Campus Event Calendar
Event Entry
What and Who
Efficient ray-tracing on GPUs - Ph.D. application talk
Stefan Popov
IMPRS
Talk
AG 1, AG 2, AG 3, AG 4, AG 5
MPI Audience
Note: We use this to send email in the morning.
Date, Time and Location
Tuesday, 21 February 2006
11:00
15 Minutes
46.1 - MPII
0.24
Saarbrücken
Abstract
Recursive ray tracing is a simple and powerful technique for rendering
high quality images. Even though recent algorithms have reached real
time frame rates, ray tracing is still very computationally intensive.
The computational advantages of GPUs over CPUs and their
shading-oriented architecture makes GPUs an attractive platform for
real-time ray tracing. However, due to architectural limitations, no
algorithm has been published that efficiently can exploit the
computational advantages of today's GPUs for use in ray tracing.
This talk presents a new ray tracing algorithm based on the recursive
KD-tree traversal algorithm, expected to have the same performance
characteristics as the latter. The algorithm is iterative and does not
require a stack, thus making it suitable for GPU implementations.
Furthermore it is a one-pass scalable algorithm, able to fully exploit
multi-threaded GPU architectures as well as GPU texture caches. A
GPU-friendly algorithm that can partially replace mail-boxing will be
presented as well.
high quality images. Even though recent algorithms have reached real
time frame rates, ray tracing is still very computationally intensive.
The computational advantages of GPUs over CPUs and their
shading-oriented architecture makes GPUs an attractive platform for
real-time ray tracing. However, due to architectural limitations, no
algorithm has been published that efficiently can exploit the
computational advantages of today's GPUs for use in ray tracing.
This talk presents a new ray tracing algorithm based on the recursive
KD-tree traversal algorithm, expected to have the same performance
characteristics as the latter. The algorithm is iterative and does not
require a stack, thus making it suitable for GPU implementations.
Furthermore it is a one-pass scalable algorithm, able to fully exploit
multi-threaded GPU architectures as well as GPU texture caches. A
GPU-friendly algorithm that can partially replace mail-boxing will be
presented as well.
Contact
Kerstin Meyer-Ross
9325 226
--email hidden
passcode not visible
logged in users only
Friederike Gerndt, 02/14/2006 17:44 -- Created document.