Memory models provide a framework for describing how memory operations behave and interact within a computing system, and a solid understanding of them is vital. They help developers avoid race conditions and undefined behavior in parallel and concurrent programs, enable software to fully exploit the underlying hardware, and play a central role in the design of programming languages and compilers that abstract hardware complexity. Recognizing this importance, the CS3211 Parallel and Concurrent Programming course at the National University of Singapore (NUS) incorporates memory models as a key topic. To balance simplicity with realism, we introduced a simplified subset of the C++20 memory model that remains faithful to the original while being more accessible to students, thus providing them with the foundation for writing correct and efficient concurrent programs. We formally proved that the CS3211 memory model is indeed a subset of the C++20 model, ensuring that programs written under it are also valid in C++. Building on this proof, we also developed and evaluated methods for teaching memory models effectively to undergraduates.