Python Parallel Programming Cookbook

Python Parallel Programming Cookbook is the Chinese translation of the book by the Italian author Giancarlo Zaccone. The original was published by Packt Publishing in 2015; the Chinese edition, translated by Zhang Long and Song Bingjin, was released by Publishing House of Electronics Industry in April 2018. The book has six chapters in a "cookbook" format: each topic revolves around a complete, runnable Python program, explaining the approach and then showing the output, so readers can start immediately and adapt the code for their own projects. Chapter 1 begins with the memory architecture and programming models of parallel computing, explains how to design parallel programs and evaluate their performance, and places Python in the parallel world along with basic concepts of processes and threads. Chapter 2 covers thread-based parallelism, using the threading module to demonstrate thread definition, Lock/RLock, semaphores, condition variables, events, the with statement, and queue-based communication, and discusses performance evaluation of multithreaded applications. Chapter 3 turns to process-based parallelism, covering process creation, naming, running in the background, termination, process pools, and inter-process synchronization, and uses mpi4py to introduce point-to-point communication, deadlock avoidance, and collective communication such as broadcast, scatter, gather, and Alltoall. Chapter 4 covers asynchronous programming, introducing concurrent.futures and asyncio's event loop, coroutines, tasks, and Futures. Chapter 5 discusses distributed Python, involving Celery, SCOOP, Pyro4, PyCSP, MapReduce on Disco, and RPyC. Chapter 6 covers GPU programming, using PyCUDA, NumbaPro, and PyOpenCL to demonstrate numerical computation on GPUs. Note that the original dates from the early Python 3 era; libraries such as NumbaPro are no longer maintained, and some APIs differ from current versions, so readers should focus on understanding principles and approaches.