Improved Performance of Large Data Visualization using Sub-Zone Load-On-Demand
Scott T. Imlay · 51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition · 2013
The size and number of datasets analyzed by post-processing and visualization tools is growing with Moore’s law. Conversely, the disk-read data transfer rate is only doubling every 36 months and is destined to be the bottleneck for traditional post-processing architectures. To eliminate this bottleneck, a sub-zone load-on-demand visualization architecture has been developed which only loads the data needed to create the desired plot. The original dataset is subdivided into sub-zones of <256 cells or nodes and these subzones are indexed on the disk using interval trees for each variable. Loading the required data starts with an ����� ���� query of the interval tree to determine which sub-zones should be loaded. The resulting visualization tool is faster and uses far less memory than the standard visualization package.