Neumann, U. Communication costs for parallel volume-rendering algorithms. IEEE Computer Graphics and Applications 14, 4 (July 1994), 49-- 58.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....causes data redistribution between processing elements. The data redistribution, or remote memory fetch, when implemented carelessly in distributed environments, is one of the most serious factors that deteriorate the speedup of parallel volume rendering, especially when the data is very large [13]. In the rendering scheme, presented in this paper, on the other hand, the entire volume is stored in compressed form at local memory, and is locally reconstructed on the fly as necessary. Since no data communication is required between processors for data redistribution, our scheme produces ....

U. Neumann. Communication costs for parallel volumerendering algorithms. IEEE Computer Graphics and Applications, 14(4):49--58, July 1994.

.... Keywords: volume visualization, splatting, ray casting, taxonomy 1 References [98] 49] 16, 48, 113] 44] 52] 107] 111, 81] 94] 16, 15] 34] 101] 40] 66] 73] 90] 116] 26] 20] 68] 6] 54, 52, 63, 28, 67] 7, 1, 42, 4, 62] 31, 75, 30, 38, 86] 41, 21, 119, 112, 118, 45, 58] [55, 65, 14, 29, 79] [115, 53, 70, 74, 87, 19, 117, 32] 72, 57, 92, 36, 7] 111, 12, 90, 97, 78, 5] 15, 27, 35, 43, 87, 28, 51, 22, 23, 59, 60, 64, 63, 66, 68, 71, 80, 81, 91, 93, 99, 47, 108] 103, 116, 83, 106, 107, 105, 2, 76, 110, 77] 11, 104, 101, 96, 89, 26, 73, 6, 17, 52, 54, 85, 114, 9, 82, 37, 24] 8, 61, ....

U. Neumann. Communication costs for parallel volume-rendering algorithms. IEEE Comput. Graph. Appl. (USA), 14(4):49--58, July 1994.

....this threshold may prove to be data dependent. Increasing the number of texture maps may increase the threshold, while decreasing the number of texture maps will decrease the threshold. 3. 2 Distributed IBRAVR The IBRAVR model maps nicely to an object order decomposition for parallel rendering [2]. The primary difference between the IBRAVR method and traditional object order parallel software volume rendering lies in the design of the partial image recombination, or gather stage of the parallel rendering operation. The intermediate images produced by each processor, each of which renders a ....

....The intermediate images produced by each processor, each of which renders a subset of a volume, must be composited together in a specific order to produce a final image. Algorithms for the image recombination stage of parallel software volume rendering have been the subject of much study [2]. Our IBRAVR implementation uses a pool of processors that perform object order, parallel volume rendering in software. Rather than recombine the intermediate images in software, the partial images are combined using low cost graphics hardware that supports two dimensional texture mapping. By ....

Ulrich Neumann, Communication Costs for Parallel Volume-Rendering Algorithms, IEEE Computer Graphics and Applications, Volume 14, Number 4, pp 49-58, July 1994.

....structures evolve over time. For visualizing large data sets, parallel processing is often used to speed up the expensive volumetric rendering process. Although the subject of rendering a single volumetric data set using a parallel computer has been studied extensively by numerous researchers [17, 16, 14, 22, 10], parallel animation of TVVD, in contrast, has received relatively little attention. Compared to parallel volume rendering of a single data set, rendering TVVD in parallel poses a di#erent set of design tradeo#s. First, because TVVD typically consists of a sequence of data volumes, the I O ....

U. Neumann, Communication Costs for Parallel Volume-Rendering Algorithms, IEEE Computer Graphics and Applications, 14 (1994), pp. 49--58.

....Sort Last Algorithm In our fourth study, we compare our hybrid partitioning algorithm to a sort last approach. For the purposes of this experiment, we have implemented a polygon rendering version of a sort last algorithm motivated by Neumann s Object Partition with Block Distribution algorithm [22]. This algorithm is used for comparison because it requires the least composition bandwidth, provides the best balance, and fits into our peer to peer image composition system better than any other sort last algorithm we are aware of. In our implementation of Neumann s algorithm, groups of ....

Ulrich Neumann. Communication costs for parallel volumerendering algorithms. IEEE Computer Graphics and Applications, 14(4):49--58, 1994.

.... processing elements of the parallel system by either: Volume partitioning: Each processing element partially computes the tasks for a portion of the volume data; or Image partitioning Each processing element computes all the tasks associated with rays cast through a sub region of the image plane [7]: The volume partitioning strategy has the advantage that processing elements are able to complete their partial tasks without the need to exchange any of the volume voxels that they hold. If the data requirements are so large that they can not be accommodated within the combined local memories of ....

U. Neumann. Communication cost for parallel volume rendering algorithms. IEEE Computer Graphics and Applications, 49--58, July 1994.

....Initiative [1] will 13 employ on the order of 10 3 10 4 high performance commercial microprocessors, in relatively conventional architectural arrangements. To date, most parallel rendering implementations on similar architectures have not scaled well beyond 100 200 processors [6] 10] 11][13]. We conjecture that this scalability barrier arises from inherent properties of the rendering problem. First of all, typical rendering applications are limited to image resolutions of about 1 megapixel or less. Given that the computations performed at a single pixel represent a relatively ....

U. Neumann. Communication costs for parallel volume-rendering algorithms. IEEE Computer Graphics and Applications, 14(4), 49-58, July 1994.

....for efficiency, and this often causes data redistribution between processing elements. The data redistribution, or remote memory fetch, when implemented carelessly, is one of the most serious factors that deteriorate the speedup of parallel volume rendering, especially when the data is very large [11]. In our implementation, the whole CT dataset of the Visible Man is compressed, and is replicated at each processing element. Since the entire dataset that is necessary for generating image segments, is available at the local memory, no data communication is needed between processors for data ....

U. Neumann. Communication costs for parallel volume-rendering algorithms. IEEE Computer Graphics and Applications, 14(4):49--58, July 1994.

....threshold surfaces in real time. 2. SYSTEM OVERVIEW Our volume renderer is a modified ray casting system that is optimized for isosurfaces. Rays are stepped through the volume to locate surface intersections, and Phong shading is computed using an interpolated gradient as the surface normal [7]. This renderer, with the image cache and isomap extensions, simultaneously handles both volumes and polygons in a uniform framework that extracts information from previous frames. Rendering models based on integration through semitransparent materials or multiple isosurfaces can also be ....

Neumann U. Communication Costs for Parallel Volume Rendering Algorithms. IEEE Computer Graphics and Applications, Vol. 14, No. 4, pp. 49 - 58, July 94.

No context found.

Neumann, U. Communication costs for parallel volume-rendering algorithms. IEEE Computer Graphics and Applications 14, 4 (July 1994), 49-- 58.

No context found.

U Neumann. Communication costs for parallel volume rendering algorithms. IEEE Computer Graphics and Applications, 14(4):49--58, 1994.

No context found.

Neumann, U. Communication Costs for Parallel Volume-Rendering Algorithms. IEEE Computer Graphics and Applications , Vol. 14, No. 4, July 1994, 49-58.

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