2011
DOI: 10.1007/s10766-011-0186-1
|View full text |Cite
|
Sign up to set email alerts
|

Understanding the Efficiency of kD-tree Ray-Traversal Techniques over a GPGPU Architecture

Abstract: Current GPU computational power enables the execution of complex and parallel algorithms, such as ray tracing techniques supported by kD-trees for 3D scene rendering in real time. This work describes in detail the study and implementation of eight different kD-tree traversal algorithms using the parallel framework NVIDIA Compute Unified Device Architecture, in order to point their pros and cons regarding performance, memory consumption, branch divergencies and scalability on multiple GPUs. In addition, two new… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
12
0

Year Published

2012
2012
2016
2016

Publication Types

Select...
3
3
2

Relationship

1
7

Authors

Journals

citations
Cited by 17 publications
(12 citation statements)
references
References 16 publications
0
12
0
Order By: Relevance
“…So instead of following a comprehensive search to identify the correct scene for the nearest intersection, only nearby objects are approved and the remaining are discarded, as recommended by Siu-Lung et al [26].…”
Section: Subdivision Surface Methods On Mobile Devicementioning
confidence: 99%
See 1 more Smart Citation
“…So instead of following a comprehensive search to identify the correct scene for the nearest intersection, only nearby objects are approved and the remaining are discarded, as recommended by Siu-Lung et al [26].…”
Section: Subdivision Surface Methods On Mobile Devicementioning
confidence: 99%
“…Another algorithm commonly used for rendering is Ray-tracing [15,16,17,25,26,27,28]. This is based on the illumination of the image, capturing the beam size and its reflection on the object.…”
Section: Related Workmentioning
confidence: 99%
“…Due to the limitation of the registers and shared memory, the stack implemented in the local memory is 600 times slower than registers do. Experiments in [36] show that the size of stack element has great impacts on the performance of GPU implementation because GPU has no valid cache system. Using a larger element size, stack can significantly increase the stack access penalty on GPU (Fig.…”
Section: A Preprocessingmentioning
confidence: 99%
“…To solve the problem of ray tracing performance for interactive applications, Santos et al [3] describe in detail the study, comparison and hardware implementation of many kD-Tree traversal algorithms. They proposed two new algorithms based on their analysis, aiming performance improvement.…”
Section: B Real Time Ray Tracingmentioning
confidence: 99%
“…However, since GPU pipelines became highly programmable with recent architectures such as NVIDIA CUDA [2], they are used as a coprocessor for executing highly parallel algorithms, including ray tracing in real time [3][4] [5]. This work exploits this architecture to implement our ray tracing pipeline for application on augmented reality, named RT 2 AR.…”
Section: Introductionmentioning
confidence: 99%