Campus Event Calendar
Event Entry
What and Who
Parallel Grid Construction for Ray Tracing on GPUs
Javor Kalojanov
International Max Planck Research School for Computer Science - IMPRS-CS
PhD Application Talk
AG 1, AG 3, AG 4, AG 5, SWS, RG1, MMCI
Public Audience
English
Note: We use this to send email in the morning.
Date, Time and Location
Monday, 19 October 2009
09:00
240 Minutes
E1 4
024
Saarbrücken
Abstract
The recent introduction of the CUDA programming model made GPUs an attractive platform to implement ray tracing on. While early research showed that algorithms and data structures that were well suited for CPU implementations also work well on graphic cards, we research an alternative approach. Grids and their variations were considered unsuited for ray tracing on CPUs, but we show that both the acceleration structure construction and its traversal map well to the parallel architecture of modern GPUs. We propose several novel parallel construction algorithms and integrate them in a ray tracing framework to show that we can achieve performance that is competitive and in some cases superior to state-of-the-art approaches for ray tracing dynamic scenes. We also propose a lazy construction algorithm for hierarchical grids with performance that scales with the amount of visible primitives instead of the total number of primitives in the scene. We hope that our results give us a better understanding of the parallel hardware architecture of modern GPUs and in the future will help us find a ray tracing implementation that can provide performance allowing complex physically based illumination simulations in dynamic scenes in real time.
Contact
imprs
225
--email hidden
passcode not visible
logged in users only
Jennifer Gerling, 10/07/2009 15:24 -- Created document.