Python Multiprocessing: Guide to Using, Optimizing & Debugging

1. Basics: What is Python Multiprocessing?

1.1 What is Multiprocessing?

Features of Multiprocessing

• Each process has its own independent memory space
• Can fully utilize CPU cores
• Inter-process communication is required (using Queue or Pipe)

Typical Use Cases

• Compute-intensive tasks (machine learning, numerical simulation)
• Tasks that fully utilize CPU (image processing, data analysis)

1.2 Difference from Multithreading

What is GIL (Global Interpreter Lock)?

1.3 Simple Multiprocessing Example in Python

import multiprocessing
import time

def worker(n):
    print(f"Process {n} start")
    time.sleep(2)
    print(f"Process {n} finished")

if __name__ == "__main__":
    process_list = []

    # Create 3 processes
    for i in range(3):
        p = multiprocessing.Process(target=worker, args=(i,))
        process_list.append(p)
        p.start()

    # Wait until all processes finish
    for p in process_list:
        p.join()

    print("All processes finished")

1.4 Precautions When Using Multiprocessing

1. The if __name__ == "__main__": guard is required on Windows

Incorrect code (causes error)

import multiprocessing

def worker():
    print("Hello from process")

p = multiprocessing.Process(target=worker)
p.start()

Correct code

import multiprocessing

def worker():
    print("Hello from process")

if __name__ == "__main__":
    p = multiprocessing.Process(target=worker)
    p.start()
    p.join()

1.5 Summary

• What is multiprocessing? → A method to run multiple processes in parallel
• Difference from multithreading → Not affected by the GIL and suitable for CPU-bound tasks
• Simple example in Python → Use multiprocessing.Process()
• Precautions on Windows → if __name__ == "__main__": is required

2. Practical Guide: Using the multiprocessing Module

2.1 Overview of the multiprocessing Module

Key Features of multiprocessing

2.2 Basic Usage of the Process Class

Creating a Basic Process

import multiprocessing
import time

def worker(n):
    print(f"Process {n} start")
    time.sleep(2)
    print(f"Process {n} finished")

if __name__ == "__main__":
    p1 = multiprocessing.Process(target=worker, args=(1,))
    p2 = multiprocessing.Process(target=worker, args=(2,))

    p1.start()
    p2.start()

    p1.join()
    p2.join()

    print("All processes finished")

2.3 Interprocess Communication (Queue & Pipe)

Sending and Receiving Data Using Queue

import multiprocessing

def worker(q):
    q.put("Hello from child process")

if __name__ == "__main__":
    q = multiprocessing.Queue()
    p = multiprocessing.Process(target=worker, args=(q,))
    p.start()
    p.join()

    # Retrieve data from child process
    print(q.get())

2.4 Shared Memory Using Value and Array

import multiprocessing

def worker(val, arr):
    val.value = 3.14  # Change the value in shared memory
    arr[0] = 42       # Change the array value

if __name__ == "__main__":
    val = multiprocessing.Value('d', 0.0)  # 'd' is double type
    arr = multiprocessing.Array('i', [0, 1, 2])  # 'i' is integer type

    p = multiprocessing.Process(target=worker, args=(val, arr))
    p.start()
    p.join()

    print(f"val: {val.value}, arr: {arr[:]}")

2.5 Process Management Using the Pool Class

Parallel Processing Using Pool

import multiprocessing

def square(n):
    return n * n

if __name__ == "__main__":
    with multiprocessing.Pool(4) as pool:
        results = pool.map(square, range(10))

    print(results)

2.6 Summary

• multiprocessing module makes parallel processing easy to implement
• Create individual processes with the Process class
• Using Queue or Pipe enables data sharing between processes
• Value and Array provide shared memory
• Using the Pool class allows efficient processing of large amounts of data

Ad

3. Advanced: Error Handling and Performance Optimization

3.1 Common Errors in multiprocessing and Solutions

Error 1: Missing if __name__ == "__main__": error on Windows

Error Message

RuntimeError: freeze_support() must be called if program is run in frozen mode

Solution

import multiprocessing

def worker():
    print("Hello from process")

if __name__ == "__main__":  # This is required
    p = multiprocessing.Process(target=worker)
    p.start()
    p.join()

Error 2: PicklingError (cannot pass functions between processes)

Error Message

AttributeError: Can't pickle local object 'main..'

Solution

import multiprocessing

def square(x):  # make it a global function
    return x * x

if __name__ == "__main__":
    with multiprocessing.Pool(4) as pool:
        results = pool.map(square, range(10))  # avoid lambda
    print(results)

Error 3: Deadlock (process remains stopped)

Solution

import multiprocessing

def worker(q):
    q.put("data")

if __name__ == "__main__":
    q = multiprocessing.Queue()
    p = multiprocessing.Process(target=worker, args=(q,))
    p.start()
    print(q.get())  # receive data
    p.join()  # normal termination here

3.2 Performance Optimization Techniques

Optimization 1: Set the number of processes appropriately

import multiprocessing

def worker(n):
    return n * n

if __name__ == "__main__":
    num_workers = multiprocessing.cpu_count()  # get number of CPU cores
    with multiprocessing.Pool(num_workers) as pool:
        results = pool.map(worker, range(100))
    print(results)

Optimization 2: Use Pool.starmap()

import multiprocessing

def multiply(a, b):
    return a * b

if __name__ == "__main__":
    with multiprocessing.Pool(4) as pool:
        results = pool.starmap(multiply, [(1, 2), (3, 4), (5, 6)])
    print(results)

Optimization 3: Leverage shared memory

import multiprocessing
import ctypes

def worker(shared_array):
    shared_array[0] = 99  # modify the value in shared memory

if __name__ == "__main__":
    shared_array = multiprocessing.Array(ctypes.c_int, [1, 2, 3])  # create shared memory
    p = multiprocessing.Process(target=worker, args=(shared_array,))
    p.start()
    p.join()
    print(shared_array[:])  # [99, 2, 3]

3.3 Summary

• multiprocessing common error avoidance methods explained
• Performance optimization points:
• Set the number of processes appropriately
• Leverage starmap()
• Speed up with shared memory

4. FAQ: Common Questions and Solutions

4.1 Which should you use in Python, multiprocessing or multithreading?

Answer

• CPU-bound (computationally intensive tasks) → Multiprocessing (multiprocessing)
• I/O-bound (file/network operations) → Multithreading (threading)

4.2 Why does multiprocessing feel “slow”?

Answer

• High cost of creating processes → Use Pool
• Too much data copying → Use shared memory (Value, Array)
• Processing many small tasks → Try concurrent.futures.ThreadPoolExecutor

import multiprocessing

def worker(n):
    return n * n

if __name__ == "__main__":
    with multiprocessing.Pool(multiprocessing.cpu_count()) as pool:
        results = pool.map(worker, range(100))
    print(results)

4.3 How to share dictionaries and lists in multiprocessing?

Answer

import multiprocessing

def worker(shared_list):
    shared_list.append(100)  # Update shared list

if __name__ == "__main__":
    with multiprocessing.Manager() as manager:
        shared_list = manager.list([1, 2, 3])
        p = multiprocessing.Process(target=worker, args=(shared_list,))
        p.start()
        p.join()
        print(shared_list)  # [1, 2, 3, 100]

4.4 Common errors with multiprocessing.Pool and how to address them?

4.5 How to debug Python multiprocessing?

Answer

import multiprocessing
import logging

def worker(n):
    logger = multiprocessing.get_logger()
    logger.info(f"Process {n} running")

if __name__ == "__main__":
    multiprocessing.log_to_stderr(logging.INFO)  # Enable logging
    p = multiprocessing.Process(target=worker, args=(1,))
    p.start()
    p.join()

Ad

5. Summary and Additional Learning Resources

5.1 Summary of This Article

Fundamentals of Multiprocessing

• What is multiprocessing? → A technique that runs multiple processes in parallel to maximize CPU utilization
• Difference from multithreading
• Multiprocessing → Suited for CPU‑bound tasks (numerical calculations, image processing, etc.)
• Multithreading → Suited for I/O‑bound tasks (file handling, network communication, etc.)

How to Use the multiprocessing Module

• Process class to create individual processes
• Using Queue and Pipe to send and receive data between processes
• Leveraging Value and Array to utilize shared memory
• With the Pool class to execute parallel processing efficiently

Error Handling and Performance Optimization

• Common errors
• If you omit if __name__ == "__main__":, you get errors on Windows
• lambda functions and local functions cause PicklingError
• Forgetting to call Queue.get() leads to deadlocks
• Performance optimization
• Use Pool to reduce the overhead of creating processes
• Use starmap() to pass multiple arguments
• Utilize multiprocessing.shared_memory to reduce data copy overhead

5.2 Additional Learning Resources

1. Python Official Documentation

• multiprocessing — Process‑based parallelism

2. Online Tutorials

• Real Python – Python’s multiprocessing Module
• GeeksforGeeks – Multiprocessing in Python

5.3 Looking Ahead to Future Applications

Technologies to Learn Next

• Asynchronous processing (asyncio) → Parallelize I/O‑bound tasks
• concurrent.futures → Unified management of threads and processes

5.4 Conclusion

Ad