unserialization in python using json, pickle and cPickle

bench.py

import cPickle
import pickle
import json
import random
from time import time
from hashlib import md5

test_runs = 1000

def float_list():
    n = 1000
    d = []
    for i in range(n):
        d.append(random.randrange(0, 99) + random.random())
    return d

def int_list():
    n = 1000
    d = []
    for i in range(n):
        d.append(random.randrange(0, 9999))
    return d

def str_list():
    n = 1000
    d = []
    for i in range(n):
        d.append(md5(str(random.random())).hexdigest())
    return d


if __name__ == "__main__":

    payload = {
        "float": float_list(),
        "int": int_list(),
        "str": str_list()
    }

    for payload_type in payload:

        data = payload[payload_type]
        
        start = time()
        for i in range(test_runs):
            serialized = json.dumps(data)
        duration = time() - start
        print("json,serialize,%s,%.3f" % (payload_type, duration))

        start = time()
        for i in range(test_runs):
            unserialized = json.loads(serialized)
        duration = time() - start
        print("json,unserialize,%s,%.3f" % (payload_type, duration))

        start = time()
        for i in range(test_runs):
            serialized = pickle.dumps(data)
        duration = time() - start
        print("pickle,serialize,%s,%.3f" % (payload_type, duration))

        start = time()
        for i in range(test_runs):
            unserialized = pickle.loads(serialized)
        duration = time() - start
        print("pickle,unserialize,%s,%.3f" % (payload_type, duration))

        start = time()
        for i in range(test_runs):
            serialized = cPickle.dumps(data)
        duration = time() - start
        print("cPickle,serialize,%s,%.3f" % (payload_type, duration))

        start = time()
        for i in range(test_runs):
            unserialized = cPickle.loads(serialized)
        duration = time() - start
        print("cPickle,unserialize,%s,%.3f" % (payload_type, duration))

result.md

Result from my MacBook using Python 2.7.5:

module	direction	data type	Duration (sec)
json	serialize	int	0.103
json	unserialize	int	0.134
pickle	serialize	int	2.462
pickle	unserialize	int	2.614
cPickle	serialize	int	0.185
cPickle	unserialize	int	0.288
json	serialize	float	1.197
json	unserialize	float	0.392
pickle	serialize	float	3.931
pickle	unserialize	float	2.144
cPickle	serialize	float	0.699
cPickle	unserialize	float	0.717
json	serialize	str	0.137
json	unserialize	str	0.368
pickle	serialize	str	4.702
pickle	unserialize	str	4.945
cPickle	serialize	str	0.572
cPickle	unserialize	str	0.631

Hi. I've forked, shortened, and added marshal and highest pickle protocol setting to your code.