M. Cox and N. Bhandari. Architectural Implications of Hardware-accelerated Bucket Rendering on the PC. In Proceedings of the Eurographics workshop on Graphics hardware, pages 25--34, 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... of the triangle s bounding box with the grid of miltchen,gws,homan,kekoa,hanrahan graphics.stanford.edu tiles [1] 3] 4] 10] 11] Molnar [8] presents an equation for the expected overlap of rectangular bounding boxes on rectangular tiles that is experimentally verified in both Molnar [9] and Cox [2]. Sorting triangles into tiles exactly, rather than by boundingbox, is certainly possible and would result in lower overlap factors, but we are not aware of an analysis of the costs and benefits of such a technique or of any systems that use it. Previously, the impact of bucket rendering has been ....

....into tiles exactly, rather than by boundingbox, is certainly possible and would result in lower overlap factors, but we are not aware of an analysis of the costs and benefits of such a technique or of any systems that use it. Previously, the impact of bucket rendering has been analyzed by Cox [2] on a specific PC based bucket rendering architecture. The paper concludes that for small tiles (32x32 pixels) the redundant storage and transfer of triangles due to high overlap factors can overwhelm current PC memory and I O systems. In that work, the impact of triangle overlap is analyzed as ....

[Article contains additional citation context not shown here]

M. Cox, N. Bhandari. Architectural Implications of HardwareAccelerated Bucket Rendering on the PC, 1997 Siggraph /Eurographics Workshop on Graphics Hardware, pp. 25-34.

....design interrelate, we examine more closely how rasterizers work. While there are many possible ways to construct a polygon rasterizer, two common approaches stand out. One is the conventional framebuffer based architecture, and the other is the tile based (or region based) architecture. [FOLE90, TORB96, COX97] With a conventional rasterizer architecture, a full screen framebuffer is used to store all of the necessary pixel information. Minimally, this includes at least two color buffers and a Z buffer. When a particular primitive is being rendered, the rasterizer computes which pixels will be ....

M. Cox and N. Bhandari, "Architectural Implications of HardwareAccelerated Bucket Rendering on the PC," 1997 Siggraph/Eurographics Workshop on Graphics Hardware, Los Angeles, CA, August 1997.

....the number of rendering servers and k is the replication factor. We have shown that k = 4 works quite well. However, data visualization applications may require an out of core method to store and traverse their databases. Previous studies have focused on out of core techniques for a single machine [5, 8]. We believe that k way replication principles can apply to out of core database management for multiple computers. For instance, it may be beneficial to store multiple copies of each data element on different disks in order to provide flexibility regarding which computers load and process data. ....

Michael Cox and Narendra Bhandari. Architectural implications of hardware-accelerated bucket rendering on the pc. In Eurographics/SIGGRAPH workshop on Graphics hardware, pages 25--34, Los Angeles, CA, 1997.

....the number of scanned pixels (pixel bandwidth) If a triangle is sent to more than one processor, the overall parallel machine does more edge slope calculations and the triangle bandwidth may more often be reached. Analytical models and simulations have been proposed to evaluate precisely its cost [18, 3, 4]. In this paper, we suppose the processors able to do clipping while drawing and that they only draw pixels that belong to their image tile or image line. This clipping requires that the distribution is static and that the distribution scheme and its parameters are hard coded in the chip. The ....

....filtering like Igehy in [13] 3.2 Parallel Architecture Our parallel engine is described is figure 4. Each node is made of a 3D engine, its cache and a private texture memory. The nodes are connected by a network that forwards the triangles from the geometry engines. As Bucket Rendering [3, 4] is not used here, the primitive order is preserved. The problem of the geometry performance and the issues involved in the geometry and image networks have been widely studied in [18, 1, 20, 19] We want to focus on new issues involving only the texture cache so we decided Video Triangles ....

M. Cox and N. Bhandari. Architectural implications of hardware-acceleratedbucket rendering on the PC. In S. Molnar and B.-O. Schneider, editors, 1997 SIGGRAPH / Eurographics Workshop on Graphics Hardware, pages 25--34, New York City, NY, Aug. 1997. ACM SIGGRAPH / Eurographics, ACM Press. ISBN 0-89791-961-0.

....architectures that can change the connectivity during rendering. First one is clipping: a stage in the rendering pipeline that removes triangles outside the view frustum (see Section 5.2) and may add new triangles along the view frustum. The other one is sorting for bucket rendering architecture [Cox and Bhandari 97, Chen et al. 98] This architecture partitions the frame bu er into multiple regions, and renders the partitions one at a time so as to reduce the frame bu er memory requirement. A view dependent sorting stage is added to the rendering pipeline that partitions a triangle mesh into multiple ....

....and sorting modi es the mesh connectivity, the original BFT mesh cannot be used in the rest of the rendering pipeline. In fact, to the best of our knowledge, the existing graphics systems convert even a triangle strip representation to independent triangles after clipping and bucket sorting [Cox and Bhandari 97, Chen et al. 98] This chapter presents techniques to create BFT encoding of the modi ed triangle meshes on the y, so that the rest of the rendering pipeline can still take advantage of BFT mesh representation. 6.2 Clipping of Compressed Mesh Recall the integrated decompression and rendering ....

[Article contains additional citation context not shown here]

Michael Cox and Narendra Bhandari. Architectural Implications of Hardware-Accelerated Bucket Rendering on the PC. 1997 SIGGRAPH / Eurographics Workshop on Graphics Hardware, pages 25-34, August 1997.

....to be less quantitative than qualitative, making it impossible to empirically compare various architectural ideas. In particular, graphics hardware papers that discuss architectural tradeoffs based on concrete measurements from real world applications only start to appear in the last two years [14, 5]. While it is difficult to solve the first problem, we believe that the second problem can be easily addressed by emulating the experimental research methodology as practiced by the CPU architecture research community for years: reporting detailed quantitative statistics from either prototype ....

M. Cox and N. Bhandari. Architectural Implications of Hardware-Accelerated Bucket Rendering on the PC. In Proceedings of ACM SIGGRAPH/Eurographics Workshop on Graphics Hardware, 1997.

....Second, most of the earlier research on 3D graphics hardware tends to be less quantitative than qualitative. In particular, graphics hardware papers that discuss architectural tradeo s based on concrete measurements from real world applications only start to appear in the last two years [HG97, CB97, CBS98, C 98] As a result, it has been impossible to empirically compare various graphics architectural ideas on an unbiased basis, and thus advance the eld by drawing on lessons distilled from the comparisons. While it is di cult to solve the rst problem, we believe the graphics hardware ....

M. Cox and N. Bhandari. Architectural Implications of Hardware-Accelerated Bucket Rendering on the PC. In Proceedings of ACM SIGGRAPH/Eurographics Workshop on Graphics Hardware, 1997.

....groups of spatially co located primitives. This paper builds upon a long history of prior work on screenspace space partitioning methods for parallel rendering [6, 38] At the highest level, approaches can be classified based on whether the decomposition of screen space into tiles is static [18, 26, 10, 23, 8] or whether the tile boundaries are determined dynamically according to the distribution of graphics primitives on the screen [37, 32, 34, 24] If there are more tiles than processors, then tiles can be assigned to processors either statically or dynamically. So far, tile shapes have been based on ....

....graphics primitives on the screen [37, 32, 34, 24] If there are more tiles than processors, then tiles can be assigned to processors either statically or dynamically. So far, tile shapes have been based on scan lines [15] horizontal strips [18, 37, 5] vertical strips [37] and rectangular areas [18, 34, 32, 39, 8, 4]. Most systems define several times more tiles than there are processors so that the rendering load can be balanced. Yet, this approach leads to inefficiencies due to high primitive tile overlap factors and the loss of spatial coherence across tile boundaries [8] Several studies have investigated ....

[Article contains additional citation context not shown here]

Michael Cox, Architectural Implications of Hardware-Accelerated Bucket Rendering on the PC, 1997 SIGGRAPH/Eurographics Workshop on Graphics Hardware, Los Angeles, CA, 1997, 25-34.

....tile overlaps is an important step in many high performance polygon rendering systems utilizing image space parallelism or bucket rendering. Different tile shapes have been used in these systems, including scan lines [14] horizontal strips [3, 20, 38] vertical strips [38] and rectangular areas [1, 6, 20, 33, 40]. Tiles have been kept static [1, 6, 7, 20, 27, 31] or dynamically adjusted based on the distribution of graphics primitives on the screen [28, 33, 38] For each graphics primitive, the rendering system must determine which 2D screen space tiles it overlaps so that it can invoke rendering ....

....polygon rendering systems utilizing image space parallelism or bucket rendering. Different tile shapes have been used in these systems, including scan lines [14] horizontal strips [3, 20, 38] vertical strips [38] and rectangular areas [1, 6, 20, 33, 40] Tiles have been kept static [1, 6, 7, 20, 27, 31] or dynamically adjusted based on the distribution of graphics primitives on the screen [28, 33, 38] For each graphics primitive, the rendering system must determine which 2D screen space tiles it overlaps so that it can invoke rendering operations on the appropriate processors. Most current ....

[Article contains additional citation context not shown here]

Michael Cox, Architectural Implications of HardwareAccelerated Bucket Rendering on the PC, 1997 SIGGRAPH /Eurographics Workshop on Graphics Hardware, Los Angeles, CA, 1997, 25-34.

No context found.

M. Cox and N. Bhandari. Architectural Implications of Hardware-accelerated Bucket Rendering on the PC. In Proceedings of the Eurographics workshop on Graphics hardware, pages 25--34, 1997.

No context found.

M. Cox and N. Bhandari. Architectural Implications of Hardware-accelerated Bucket Rendering on the PC. In Proceedings of the Eurographics workshop on Graphics hardware, pages 25--34, 1997.

No context found.

Michael Cox and Narendra Bhandari. Architectural implications of hardware-accelerated bucket rendering on the PC. In Proceedings of the 1997.

No context found.

Michael Cox and Narendra Bhandari. Architectural implications of hardware-accelerated bucket rendering on the PC. In Proceedings of the 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC