Scalable scientific data

Question Hierarchial Z-Order Evaluation How can we present hundreds or thousands of gigabytes of scientific data to a user for analysis and interpretation? • The Scientific Computing and Imaging Institute is responsible for helping scientists visualize massive amounts of data. • Sources of large scientific data include medical imaging equipment (CAT, PET, MRI, etc.), fluid dynamics simulations, and genetic sequence mapping • Some of these simulations produce hundreds of gigabytes of data per simulation time step. Evaluating the speed of loading a set of random samples from an 8GB 3D image showed that: •Both Z and HZ-order significantly outperform the standard Row Major mode representation •HZ-order also outperforms Z-order for progressive requests Based on the Lebesque curve • Indexes Z-curve resolution levels in hierarchical order from coarser to finer. • Maintains the same geometric locality for each Z-curve resolution level • Beneficial for progressive resolution requests. (e.g. an "object search" application may first attempt to perform filtering on a coarser resolution)