Python多处理嵌套字典

Question

I try to use the multiprocessing tool in python (3.4.2) to speed up my performance. My normal programm is running ok, but I have to process a large and nested dictionary in it. Attached a simplified version, which shows my problem. If I use the TestProc directly it is working, with multiprocessing not.

Thanks for your help!

import multiprocessing

def TestProc(liste, results):

    for i in liste:

        results[i] = {'power':{'square': float(i)**2, 'cubic': 
        float(i)**3},'root':{'square': float(i)**(1/2), 'cubic': float(i)**(1/3)}}

if __name__ == "__main__":

    multiprocessing.freeze_support()
    results = multiprocessing.Manager().dict()

    results = {}
    liste = ['1','2','3','4','5']

    for i in liste:

        results[i] = multiprocessing.Manager().dict()

    print(results)

    #TestProc(liste, results)

    p1 = multiprocessing.Process(target=TestProc, args=(liste,results,))
    p1.start()
    p1.join()

    print(results)

Answer 1

When you do:

results[i] = ...

you are not adding values to the DictProxy at results[i] , you are replacing it with another brand new dict , inside the outer results dict . Therefore, the only one you need to be synchronized is that outer dict .

Replace line 15 with results = multiprocessing.Manager().dict() , and remove the loop line 18-20 that never did anything.