无法在python中使用mutliprocessing修改全局dict变量 [英] Can't modify global dict variable using mutliprocessing in python

问题描述

我尝试使用多重处理来处理numpy数组.但是我不知道如何将处理结果返回到dict变量.

I try to use multiprocessing to process numpy array. But I don't know how to return the process result back to the dict variable.

使用注释掉代码可以产生我所期望的.但是当我尝试使用多处理时，对于max_rslts我什么也收不到，这是一个空洞的字典.

Use the comments out code can produce what I expected. But when I try to use multiprocessing, I can't get anything for the max_rslts, it is an empty dict.

示例代码为:

from __future__ import print_function

from multiprocessing import Process
import numpy as np

def calc_max(a, step):
    global max_rslts
    max_rslts[step] = a[::step].max()


if __name__ == '__main__':
    max_rslts = dict()

    a1 = np.arange(16).reshape(4,4)

#    for i in range(1,4):
#        calc_max(a1, i)
#    print(max_rslts)

    jobs = []
    for i in range(1,4):
        p = Process(target=calc_max, args=(a1, i))
        jobs.append(p)
        p.start()

    for j in jobs:
        j.join()

    print(a1)
    print(max_rslts)

推荐答案

来自

From the multiprocessing programming guidelines:

避免共享状态

Avoid shared state

分叉每个子流程后，进程之间不会共享

global.使用其他方式在进程之间共享状态 . dict管理器将在这里工作:

globals are not shared between processes after forking off each sub-process. Use different means to share state between processes instead. A dict manager would work here:

from __future__ import print_function

from multiprocessing import Process, Manager
import numpy as np


def calc_max(a, step, max_rslts):
    max_rslts[step] = a[::step].max()


if __name__ == '__main__':
    manager = Manager()
    max_rslts = manager.dict()
    a1 = np.arange(16).reshape(4,4)
    jobs = []
    for i in range(1,4):
        p = Process(target=calc_max, args=(a1, i, max_rslts))
        jobs.append(p)
        p.start()

    for j in jobs:
        j.join()

    print(a1)
    print(max_rslts)

这篇关于无法在python中使用mutliprocessing修改全局dict变量的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！