Visualizing Particle-Based Simulation Datasets on the Desktop Christiaan P. Gribble, Abraham J. Stephens, James E. Guilkey, and Steven G. Parker Scientific Computing and Imaging Institute, University of Utah 50 S Central Campus Drive, MEB 3490 Salt Lake City, Utah 84105 United States of America cgribble@sci.utah.edu, abe@sci.utah.edu, james.guilkey@utah.edu, sparker@sci.utah.edu We present an approach to rendering large, time-varying particle-based simulation datasets using programmable graphics hardware on desktop computer systems. Particle methods are used to model a wide range of complex phenomena, and effective visualization of the resulting data requires communicating subtle changes in the three-dimensional structure, spatial organization, and qualitative trends within a simulation as it evolves, as well as allowing easier navigation and exploration of the data through interactivity. We highlight the critical components of our approach, and introduce an extension to the coherent hierarchical culling algorithm that often improves temporal coherence and leads to better average performance for time-varying datasets. Our approach performs competitively with current particle visualization systems based on interactive ray tracing that require tightly coupled supercomputers. Moreover, our system runs on hardware that is a fraction of the cost of these systems, making particle visualization and data exploration more accessible. We thus advance the current state-of-the-art by bringing visualization of particle-based simulation datasets to the desktop. Key words: particle-based simulation, particle visualization, interactive rendering, occlusion algorithms 1. INTRODUCTION Over the past several years, commodity graphics processing units (GPUs) have become an attractive option for visualizing a wide variety of scientific datasets. This hardware has evolved quite rapidly from a fixed-functionality pipeline into a programmable machine with powerful vertex and fragment processors. Moreover, extreme parallelism and the advent of high-level programming languages for this hardware have led to flexible processors providing compute power in the multi-GFLOPS to TFLOPS range [1]. We investigate the use of programmable GPUs to interactively visualize the results of particle-based simulations on desktop computer systems. Particle methods are commonly used to simulate complex phenomena in many scientific domains, including astronomy, biology, chemistry, physics, and others. Using particle-based simulation techniques, computational scientists model such phenomena as a system of discrete particles that obey certain laws and possess certain properties. These methods are particularly attractive because they can be used to solve time-dependent problems on scales from the atomic to the cosmological. Frequently, millions of particles are required to capture the behavior of a system accurately, leading to very large, very complex datasets. Investigators use particle visualization to assist efforts in data analysis and feature detection, as well as in debugging ill-behaved solutions. As a result, effective visualization of particle-based simulation data requires communication of subtle changes in the three-dimensional structure, spatial organization, and qualitative trends within a simulation as it evolves. Most importantly, an effective visualization method will enable investigators to interrogate their data interactively. Unfortunately, the size and complexity of typical particle datasets make interactive visualization a difficult task. Recently, Bigler et al. [2] have described an approach that is based on interactive ray tracing and requires tightly coupled supercomputing platforms. This system represents the current state-of-the-art in particle-based simulation data visualization, and while the approach satisfies the requirements of effective particle visualization, the hardware costs are prohibitive and thus impede accessibility. British HCI 2006 Workshop on Combining Visualization and Interaction to Facilitate Scientific Exploration and Discovery (September 2006) (to appear)