|
Abstract:
Advances in computing technology, particularly in the field of microprocessor
design, have enabled natural sciences CFD simulations to be carried out at
unprecedented scales, yielding torrents of numerical data. However, for many
CFD researchers the greatest challenge now faced is not performing the
simulation, but extracting meaningful information from the outputs that
result. This talk will describe our approach to addressing this problem by
exploiting the power of progressive data access (PDA). Our novel,
wavelet-based PDA scheme enables the exploration and manipulation of
terascale size data sets using only a commodity, desktop PC. We have created
a prototype analysis system that combines advanced 3D visualization with
traditional statistical analysis capabilities to create a powerful
environment for investigating large, CFD data sets. With NSF ITR funding we
will develop our prototype into a robust, open source tool suite, freely
available to the scientific community.
|