Python 线程池 和 进程池 使用

[TOC]

概念

• 并发：串行
• 并行：并排运行

并行编程测试

from concurrent import futures
import time
import os


print("cpu 数量：", os.cpu_count())  # 测试笔记本是4核


def test(id, num):  # 涉及到多参数问题
    time.sleep(num)
    print(f"task {id} finished")
    return id



# 线程池 默认的是cpu的数目*5
start = time.time()
with futures.ThreadPoolExecutor() as tp:
    task_list = [tp.submit(test, i, i) for i in range(8, 0, -1)]
    # task_result = [t.result() for t in task_list]  # 有任务结束，就返回，有序结果
    task_result = [t.result() for t in futures.as_completed(task_list)]  # 有任务结束，就返回结果，无序
    print(task_result)
print(time.time() - start)


start = time.time()
with futures.ThreadPoolExecutor() as tp:
    task_result = tp.map(lambda args: test(*args), [(i, i) for i in range(8, 0, -1)])
    print(list(task_result))  # 原始顺序
print(time.time() - start)


# 进程池 默认的是cpu的数目
start = time.time()
with futures.ProcessPoolExecutor() as pp:
    task_list = [pp.submit(test, i, i) for i in range(8, 0, -1)]
    # task_result = [t.result() for t in task_list]  # 有任务结束，就返回，有序结果
    task_result = [t.result() for t in futures.as_completed(task_list)]  # 有任务结束，就返回，无序结果
    print(task_result)
print(time.time() - start)

# 进程池的map测试，需要修改测试函数

# 测试结果：线程池速度优于进程池

参考

• https://www.cnblogs.com/haiyan123/p/7461294.html