Large Data Visualization on the K computer and Fugaku

Abstract

This paper presents and discuss the large data visualization environment on the K computer and supercomputer Fugaku. These flagship-class supercomputers have been designed to provide maximum performance during simulation runs, and they are capable of executing extreme-scale numerical simulations, which can generate large amounts of simulation results. In addition to these simulation results, large amounts of sensor measurements and system data have also been stored in form of time-varying multi-variate log data. We will present and discuss some efforts to provide large data visualization-oriented tools and applications on the provided hardware infrastructure which includes the HPC system itself and auxiliary pre- and post-processing systems. This includes some results from collaborative research work done with domestic and international academic institutions.