This training aims is to introduce the basic principles of GPU technology and programming to interested users. The course is designed for those who wish to understand the operating principles of current GPUs and explore how this new hardware-software eco-system can support research and industrial developments. In addition to teaching programming, the course also covers GPU hardware technology, since knowledge of the GPU execution mechanism is essential for the development of efficient GPU programs. The training consists of theoretical and practical parts. In the theoretical part, we will learn about the basics of GPU technology, the internal architecture and operation of GPU chips, followed by the basic principles of the CUDA programming language, and the design strategies of parallel GPU programs. In the practical session, participants will learn about compiling and running CUDA programs on the Komondor supercomputer, and explore various computational tasks with the help of simple sample programs.
Lecture: Dr. Zoltán Juhász, HPC Competence Centre