Want to keep learning?

This content is taken from the Partnership for Advanced Computing in Europe (PRACE)'s online course, Python in High Performance Computing. Join the course to learn more.

Skip to 0 minutes and 11 seconds To understand better parallel computing we’ll be discussing

Skip to 0 minutes and 15 seconds the two main types of execution units: processes and threads in more detail. Let’s first look at threads. If you have a parallel program that starts as a single process here we have a single process executing a parallel program. If it uses threads at some point the program starts something called a parallel region that starts multiple threads that will start executing the program. At this point you will get multiple threads executing the same program. These threads will then execute the instructions in the program in parallel. These threads are short-lived and the main feature is that they share the memory of the parent process, so they are sharing the memory of each other.

Skip to 1 minute and 21 seconds Therefore this kind of parallelization is often called shared memory parallelization. At some point in the program execution you might want to close the parallel region and at this point all the threads are closed or joined together and afterwards you will have a serial execution of the program again. These parallel regions you can have multiple of them in the same program so you could open a new parallel region after this and so forth. The main thing here is that you have multiple threads and they are all sharing the same memory space.

Skip to 2 minutes and 7 seconds In other languages than Python, for example C or Fortran This would be commonly used with something called OpenMP to create the threads and execute them in parallel Unfortunately in Python this kind of the threaded approach is quite limited so let’s next look at in more detail something called processes and how to use those in parallel. So, now instead of having a single process we will have multiple processes in parallel. So we’ll have one, two, three, and four processes they are independent but they are executing the same program code. These independent processes they have their own memory space and they are completely independent from each other. Therefore, to exchange information between these processes.

Skip to 3 minutes and 19 seconds you need to explicitly send and receive messages between the processes.

Skip to 3 minutes and 28 seconds This we will discuss in more detail using something called MPI.

Skip to 3 minutes and 44 seconds you

Processes and threads

Parallel processing is normally based either on using multiple processes or multiple threads.

In this video we discuss the most important differences between them.

Share this video:

This video is from the free online course:

Python in High Performance Computing

Partnership for Advanced Computing in Europe (PRACE)