Thomas W. Crockett · 1996
In computer graphics, rendering is the process by which an abstract description of a scene is converted to an image. When the scene is complex, or when high-quality images or high frame rates are required, the rendering process becomes computationally demanding. To provide the necessary levels of performance, parallel computing techniques must be brought to bear. Although parallelism has been exploited in computer graphics since the early days of the field, its initial use was primarily in specialized applications. The VLSI revolution of the late 1970's and the advent of scalable parallel computers during the late 1980's changed this situation. Today, parallel hardware is routinely used in graphics workstations, and numerous software-based rendering systems have been developed for general-purpose parallel architectures. This article provides a broad introduction to the subject of parallel rendering, encompassing both hardware and software systems. The focus is on the underlying concept...
Cited by 0
Thomas W. Crockett · 1995
In computer graphics, rendering is the process by which an abstract description of a scene is converted to an image. When the scene is complex, or when high-quality images or high frame rates are required, the rendering process becomes computationally demanding. To provide the necessary levels of performance, parallel computing techniques must be brought to bear. Although parallelism has been exploited in computer graphics since the early days of the field, its initial use was primarily in specialized applications. The VLSI revolution of the late 1970's and the advent of scalable parallel computers during the late 1980's changed this situation. Today, parallel hardware is routinely used in graphics workstations, and numerous software-based rendering systems have been developed for general-purpose parallel architectures. This article provides a broad introduction to the subject of parallel rendering, encompassing both hardware and software systems. The focus is on the underlying concept...
Cited by 0
Crockett, Thomas W. · NASA STI Repository (National Aeronautics and Space Administration)
This article provides a broad introduction to the subject of parallel rendering, encompassing both hardware and software systems. The focus is on the underlying concepts and the issues which arise in the design of parallel rendering algorithms and systems. We examine the different types of parallelism and how they can be applied in rendering applications. Concepts from parallel computing, such as data decomposition, task granularity, scalability, and load balancing, are considered in relation to the rendering problem. We also explore concepts from computer graphics, such as coherence and projection, which have a significant impact on the structure of parallel rendering algorithms. Our survey covers a number of practical considerations as well, including the choice of architectural platform, communication and memory requirements, and the problem of image assembly and display. We illustrate the discussion with numerous examples from the parallel rendering literature, representing most of the principal rendering methods currently used in computer graphics.
Cited by 0Open access
Thomas W. Crockett · 1995
This article provides a broad introduction to the subject of parallel rendering, encompassing both hardware and software systems. The focus is on the underlying concepts and the issues which arise in the design of parallel rendering algorithms and systems. We examine the different types of parallelism and how they can be applied in rendering applications. Concepts from parallel computing, such as data decomposition, task granularity, scalability, and load balancing, are considered in relation to the rendering problem. We also explore concepts from computer graphics, such as coherence and projection, which have a significant impact on the structure of parallel rendering algorithms. Our survey covers a number of practical considerations as well, including the choice of architectural platform, communication and memory requirements, and the problem of image assembly and display. We illustrate the discussion with numerous examples from the parallel rendering literature, representing most of the principal rendering methods currently used in computer graphics.
Cited by 35
Chalmers Alan, Davis E. Timothy, Reinhard Erik · Bristol Research (University of Bristol) · 2002
Realistic computer graphics is an area of research which develops algorithms and methods to render images of artifical models of worlds as realistically as possible. Such algorithms are known for their unpredictable data accesses and their high computational complexity. Rendering a single high quality image make take several hours, or even days. Parallel processing offers the potential for solving such complex problems in reasonable times.
Cited by 0
2002
Cited by 24
Wai-sum Lin · 1999
Cited by 1Open access
Alan Chalmers, Timothy A. Davis, Toshi Kato et al. · 2002
Cited by 0
Michael Cox · 1995
This dissertation investigates which parallel algorithms and architectures are the best candidates likely to enable rates for renderers that support such features as high-level geometric primitives, tesselation for accurate representation of curved surfaces, and anti-aliasing by oversampling.
Parallel algorithms are distinguished by where in the graphics pipeline they redistribute data in order to sort from object-space to image-space: very early in the pipeline (sort-first), after the natural translation in the pipeline to image-space (sort-middle), and after pixels are calculated (sort-last).
The main modeling results give a tradeoff between sort-first, sort-middle, and sort-last. In particular, sort-first renderers that support high-level primitives and extensive oversampling are favored by low redistribution costs. However, sort-last renderers that support extensive oversampling require very substantial network bandwidth. Modeling shows that sort-middle is intermediate between sort-first and sort-last with respect to costs and benefits.
To determine whether these results apply in practice, I have modified RenderMan, a commercial package, to support sort-first and sort-last. It is found that sort-first redistribution costs are indeed very low, but that in contrast to previously published results, redundancy is an impediment to scalability. Load imbalance, however, is the more significant cause of inefficiency for moderately-sized machines. For sort-last, redistribution costs are indeed a problem if rates are to be achieved. In addition, the particular renderer studied is extremely vulnerable to synchronization costs when implemented as sort-last; however, this is a limitation of the underlying uniprocessor algorithm rather than a limitation of sort-last. Thus, sort-first is a good candidate for except for load imbalance, sort-last except for redistribution costs.
Combining theoretical and empirical findings, I propose and implement a hybrid algorithm that employs sort-last to balance load in sort-first. This hybrid is found to be effective at reducing load imbalance. Addressing sort-last redistribution costs, I propose and study two techniques to reduce network bandwidth requirements: the inclusion of local z-buffers, and the use of snooping on a shared bus in order to reduce the number of pixels that must be redistributed. These techniques are investigated by modeling and trace-driven simulations; it is shown that the first is effective at reducing network bandwidth requirements, the second is extremely effective.
This dissertation concludes with an identification of further problems that must be solved if real-time realistic rendering is to be realized. (Abstract shortened by UMI.)
Cited by 22
Michael Carter · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 1993
In this dissertation, the step-by-step development of a scalable parallel hierarchical radiosity renderer is documented. First, a new look is taken at the traditional radiosity equation, and a new form is presented in which the matrix of linear system coefficients is transformed into a symmetric matrix, thereby simplifying the problem and enabling a new solution technique to be applied. Next, the state-of-the-art hierarchical radiosity methods are examined for their suitability to parallel implementation, and scalability. Significant enhancements are also discovered which both improve their theoretical foundations and improve the images they generate. The resultant hierarchical radiosity algorithm is then examined for sources of parallelism, and for an architectural mapping. Several architectural mappings are discussed. A few key algorithmic changes are suggested during the process of making the algorithm parallel. Next, the performance, efficiency, and scalability of the algorithm are analyzed. The dissertation closes with a discussion of several ideas which have the potential to further enhance the hierarchical radiosity method, or provide an entirely new forum for the application of hierarchical methods.
Cited by 9Open access
Paulo Santos, Alan Chalmers · 2003
Interactive global illumination at interactive rates is still an unachieved goal for medium to large complexity scenes and high fldelity illumination models, such as the one used by Radiance. In order to achieve this goal several difierent optimization approaches must be exploited, including parallelism, perceptual issues and low{level optimizations. This work concentrates on developing a parallel message passing prototype of Radiance’s renderer targeted towards producing animations. This prototype is used to study scalability issues and perceptual optimizations that will reduce rendering times and will, hopefully, be unnoticed by the viewer. The preliminary results obtained suggest that efiorts should concentrate on reducing idle times due to suboptimal load distribution and indirect irradiance calculation times. Also psychophysical experiments and perceptual metrics should be carried on, in order to assess the impact of selective rendering on perceived quality.
Cited by 1
Binary-swap Compositing, Kwan-liu Ma, James S. Painter et al. · 2014
This algorithm distributes data and computations to individual processing nodes for rendering subimages, then composites the final image with a method that uses all nodes at all times. E xisting volume rendering methods, though capable of very effective visualizations, are computationally intensive and therefore fail to achieve interactive rendering rates for large data sets. Although computing technology continues to advance, computer processing power never seems to catch up to the increases in data size. Several observations motivated the work we describe here.
Cited by 0
Luís Paulo Santos, Dep Informática, Alan Chalmers · 2003
Interactive global illumination at interactive rates is still an unachieved goal for medium to large complexity scenes and high fidelity illumination models, such as the one used by Radiance. In order to achieve this goal several different optimization approaches must be exploited, including parallelism, perceptual issues and low–level optimizations. This work concentrates on developing a parallel message passing prototype of Radiance’s renderer targeted towards producing animations. This prototype is used to study scalability issues and perceptual optimizations that will reduce rendering times and will, hopefully, be unnoticed by the viewer. The preliminary results obtained suggest that efforts should concentrate on reducing idle times due to suboptimal load distribution and indirect irradiance calculation times. Also psychophysical experiments and perceptual metrics should be carried on, in order to assess the impact of selective rendering on perceived quality. 1
Cited by 0
Michael Brannon Carter · 1993
Cited by 2
Datta, S., Guha, R. · Journal of International Crisis and Risk Communication Research · 1995
In this parallel surface rendering algorithm based on dividing cube2 on a SIMD machine MasPar MP-1, we address the problem of load balancing and image composition. We divide a 3D array of Nx × Ny × Nz volume data into Nx × Ny columns, each Nz deep. Each processor in the mesh receives a subvolume of such data columns. All processors synchronously traverse its subvolume to determine the voxels intersecting the isosurface. Intersecting voxels are called isovoxels. Partial load balancing distributes the isovoxels contained in a row of processors evenly among the processors in that row to reduce the network traffic and complexity of the rendering phase. Each isovoxel is subdivided into point primitives using dividing cube algorithm. Rendering algorithm transforms the surface points and their normals and projects them into the view plane.
Cited by 0Open access
Charles Hansen, M. Krogh, James Painter · University of North Texas Digital Library (University of North Texas) · 1995
As the resolution of simulation models increases, scientific visualization algorithms which take advantage of the large memory. and parallelism of Massively Parallel Processors (MPPs) are becoming increasingly important. For large applications rendering on the MPP tends to be preferable to rendering on a graphics workstation due to the MPP`s abundant resources: memory, disk, and numerous processors. The challenge becomes developing algorithms that can exploit these resources while minimizing overhead, typically communication costs. This paper will describe recent efforts in parallel rendering for polygonal primitives as well as parallel volumetric techniques. This paper presents rendering algorithms, developed for massively parallel processors (MPPs), for polygonal, spheres, and volumetric data. The polygon algorithm uses a data parallel approach whereas the sphere and volume render use a MIMD approach. Implementations for these algorithms are presented for the Thinking Ma.chines Corporation CM-5 MPP.
Cited by 0Open access
Thomas W. Crockett, Charles Hansen, Scott Whitman et al. · 1993
Cited by 16
Matthew Nicholas Pappas, D. Brookshire Conner, Oren J. Tversky et al. · 1992
this paper, we discuss the design and implementation of some three-dimensional rendering algorithms on massively parallel computers. These algorithms employ SIMD-style parallelism. That is, each processor runs the same instructions on different data. We have implemented some of these algorithms on the Connection Machine 2, a computer manufactured by the Thinking Machines Corporation. This paper should be accessible to people interested in computer graphics or massively parallel algorithms. We have taken great care to introduce both fields sufficiently that this paper should be quite self-contained. As a result, this paper should be readable by anyone familiar with some elementary theoretical computer science. This paper is intended both for the graphics community and for the parallel processing community. In view of this, we have felt it necessary to review certain aspects of both areas, resulting in a lengthy report. The reader who finds our explanations of either area too terse should refer to the references for further reading
Cited by 0
Thomas W. Crockett · Parallel Computing · 1997
Cited by 69
Matthew Eldridge · DSpace@MIT (Massachusetts Institute of Technology) · 1995
on May 12, 1995, in partial ful llment of the requirements for the degrees of Bachelor of Science and Master of Science This thesis describes the design and implementation of a polygonal rendering system for a large one dimensional single instruction multiple data (SIMD) array of processors. Polygonal rendering enjoys a traditionally high level of available parallelism, but e cient realization of this parallelism requires communication. On large systems, such as the 1024{processor Princeton Engine studied here, communication overhead limits performance. Special attention is paid to study of the analytical and experimental scalability of the design alternatives. Sort{last communication is determined to be the only design to o er linear performance improvements in the number of processors. We
Cited by 0Open access