Question

我试图处理一个函数20次，并且试图使用多线程来减少耗时，但是串行处理的实际执行时间比mutithread处理要快得多，我有点困惑关于python中的多线程，在我看来，它会使程序更快吗？还是应该使用多处理或池？

start = time.time()
N = 100
M = 100
NumOfThread = 20
threads = []
start = time.time()
for i in range(NumOfThread):
    t = threading.Thread(target = mccala,args=(N,M))
    t.start()
    threads.append(t)

for thread in threads:
    thread.join()

end = time.time()
print(end - start )

这将返回4秒，但是像这样的串行处理

start = time.time()
for i in range(NumOfThread):
    t = mccala(N,M)
end = time.time()

print(end- start)

让我返回1秒

mccala函数如下所示

def mccala (searchRange,density):

selectedCore = [0]*(searchRange*density)


i = random.randint(0,searchRange*density)
core = i/ (searchRange*density)
selectedCore[i] = 1
k =0
while k < K :
    lam = random.uniform(core-1/searchRange,core+1/searchRange)
    if lam > 0 and lam < 1:
        response = rm.binomial(1,func(lam),1)[0]
        if response == 1:

            if lam < core :
                core = core - 1/ (searchRange*density)
                i -=1
            else:
                core = core + 1/(searchRange*density)
                i+= 1
    selectedCore[i] += 1
    k += 1
return q.put([i/K*10 for i in selectedCore])

导入包是

import math
import numpy.random as rm
import matplotlib.pyplot as plt
from scipy.interpolate import UnivariateSpline
import numpy as np
import random
import threading
import queue,time
from multiprocessing import Pool

多线程比串行处理慢

0 个答案: