Fourth Eurographics Workshop on Parallel Graphics and Visualization (2002) D. Bartz, X. Pueyo, E. Reinhard (Editors) Interactive Ray Tracing of Time Varying Data Erik Reinhard, Charles Hansen and Steve Parker School of Computing, University of Utah Abstract We present a simple and effective algorithm for ray tracing iso-surfaces of time varying data sets. Each time step is partitioned into separate ranges of potentional iso-surface values. This creates a large number of relatively small files. Out-of-core rendering is implemented by reading for each time step the relevant iso-surface file, which con- tains its own spatial subdivision as well as the volumetric data. Since any of these data partitions is smaller than a single time step, the I/O bottleneck is overcome. Our method capitalizes on the ability of modern architectures to stream data off disk without interference of the operating system. Additionally, only a fraction of a time-step is held in memory at any moment during the visualization, which significantly reduces the required amount of internal memory. Categories and Subject Descriptors (according to ACM CCS): I.3.1 [Hardware Architectures]: Parallel Processing I.3.7 [Three-Dimensional Graphics and Realism]: Ray tracing 1. Introduction Researchers in many science and engineering fields rely on insight gained from instruments and simulations that pro- duce discrete samplings of three-dimensional scalar fields. Visualization methods allow for more efficient data anal- ysis to guide researchers. Iso-surface extraction is an im- portant technique for visualizing three-dimensional scalar fields by exposing contours of constant value 10 . These con- tours isolate surfaces of interest, focusing attention on im- portant features in the data such as material boundaries and shock waves while suppressing extraneous information. Sev- eral disciplines, including medicine 12 22 , computational fluid dynamics (CFD) 56 , and molecular dynamics 9 14 , have used this method effectively. Understanding the dynamic behavior of a data set re- quires the visualization of its changes with respect to time. However, most high performance computers possess nei- ther the disk space nor the amount of memory necessary to store and manipulate large time-varying data sets effi- ciently. While visualization research has begun to address this problem 2 3 21 20 17 24 , data sets from both computational and measurement sources have continued to increase in size, putting pressure on storage systems. Simulations that com- pute and store multiple time steps further increase the de- mand for storage space, commonly producing data sets on the order of one half to one gigabyte per time step with hun- dreds of time steps. With this vast amount of data to process, the interactive iso-surface visualization problem is impacted by consuming large amounts of time reading a multitude of huge files from disk and potentially performing swapping due to limited physical memory. Without a high degree of interactivity, the user loses the visual cues necessary to un- derstand the structure of the field, reducing the effectiveness of the visualization. We present an algorithm for the interactive visualization of iso-surfaces in time-varying fields that minimizes the im- pact of the I/O bottleneck. By preprocessing the data into ef- fective iso-contour ranges, the amount of data read is limited. By streaming the data from disk, the potential overall size of a time-varying simulation is bounded only by disk capacity, not by the I/O rate. Coupled with a parallel ray tracing en- gine, we achieve interactive time varying data visualization. In the following sections, we first discuss related work and then present our algorithm for streaming data for iso- surface visualization of time-varying fields. We then provide experimental results, demonstrating the performance of the algorithm on several large time-varying data sets. Finally, we draw conclusions and suggest directions for future work. c The Eurographics Association 2002.