Visualization for Supercomputer System: A Survey

With the rapid development of science and technology, supercomputer systems are growing in size and complexity, which poses challenges for their usage and maintenance. Visualization provides an intuitive way to reveal the status of supercomputer systems. It enables users to gain a deeper understanding of the internal mechanisms of the systems, making it easier to use and optimize them. This survey summarizes the related work on supercomputer system visualization and classifies them into three categories: status monitoring, performance optimization, and system maintenance. Our survey shows that visualization helps analyze the status of internal processes and data for status monitoring; identify and optimize performance bottlenecks for performance optimization; and discover and analyze anomalies for system maintenance. Finally, we discuss future research opportunities, such as large-scale data processing, global optimization, and downstream task migration.