Memory Access Microbenchmark

faster_py.py

import time

def initTrades(trades):
    for i, trade in enumerate(trades):
        if i % 2:
            trade["side"] = 'S'
        else:
            trade["side"] = 'B'
        trade["tradeId"] = i
        trade["clientId"] = 1
        trade["venueId"] = 123
        trade["instrumentCode"] = 321
        trade["price"] = i
        trade["quantity"] = i

def perfRun(runNum, trades):
    start = time.time() * 1000
    initTrades(trades)
    buyCost, sellCost = 0, 0
    for trade in trades:
        if trade["side"] == 'B':
            buyCost += trade["price"] * trade["quantity"]
        else:
            sellCost += trade["price"] * trade["quantity"]
    print runNum, "- duration" , ((time.time() * 1000) - start), "ms"
    print "buyCost = ", buyCost, " sellCost = ", sellCost

if __name__ == '__main__':
    NUM_RECORDS = 500 * 1000# * 444
    PyMemTrade = {"tradeId": 0, "clientId": 0, "venueId": 0, 
                  "instrumentCode": 0, "price": 0, "quantity": 0, "side": "0"}
    trades = [PyMemTrade.copy() for i in xrange(NUM_RECORDS)]
    for i in range (0, 5):
        perfRun(i, trades)

faster_pypy.py

import time

NUM_RECORDS = 50 * 1000 * 50

class PyMemTrade():
    def __init__(self, *args, **kwargs):
        self.__dict__.update(kwargs)

def initTrades(trades):
    for i, trade in enumerate(trades):
        trade.tradeId = i
        trade.clientId = 1
        trade.venueId = 123
        trade.instrumentCode = 321
        trade.price = i
        trade.quantity = i
        if (i % 2 == 0):
            trade.side = 'B'
        else:
            trade.side = 'S'

def perfRun(runNum, trades):
    start = time.time() * 1000
    initTrades(trades)
    buyCost, sellCost = 0, 0
    for trade in trades:
        if trade.side == 'B':
            buyCost += trade.price * trade.quantity
        else:
            sellCost += trade.price * trade.quantity      
    print runNum, "- duration" , ((time.time() * 1000) - start), "ms"
    print "buyCost = ", buyCost, " sellCost = ", sellCost

if __name__ == '__main__':
    trades = [PyMemTrade(0,0,0,0,0,0,'0') for i in xrange(NUM_RECORDS)]
    for i in xrange(5):
        perfRun(i, trades)

output.txt

python faster_py.py
0 - duration 196.531982422 ms
buyCost =  20833208333500000  sellCost =  20833333333250000
1 - duration 191.945068359 ms
buyCost =  20833208333500000  sellCost =  20833333333250000
2 - duration 192.145996094 ms
buyCost =  20833208333500000  sellCost =  20833333333250000
3 - duration 192.202148438 ms
buyCost =  20833208333500000  sellCost =  20833333333250000
4 - duration 192.083007812 ms
buyCost =  20833208333500000  sellCost =  20833333333250000

python pandas_python.py 
buyCost = 20833208333500000 sellCost = 20833333333250000
0 - duration 71.9406738281 ms
buyCost = 20833208333500000 sellCost = 20833333333250000
1 - duration 69.2270507812 ms
buyCost = 20833208333500000 sellCost = 20833333333250000
2 - duration 68.9450683594 ms
buyCost = 20833208333500000 sellCost = 20833333333250000
3 - duration 68.9328613281 ms
buyCost = 20833208333500000 sellCost = 20833333333250000
4 - duration 68.9848632812 ms

pandas_py.py

import time
from pandas import DataFrame
import numpy as np

def initTrades(trades):
    trades['side'][trades.index % 2 == 0] = 'B'
    trades['side'][trades.index % 2 == 1] = 'S'
    for k in ['tradeId', 'price', 'quantity']:
        trades[k] = np.array(trades.index)
    trades["venueId"] = 123
    trades["instrumentCode"] = 321

def perfRun(runNum, trades):
    start = time.time() * 1000
    initTrades(trades)
    result = trades['price'] * trades['quantity']
    buyCost = result[trades["side"] == 'B'].sum()
    sellCost = result[trades["side"] == 'S'].sum()
    print "buyCost =", buyCost, "sellCost =", sellCost
    print runNum, "- duration" , ((time.time() * 1000) - start), "ms"

if __name__ == '__main__':
    NUM_RECORDS = 500 * 1000 #* 444
    PyMemTrades = {}
    for k in ["tradeId", "clientId", "venueId", "instrumentCode", "price",
              "quantity"]:
        PyMemTrades[k] = np.zeros(NUM_RECORDS)
    PyMemTrades["side"] = np.chararray(NUM_RECORDS)
    PyMemTrades["side"].fill('0')
    trades = DataFrame(PyMemTrades)
    for i in xrange(0, 5):
        perfRun(i, trades)