Adding Interlacing to In-Situ Data Compression for Multi-Resolution Visualization of Large-Scale Scientific Simulations

Abstract

Large-scale simulations in HPC environments produce massive data sets in the order of several terabytes which pose challenges both to data storage and data analysis. A promising approach for reducing the amount of data written out during simulation runs is in-situ compression. However, even compressed data sets are typically still much too large for interactive visual data exploration. We propose an interlacing scheme applicable to in-situ compression in order to write out results of HPC simulations in a compressed multi-resolution format. This data format allows progressive loading of large scientific data sets in high-end immersive displays. Low-resolution views are easily extracted for visualizations on low-end displays such as hand-held devices and web browsers.