National University of Defense Technology, China, China
With the coming of the next-generation of exascale supercomputer systems, the scale of high-performance computer systems is becoming increasingly large and complex, and the scale of HPC applications and the amount of data have increased sharply. This requires high-performance computing professionals to comprehensively use a variety of knowledge from the perspective of high-performance computer systems to carry out theoretical study and engineering practice, and to have the system ability to solve HPC practical problems. Multi-layered various knowledge includes parallel computer architecture, operating system/resource management, compilation and library optimization, domain-oriented algorithms and more. This talk focuses on the goal of cultivating the system-level comprehensive capability of high-performance computing majors and elaborates on the practice, experience and challenges faced by China’s National University of Defense Technology in developing a series of high-performance computing courses. NUDT has developed the Tianhe-2 supercomputer, which was the world’s fastest supercomputer, according to the TOP500 list, six times between June 2013 and November 2015, and now ranks the fifth in the TOP500 list. NUDT has accumulated rich experience in the development of high-performance computer systems, and also offers a series of high-performance computing-related courses from freshman to postgraduate. NUDT paid special attention to the cultivation of students’ system-level comprehensive abilities in terms of course-to-course integrity, relevance and level. NUDT has achieved good results in the past ten years of practical exploration.