python 性能测试（二）：for，函数向量化，计算向量化

 

import math
import numpy as np
from datetime import datetime as dt
from multiprocessing import Pool

def fun(v):
    return math.exp(v)

def FunVec():
    init()    
    fun_vecterize = np.vectorize(fun,otypes = [float])
    x = fun_vecterize(v)
    return x

def FunFor():
    init()
    x = np.zeros((iter,1))
    for i in range(iter):
        x[i] = math.exp(v[i])
    return x

def Vec():
    init()    
    x = np.exp(v)
    return x

def init():
    global iter,v
    iter = 1000
    v = np.random.randn(iter,1)   

if __name__ == '__main__':
    iter = 1000
    v = np.random.randn(iter,1)
    x = Vec()

> python -m timeit -s "import hellopy" "hellopy.Vec()"
5000 loops, best of 5: 44.3 usec per loop
> python -m timeit -s "import hellopy" "hellopy.FunVec()"
1000 loops, best of 5: 336 usec per loop
> python -m timeit -s "import hellopy" "hellopy.FunFor()"
500 loops, best of 5: 681 usec per loop

for循环，函数向量化和计算向量化之间的性能对比

函数向量化提升的效果没有计算向量化的提升明显