Novelties in Teaching High Performance Computing

摘要

High Performance Computing (HPC) involves solving intensive problems beyond desktop computing. Teaching HPC to undergraduate students requires special consideration of taking students comprehension to clusters of parallel machines - which can solve intensive tasks. In addition, the concepts of HPC should be well-aided with programming examples and assignments, in order make them understand the concepts, clearly. This paper focuses on the aspect of teaching HPC curriculum with a practical approach. The HPC course described in this paper, utilizes development and solving programming assignments on multiple HPC platforms in order to teach different paradigms of HPC. These include data-intensive computing, CPU-intensive computing, shared-memory, and message passing. The course also focuses on developing problem solving skills among students by aiding lecture contents with programming examples. In addition, the course focuses on maintaining effective interaction between the students and the instructor in order to develop their motivation and include topics of their interest. Evaluations revealed increased popularity of the course and improved student comprehension.