如何实现具有多个队列的多处理 IPC？

Question

I'm new to multiprocessing and multithreading. For learning purpose, i'm trying to achieve IPC using queues.

Code

from multiprocessing import Process, Queue, Lock

import math



def calculate_square(sq_q, sqrt_q):
    itm = sq_q.get()
    print(f"Calculating sq of: {itm}")
    square = itm * itm
    sqrt_q.put(square)

def calculate_sqroot(sqrt_q, result_q):
    itm = sqrt_q.get()
    print(f"Calculating sqrt of: {itm}")
    sqrt = math.sqrt(itm)
    result_q.put(sqrt)



sq_q = Queue()
sqrt_q = Queue()
result_q = Queue()


for i in range(5, 20):
    sq_q.put(i)


p_sq = Process(target=calculate_square, args=(sq_q, sqrt_q))
p_sqrt = Process(target=calculate_sqroot, args=(sqrt_q, result_q))


p_sq.start()
p_sqrt.start()



p_sq.join()
p_sqrt.join()

while not result_q.empty():
    print(result_q.get())

Explanation

Here i'm trying to run two of the functions with two different processes each of which calculates the square of the number and again calculate the square root of the number.

Queues

• sq_q: Queue containing the initial number whose square root is to calculated.
• sqrt_q: Queue containing the numbers whose square root has to be calculated
• result_q: Queue containing final result.

Issue

Only the first item of the sq_q is consumed.

Output:

5.0

I expect the output to be: [5, 6, 7, 8, .. , 19]

Please note that this is purely for learning purpose and i want to implement the IPC with multiple queues though it could have been possible to achieve with shared objects Locks and arrays.

Answer 1

You only call the functions once, so only the first value 5 is taken, you then want to loop all the values in the queue.

while not sq_q.empty():
    itm = sq_q.get()
    print(f"Calculating sq of: {itm}")
    square = itm * itm
    sqrt_q.put(square)

Same goes for the other function, but the condition here will be till result_q is full (give a max size to the result queue for the condition to work), then the final result will have the values.

    while not result_q.full():
        itm = sqrt_q.get()
        print(f"Calculating sqrt of: {itm}")
        sqrt = math.sqrt(itm)
        result_q.put(sqrt)

Complete Code

import math
from multiprocessing import Process, Queue


def calculate_square(sq_q, sqrt_q):
    while not sq_q.empty():
        itm = sq_q.get()
        print(f"Calculating sq of: {itm}")
        square = itm * itm
        sqrt_q.put(square)


def calculate_sqroot(sqrt_q, result_q):
    while not result_q.full():
        itm = sqrt_q.get()
        print(f"Calculating sqrt of: {itm}")
        sqrt = math.sqrt(itm)
        result_q.put(sqrt)


if __name__ == "__main__":
    sq_q = Queue()
    sqrt_q = Queue()
    result_q = Queue(5)

    for i in range(5):
        sq_q.put(i)

    p_sq = Process(target=calculate_square, args=(sq_q, sqrt_q))
    p_sqrt = Process(target=calculate_sqroot, args=(sqrt_q, result_q))

    p_sq.start()
    p_sqrt.start()

    p_sq.join()
    p_sqrt.join()

    while not result_q.empty():
        print(result_q.get())

Output

Calculating sq of: 0
Calculating sq of: 1
Calculating sq of: 2
Calculating sq of: 3
Calculating sq of: 4
Calculating sqrt of: 0
Calculating sqrt of: 1
Calculating sqrt of: 4
Calculating sqrt of: 9
Calculating sqrt of: 16
0.0
1.0
2.0
3.0
4.0

Edit:

As the calculate_sqroot now depends on result_q a delay is no longer needed.

Answer 2

Also we can use JoinableQueue which provides two of the most convenient method so that we can block the main process until the items in the queue are fully consumed.

i. task_done()

Indicate that a formerly enqueued task is complete. Used by queue consumers. For each get() used to fetch a task, a subsequent call to task_done() tells the queue that the processing on the task is complete.

If a join() is currently blocking, it will resume when all items have been processed (meaning that a task_done() call was received for every item that had been put() into the queue).

Raises a ValueError if called more times than there were items placed in the queue.

ii. join()

Block until all items in the queue have been gotten and processed.

The count of unfinished tasks goes up whenever an item is added to the queue. The count goes down whenever a consumer calls task_done() to indicate that the item was retrieved and all work on it is complete. When the count of unfinished tasks drops to zero, join() unblocks.

Solution

from multiprocessing import Process, Lock
from multiprocessing import Queue
from multiprocessing import JoinableQueue

import math, time

def calculate_square(sq_q, sqrt_q):
    while True:
        itm = sq_q.get()
        print(f"Calculating sq of: {itm}")
        square = itm * itm
        sqrt_q.put(square)
        sq_q.task_done()

def calculate_sqroot(sqrt_q, result_q):
    while True:
        itm = sqrt_q.get() # this blocks the process unless there's a item to consume
        print(f"Calculating sqrt of: {itm}")
        sqrt = math.sqrt(itm)
        result_q.put(sqrt)
        sqrt_q.task_done()
        
items = [i for i in range(5, 20)]

sq_q = JoinableQueue()
sqrt_q = JoinableQueue()
result_q = JoinableQueue()

for i in items:
    sq_q.put(i)
    
p_sq = Process(target=calculate_square, args=(sq_q, sqrt_q))
p_sqrt = Process(target=calculate_sqroot, args=(sqrt_q, result_q))

p_sq.start()
p_sqrt.start()

sq_q.join()
sqrt_q.join()
# result_q.join() no need to join this queue

while not result_q.empty():
    print(result_q.get())