Spatiotemporal wavelet compression for visualization of scientific simulation data

S. LI; S. SANE; L. ORF; P.D. MININNI; J. CLYNE; H. CHILDS

Honolulu

Conferencia; 2017 IEEE International Conference on Cluster Computing (CLUSTER); 2017

IEEE

Data reduction through compression is emergingas a promising approach to ease I/O costs for simulation codeson supercomputers. Typically, this compression is achieved bytechniques that operate on individual time slices. However, assimulation codes advance in time, outputting multiple timeslices as they go, the opportunity for compression incorporatingthe time dimension has not been extensively explored. Moreover, recent supercomputers are increasingly equipped withdeeper memory hierarchies, including solid state drives andburst buffers, which creates the opportunity to temporarilystore multiple time slices and then apply compression tothem all at once, i.e., spatiotemporal compression. This paperexplores the benefits of incorporating the time dimension intoexisting wavelet compression, including studying its key parameters and demonstrating its benefits in three axes: storage,accuracy, and temporal resolution. Our results demonstratethat temporal compression can improve each of these axes,and that the impact on performance for real systems, includingtradeoffs in memory usage and execution time, is acceptable.We also demonstrate the benefits of spatiotemporal waveletcompression with real-world visualization use cases and tailored evaluation metrics.