Python 線程與進程

Python由于有全鎖局的存在（同一時間只能有一個線程執(zhí)行）雏赦，并不能利用多核優(yōu)勢挎袜。所以助琐，如果程序的多線程進程是CPU密集型的符隙，那多線程并不能帶來效率上的提升憎账，相反還可能會因為線程的頻繁切換套硼，導致效率下降；如果是IO密集型胞皱，多線程進程可以利用IO阻塞等待時的空閑時間執(zhí)行其他線程邪意，提升效率看政。

但我們總會有在程序中實現多并發(fā)來提升程序運行效率的情形。在這些情形下抄罕，可以適當的利用多進程來實現提升效率允蚣。

另外，python中呆贿，為了解決網絡請求密集中嚷兔，延時等待的問題，我們還可是使用協程來提交效率。

在于IO密集型程序中饱狂，多線程應用較多惜颇。
但在網絡請求密集中，協程比多線程強上很多壶运。
在CPU密集中，還是進程應用更多浪秘。
以下是三個例子：

線程：

#!/usr/bin/python
# -*- coding: UTF-8 -*-

import threading
import time, datetime


# 為線程定義一個函數
def print_time(thread_name):
    for i in range(3):
        now = datetime.datetime.now()
        print(now, thread_name)
        time.sleep(1)

# 不帶線程處理的程序
for i in range(5):
    threadname = "threadName" + str(i)
    print_time(threadname)

#線程處理
# for i in range(5):
#     threadname = "threadName"+str(i)
#     t = threading.Thread(target=print_time(threadname))
#     t.start()

運行發(fā)現蒋情，不帶線程處理的程序和線程處理的程序運行順序是一樣的：

/Library/Frameworks/Python.framework/Versions/3.6/bin/python3.6 /Users/caobo/PycharmProjects/ThreadTest/threadTest.py

2017-11-08 11:55:32.889184 threadName0
2017-11-08 11:55:33.890456 threadName0
2017-11-08 11:55:34.894145 threadName0
2017-11-08 11:55:35.899107 threadName1
2017-11-08 11:55:36.900408 threadName1
2017-11-08 11:55:37.903821 threadName1
2017-11-08 11:55:38.907008 threadName2
2017-11-08 11:55:39.909538 threadName2
2017-11-08 11:55:40.914680 threadName2
2017-11-08 11:55:41.918500 threadName3
2017-11-08 11:55:42.921579 threadName3
2017-11-08 11:55:43.925748 threadName3
2017-11-08 11:55:44.928359 threadName4
2017-11-08 11:55:45.931913 threadName4
2017-11-08 11:55:46.932414 threadName4

Process finished with exit code 0

每個線程執(zhí)行完需要3秒，依次執(zhí)行線程耸携，總耗時15秒棵癣。

進程

# 進程處理
if __name__ == "__main__":
    for i in range(5):
        threadName = "threadName" + str(i)
        p = multiprocessing.Process(target=print_time, args=(threadName,))
        p.start()

進程處理運行結果如下：

/Library/Frameworks/Python.framework/Versions/3.6/bin/python3.6 /Users/caobo/PycharmProjects/ThreadTest/threadTest.py
2017-11-08 11:58:44.045175 threadName0
2017-11-08 11:58:44.046196 threadName1
2017-11-08 11:58:44.047036 threadName2
2017-11-08 11:58:44.048071 threadName3
2017-11-08 11:58:44.049010 threadName4
2017-11-08 11:58:45.045841 threadName0
2017-11-08 11:58:45.046783 threadName1
2017-11-08 11:58:45.048291 threadName2
2017-11-08 11:58:45.048476 threadName3
2017-11-08 11:58:45.050095 threadName4
2017-11-08 11:58:46.046989 threadName0
2017-11-08 11:58:46.047063 threadName1
2017-11-08 11:58:46.048629 threadName2
2017-11-08 11:58:46.049239 threadName3
2017-11-08 11:58:46.050937 threadName4

Process finished with exit code 0

每個進程執(zhí)行完需要3秒，并發(fā)執(zhí)行線程夺衍，總耗時3秒狈谊。

進程池

# 進程池處理
if __name__ == "__main__":
    pool = multiprocessing.Pool(processes=4)
    for i in range(5):
        threadName = "threadName" + str(i)
        pool.apply_async(print_time, (threadName,))
    pool.close()
    pool.join()
    print("Sub-process(es) done.")

進程池處理結果如下：

/Library/Frameworks/Python.framework/Versions/3.6/bin/python3.6 /Users/caobo/PycharmProjects/ThreadTest/threadTest.py
2017-11-08 12:01:03.557402 threadName0
2017-11-08 12:01:03.557552 threadName1
2017-11-08 12:01:03.557686 threadName2
2017-11-08 12:01:03.557827 threadName3
2017-11-08 12:01:04.558322 threadName2
2017-11-08 12:01:04.558306 threadName0
2017-11-08 12:01:04.558311 threadName1
2017-11-08 12:01:04.558322 threadName3
2017-11-08 12:01:05.558841 threadName1
2017-11-08 12:01:05.558833 threadName2
2017-11-08 12:01:05.558841 threadName0
2017-11-08 12:01:05.558845 threadName3
2017-11-08 12:01:06.560105 threadName4
2017-11-08 12:01:07.561340 threadName4
2017-11-08 12:01:08.561577 threadName4
Sub-process(es) done.

Process finished with exit code 0

由于設置了進程并發(fā)的數量為4，所以沟沙，前三秒執(zhí)行的都是前四個進程的內容（每個進程執(zhí)行完需要三秒）河劝，進程5只能在前四個進程執(zhí)行完成之后，才開始執(zhí)行矛紫∈晗梗總耗時6秒。

修改進程并發(fā)數量為5：

# 進程池處理
if __name__ == "__main__":
    pool = multiprocessing.Pool(processes=5)
    for i in range(5):
        threadName = "threadName" + str(i)
        pool.apply_async(print_time, (threadName,))
    pool.close()
    pool.join()
    print("Sub-process(es) done.")

運行結果如下：

/Library/Frameworks/Python.framework/Versions/3.6/bin/python3.6 /Users/caobo/PycharmProjects/ThreadTest/threadTest.py
2017-11-08 12:12:17.210982 threadName0
2017-11-08 12:12:17.211084 threadName1
2017-11-08 12:12:17.211188 threadName2
2017-11-08 12:12:17.211335 threadName3
2017-11-08 12:12:17.211451 threadName4
2017-11-08 12:12:18.211480 threadName0
2017-11-08 12:12:18.211481 threadName2
2017-11-08 12:12:18.211480 threadName1
2017-11-08 12:12:18.211785 threadName3
2017-11-08 12:12:18.211787 threadName4
2017-11-08 12:12:19.211773 threadName2
2017-11-08 12:12:19.211784 threadName0
2017-11-08 12:12:19.212676 threadName1
2017-11-08 12:12:19.212679 threadName4
2017-11-08 12:12:19.212679 threadName3
Sub-process(es) done.

Process finished with exit code 0

修改設置進程并發(fā)的數量為5含衔，所以煎娇，所有5個進程能夠同步執(zhí)行。每個進程執(zhí)行完需要三秒贪染，總耗時3秒缓呛。