Multiprocessing

It's common to need to execute more than one thing in parallel in a modern software system. Machine learning programs and scientific simulations benefit from using the multiple cores available in a modern processor, dividing their work up between concurrent threads operating on the parallel hardware. Graphical user interfaces and network servers do their work "in the background," leaving a thread available to respond to user events or new requests.

As a simple example, suppose your program had to execute three steps: A, B, and C. These steps are not dependent on each other, meaning they can be completed in any order. Usually, you would simply execute them in order, as follows:

Figure 9.14: Processing with a single thread

However, what if you could do all of these steps at the same time, rather than waiting for one to complete before moving onto the next? Our workflow would look like this:

Figure 9.15:...