pythonobjectmultiprocessingpoolstarmap

Multiprocessing for creating objects + calling functions using starmap() Python


I would like to create objects of class Training and create multiple processes which call the print() function.

I have a class Training:

class Training():

    def __init__(self, param1, param2):
        self.param1 = param1
        self.param2 = param2
    
    def print(self):
        print(self.param1)
        print(self.param2)

I have tried to use the starmap function to create 5 processes in the following way:

import multiprocessing as mp

num_devices = 5

func_args = []
for i in range (0, num_devices):
    func_args.append((i, i*10))

with mp.Pool(num_devices) as pool:
    obj = pool.starmap(Training, func_args, chunksize=1)
    obj[0].print()
    obj[1].print()
    obj[2].print()
    obj[3].print()
    obj[4].print()

However, this code is is creating multiple processes to create the objects, and not to run the print() function. How can I do it in the correct way?


Solution

  • Create a list of objects initialised with the required parameters, and a helper function to call the method of object:

    def helper_func(obj):
        obj.print()
    
    func_args = []
    for i in range (0, num_devices):
        obj = Training(i, i*10)
        func_args.append(obj)
    
    with mp.Pool(num_devices) as pool:
        pool.map(helper_func, func_args, chunksize=1)