Let’s Put the Memory Model Front and Center When Teaching Parallel Programming in C++

When teaching parallel programming in C++, the memory model is often treated as an afterthought. Even if it is included in the lectures, it may only be covered as an isolated topic near the end of the whole course. We have reorganized the Parallel Computing course at the University of Vienna to make the memory model an integral part of the course, starting from the very first lecture. Being aware of the memory model and understanding its basic principles helps the students better comprehend parallel programming in C++, even when the memory model is only discussed informally. In the paper, we describe how we integrated the memory model into the course. Based on test results, assignments, and feedback from the students, we consider this approach to be successful.