並列コンピューティング

In numerical simulation, parallel computing is essential for scaling up and achieving high performance. This course will cover the parallel programming required for these purposes through lectures and hands-on exercises. In particular, it aims to equip students with the necessary skills from the perspective of parallelizing simulation codes.

Specifically, students will learn about CPU parallel computing using OpenMP, GPU parallel computing using OpenACC and CUDA, and parallel computing using MPI. The course covers the fundamentals of each programming approach. As examples of the application of these programming techniques to numerical computation, students will learn how to apply them to diffusion equations and acquire the practical skills necessary for parallelizing simulation codes. Students will also learn about parallelizing simulation codes using GPUs, which are equipped in many modern supercomputers. However, this course does not cover machine learning or AI..

In the hands-on exercises, students will use the Wisteria/BDEC-01 supercomputer operated by the University of Tokyo or the Miyabi supercomputer at the Joint Center for Advanced High Performance Computing (JCAHPC, jointly operated by the Center for Computational Sciences at the University of Tsukuba and the Information Technology Center at the University of Tokyo). Students will also learn the basics of how to use supercomputer systems.