EGPGV04: Eurographics Symposium on Parallel Graphics and Visualization

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/362

Browse

Now showing 1 - 14 of 14

Case Study of Multithreaded In-core Isosurface Extraction Algorithms
(The Eurographics Association, 2004) Zhang, Huijuan; Newman, Timothy S.; Zhang, Xiang; Dirk Bartz and Bruno Raffin and Han-Wei Shen
A comparative, empirical study of the computational performance of multithreading strategies for Marching Cubes isosurface extraction is presented. Several representative data-centric strategies are considered. Focus is on in-core computation that can be performed on desktop (single- or dual-CPU) computers. The study's empirical results are analyzed on the metrics of initialization overhead, individual surface extraction time, and total run time. In addition, an analysis of cache behavior and memory storage requirements is presented.
Fast Remote Isosurface Visualization With Chessboarding
(The Eurographics Association, 2004) Neeman, A.; Sulatycke, P.; Ghose, K.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
Isosurface rendering is a technique for viewing and understanding many large data sets from both science and engineering. With the advent of multi-gigabit-per-second network backbones such as Internet2 and fast local networking technology, scientists are looking at new ways to share and explore large data sets remotely. Telemedicine, which encompasses both videoconferencing and remote visualization is likely to be in widespread use with these advances. Despite the availability of increased bandwidth, two challenges remain. First, the time it takes to locate cells intersecting an isosurface of interest must be reduced for large data sets; a cell extraction technique that scales with data size is also critical. The second challenge has to do with the mitigating the effects of network latency on the overall isosurface visualization time. We present a remote isosurface visualization technique that addresses these two challenges. Isosurface extraction delays are reduced through the use of a search-optimized, chessboarded interval tree data structure on the disk. Network transport delays are reduced by sending cells extracted from the chessboarded data on the server, compressing it by about 87%. In addition, network transport delays are hidden effectively by overlapping data transport with server side functions. On a 100 Mbits/sec. switched LAN, the remote visualization time - the time between the issue of a query from the client side to the server and the displaying of a complete image on the client is only a few seconds for most isovalues in the well-known visible woman data set.
A hierarchical and view dependent visualization algorithm for tree based AMR data in 2D or 3D
(The Eurographics Association, 2004) Pino, Stéphane Del; Dirk Bartz and Bruno Raffin and Han-Wei Shen
In this paper, a solution to the visualization of huge amount of data provided by solvers using tree based AMR method is proposed. This approach strongly relies on the hierarchical structure of data and view dependent arguments: only the visible cells will be drawn, reducing consequently the amount of rendered data, selecting only the cells that intersect the screen and whose size is bigger than one pixel. After a brief statement of the problem, we recall the main principles of AMR methods.We then proceed to the data analysis which shows notable differences related to the dimension (2 or 3). A natural view dependent decimation algorithm is derived in the 2D case (only visible cells are plotted), while in 3D the treatment is not straightforward. The proposed solution relies then on the use of perspective in order to keep the same guidelines that were used in 2D. We then give a few hints about implementation and perform numerical experiments which confirm the efficiency of the proposed algorithms.We finally discuss this approach and give the sketch for future improvements.
Hierarchical Visualization and Compression of Large Volume Datasets Using GPU Clusters
(The Eurographics Association, 2004) Strengert, Magnus; Magallón, Marcelo; Weiskopf, Daniel; Guthe, Stefan; Ertl, Thomas; Dirk Bartz and Bruno Raffin and Han-Wei Shen
We describe a system for the texture-based direct volume visualization of large data sets on a PC cluster equipped with GPUs. The data is partitioned into volume bricks in object space, and the intermediate images are combined to a final picture in a sort-last approach. Hierarchical wavelet compression is applied to increase the effective size of volumes that can be handled. An adaptive rendering mechanism takes into account the viewing parameters and the properties of the data set to adjust the texture resolution and number of slices. We discuss the specific issues of this adaptive and hierarchical approach in the context of a distributed memory architecture and present solutions for these problems. Furthermore, our compositing scheme takes into account the footprints of volume bricks to minimize the costs for reading from framebuffer, network communication, and blending. A detailed performance analysis is provided and scaling characteristics of the parallel system are discussed. For example, our tests on a 16-node PC cluster show a rendering speed of 5 frames per second for a 2048x1024x1878 data set on a 10242 viewport.
I/O Strategies for Parallel Rendering of Large Time-Varying Volume Data
(The Eurographics Association, 2004) Yu, Hongfeng; Ma, Kwan-Liu; Welling, Joel; Dirk Bartz and Bruno Raffin and Han-Wei Shen
This paper presents I/O solutions for the visualization of time-varying volume data in a parallel and distributed computing environment. Depending on the number of rendering processors used, our I/O strategies help signifi- cantly lower interframe delay by employing a set of I/O processors coupled with MPI parallel I/O support. The targeted application is earthquake modeling using a large 3D unstructured mesh consisting of one hundred millions cells. Our test results on the HP/Compaq AlphaServer operated at the Pittsburgh Supercomputing Center demonstrate that the I/O strategies effectively remove the I/O bottlenecks commonly present in time-varying data visualization. This high-performance visualization solution we provide to the scientists allows them to explore their data in the temporal, spatial, and visualization domains at high resolution. This new high-resolution explorability, likely not presently available to most computational science groups, will help lead to many new insights.
Interactive Parallel Visualization of Large Particle Datasets
(The Eurographics Association, 2004) Liang, Kevin; Monger, Patricia; Couchman, Huge; Dirk Bartz and Bruno Raffin and Han-Wei Shen
This paper presents a new interactive parallel method for direct visualization of large particle datasets. Based on a parallel rendering cluster, a frame rate of 9 frames-per-second is achieved for 2563 particles using 7 render nodes and a display node. This provides real time interaction and interactive exploration of large datasets, which has been a challenge for scientific visualization and other real time data mining applications. The system allows scientists to study and to analyze the simulation results by viewing the particle cube from different perspectives, flying through the simulation field, or diving into the internal structure of the particles. A dynamic data distribution technique is designed for visualizing a highlighted subset of the particle volume. It maintains the load balance of the system and minimize the network traffic by updating the rendering pipeline through reconfiguration of the rendering chain. The method can be easily extended to other large datasets such as hydrodynamic turbulence, fluid dynamics, and so on.
Massive Data Pre-Processing with a Cluster Based Approach
(The Eurographics Association, 2004) Borgo, R.; Pascucci, V.; Scopigno, R.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
Data coming from complex simulation models reach easily dimensions much greater than available computational resources. Visualization of such data still represents the most intuitive and effective tool for scientific inspection of simulated phenomena. To ease this process several techniques have been adopted mainly concerning the use of hierarchical multi-resolution representations. In this paper we present the implementation of a hierarchical indexing schema for multiresolution data tailored to overwork the computational power of distributed environments.
Memory-Savvy Distributed Interactive Ray Tracing
(The Eurographics Association, 2004) DeMarle, David E.; Gribble, Christiaan P.; Parker, Steven G.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
Interactive ray tracing in a cluster environment requires paying close attention to the constraints of a loosely coupled distributed system. To render large scenes interactively, memory limits and network latency must be addressed efficiently. In this paper, we improve previous systems by moving to a page-based distributed shared memory layer, resulting in faster and easier access to a shared memory space. The technique is designed to take advantage of the large virtual memory space provided by 64-bit machines. We also examine task reuse through decentralized load balancing and primitive reorganization to complement the shared memory system. These techniques improve memory coherence and are valuable when physical memory is limited.
Parallel Implicit Integration for Cloth Animations on Distributed Memory Architectures
(The Eurographics Association, 2004) Keckeisen, M.; Blochinger, W.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
We present a parallel cloth simulation engine designed for distributed memory parallel architectures, in particular clusters built of commodity components. We focus on efficient parallel processing of irregularly structured and real-world sized problems typically occurring in the simulation of garments. We report on performance measurements showing a high degree of parallel efficiency and scalability indicating the usefulness of our approach.
Parallel Multiresolution Volume Rendering of Large Data Sets with Error-Guided Load Balancing
(The Eurographics Association, 2004) Wang, Chaoli; Gao, Jinzhu; Shen, Han-Wei; Dirk Bartz and Bruno Raffin and Han-Wei Shen
We present a new parallel multiresolution volume rendering algorithm for visualizing large data sets. Using the wavelet transform, the raw data is first converted into a multiresolution wavelet tree. To eliminate the parent-child data dependency for reconstruction and achieve load-balanced rendering, we design a novel algorithm to partition the tree and distribute the data along a hierarchical space-filling curve with error-guided bucketization. At run time, the wavelet tree is traversed according to the user-specified error tolerance, data blocks of different resolutions are decompressed and rendered to compose the final image in parallel. Experimental results showed that our algorithm can reduce the run-time communication cost to a minimum and ensure a well-balanced workload among processors when visualizing gigabytes of data with arbitrary error tolerances.
A Scalable Cluster-based Parallel Simplification Framework for Height Fields
(The Eurographics Association, 2004) Gouranton, V.; Limet, S.; Madougou, S.; Melin, E.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
In this paper, we present a method to interactively render 3D large datasets on a PC Cluster. Classical methods use simplification to fill up the gap between such models and graphics card capabilities. Unfortunatelly, simplification algorithms are time and memory consuming and they allow real-time interaction only for a restricted size of models. This work focuses on parallelizing Rottger's simplification algorithm for height fields but the main ideas can be generalized to other scientific areas. The method benefits from the scalable computating power of clusters. As our results show it, this permits us to achieve a data scaling while maintaining an acceptable frame rate with real-time interaction. Moreover, the scheme can take avantage of tiled-display environments.
Through the Concurrency Gateway: a Challenge from the Near Future of Graphics Hardware
(The Eurographics Association, 2004) Welch, P. H.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
The computer graphics industry, and in particular those involved with films, games and virtual reality, continue to demand more and more realistic computer generated images. The complexity of the scenes being modelled and the high fidelity required of the images means that rendering is simply not possible in a reasonable time (let alone real-time) on a single computer[BrW03]. Interactive ray tracing exists today[WSB*01], but real-time global illumination remains a major challenge. Fortunately, "computer graphics cards are developing at Moore's law cubed" [David Kirk, Chief Scientist, nVIDIA]. Such performance increases are directly due to the inherent parallel nature of modern graphics cards. If this trend continues, they will be 100 times faster in a mere 3.5 years time, 1000 times faster in 5 years and they will be massively parallel. Unfortunately, past experiences in designing systems that can exploit parallel processors in anything beyond embarrassingly trivial ways are not encouraging. For real-time interaction with high fidelity images, the parallel processing requirements will not be embarrassingly trivial! Regular and irregular patterns of synchronisation and communication will have to be managed over networks of fine-grained (for accuracy) model components whose scale, topology and physical distribution are dynamically evolving. This paper reviews weaknesses in our standard approaches to the design and implementation of concurrent systems and describes ways forward that are mature and practical - both for the programmer to program and the hardware to execute. They are built on decades of research into process algebrae (CSP and the p-calculus), but are able to preserve and exploit traditional skills and capabilities of serial software engineering and von Neumann architecture (components of which will still form the processor base of parallel systems for at least the next decade). The changes are, therefore, evolutionary rather than revolutionary - but are nevertheless essential both in the field of graphics and for the wider Grand Challenges of computer science.
Tuning of Algorithms for Independent Task Placement in the Context of Demand-Driven Parallel Ray Tracing
(The Eurographics Association, 2004) Plachetka, T.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
This paper investigates assignment strategies (load balancing algorithms) for process farms which solve the problem of online placement of a constant number of independent tasks with given, but unknown, time complexities onto a homogeneous network of processors with a given latency. Results for the chunking and factoring assignment strategies are summarised for a probabilistic model which models tasks' time complexities as realisations of a random variable with known mean and variance. Then a deterministic model is presented which requires the knowledge of the minimal and maximal tasks' complexities. While the goal in the probabilistic model is the minimisation of the expected makespan, the goal in the deterministic model is the minimisation of the worstcase makespan. We give a novel analysis of chunking and factoring for the deterministic model. In the context of demand-driven parallel ray tracing, tasks' time complexities are unfortunately unknown until the actual computation finishes. Therefore we propose automatic self-tuning procedures which estimate the missing information in run-time. We experimentally demonstrate for an "everyday ray tracing setting" that chunking does not perform much worse than factoring on up to 128 processors, if the parameters of these strategies are properly tuned. This may seem surprising. However, the experimentally measured efficiencies agree with our theoretical predictions.
Visualization of Seismic Wave Propagation from Recent Damaging Earthquakes in Japan: Dense Array Observations and Parallel Simulations Using the Earth Simulator
(The Eurographics Association, 2004) Furumura, T.; Chen, L.; Dirk Bartz and Bruno Raffin and Han-Wei Shen
This paper presents recent developments of large-scale parallel simulation of seismic wave propagation and concurrent visualization of 3D seismic wavefield using the Earth Simulator supercomputer. The developments of high-performance computing facilities and dense networks of strong ground motion instruments in Japan (K-NET and KiK-net) have now made it possible to directly visualize regional seismic wave propagation during large earthquakes. Our group has developed efficient parallel finite difference method (FDM) code for modeling the seismic wavefield, and a 3D visualization technique, both suitable for implementation on the Earth Simulator. Large-scale 3D simulations of seismic wave propagation have been conducted using these techniques to recast strong ground motions during recent damaging earthquakes such as the 2000 Tottori-ken Seibu (MJ7.3) earthquake, the 1923 great Kanto earthquake (MJ 7.9), and the 1855 Ansei Edo (M7) earthquake. Comparison of the simulation results with the dense array observations provides insights into the nature of complex seismic wave propagation through the heterogeneous subsurface structure in Japan. The simulation results are in very good agreement with the observations in terms of the features of the waveform and the regional intensity pattern, indicating that the simulation is already at a suitable level for use in investigating the expected pattern of ground motions for future earthquake scenarios.

Browse

Browsing EGPGV04: Eurographics Symposium on Parallel Graphics and Visualization by Title

Results Per Page

Sort Options