Python gevent vs thread vs sequence

最新推荐文章于 2024-09-26 07:45:00 发布

拖拉机拖拉机

最新推荐文章于 2024-09-26 07:45:00 发布

阅读量1.5k

点赞数

分类专栏：个人总结 python 周边网络编程文章标签： python thread

本文链接：https://blog.youkuaiyun.com/xluren/article/details/40782683

版权

个人总结同时被 3 个专栏收录

66 篇文章

订阅专栏

python 周边

46 篇文章

订阅专栏

网络编程

13 篇文章

订阅专栏

本文通过实验对比了同步、多线程、协程及多进程四种并发模型在单核CPU环境下的执行效率。测试结果显示，使用gevent库实现的协程模型在处理大量任务时表现最优。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

测试机器单核

sequence：

(pythonenv)[xluren@test thread_coroutine_demo]$ cat sequence.py 
import time

def sync_task(i):
    #do something
    time.sleep(1)
    print i,"done"
def sync_run():
    start = time.time()
    for i in range(10):
        sync_task(i)
    end = time.time()

    print("sync task executed in %f second"%(end-start))

sync_run()

结果：

sync task executed in 10.011511 second
(pythonenv)[xluren@test thread_coroutine_demo]$

multithread

(pythonenv)[xluren@test thread_coroutine_demo]$ cat multithread.py 
import threading
import time
def sync_task(i):
    #do something
    time.sleep(1)
    print i,"done"
def multi_thread_run():
    start = time.time()
    a=[]
    for i in range(1000):
        t = threading.Thread(target=sync_task,args=(i,))
        a.append(t)
        t.start()
    for i in a:
        i.join()
    end = time.time()

    print("multi thread task executed in %f second"%(end-start))
multi_thread_run()
(pythonenv)[xluren@test thread_coroutine_demo]$

结果：

999 done
multi thread task executed in 1.407100 second

gevent：

(pythonenv)[xluren@test thread_coroutine_demo]$ cat gevent_demo.py 
import gevent
import time
def async_task(i):
    #do something
    gevent.sleep(1)
    print i,"done"
def async_run():
    start = time.time()
    coroutins = []

    for i in range(1000):
        coroutins.append(gevent.spawn(async_task,i))
    gevent.joinall(coroutins)
    end = time.time()
    print("async task executed in %f second"%(end-start))
async_run()
(pythonenv)[xluren@test thread_coroutine_demo]$

结果：

998 done
999 done
async task executed in 1.111667 second
(pythonenv)[xluren@test thread_coroutine_demo]$ cat gevent_demo.py

multiprocess

(pythonenv)[xluren@test thread_coroutine_demo]$ cat multiprocess.py 
import multiprocessing
import time
def sync_task(i):
    #do something
    time.sleep(1)
    print i,"done"
def multi_thread_run():
    start = time.time()
    a=[]
    for i in range(1000):
        t = multiprocessing.Process(target=sync_task,args=(i,))
        a.append(t)
        t.start()
    for i in a:
        i.join()
    end = time.time()

    print("multi process task executed in %f second"%(end-start))
multi_thread_run()
(pythonenv)[xluren@test thread_coroutine_demo]$

结果：

multi process task executed in 12.607443 second
(pythonenv)[xluren@test thread_coroutine_demo]$ cat multiprocess.py

到时再来个多核的测试，gevent 很好用。尤其他的monkey加上之后。