Refine
Year of publication
Document Type
- In Proceedings (165)
- Article (135)
- ZIB-Report (62)
- Book chapter (16)
- Other (14)
- In Collection (9)
- Book (3)
- Poster (3)
Keywords
- volume rendering (3)
- 3D texture mapping (2)
- DVR (2)
- visualization (2)
- 3D neural network (1)
- AMR (1)
- AMR hierarchies (1)
- AMR tree (1)
- Amira (1)
- CCTK (1)
Institute
- Visual and Data-centric Computing (359)
- Visual Data Analysis (348)
- Visual Data Analysis in Science and Engineering (242)
- Therapy Planning (76)
- Image Analysis in Biology and Materials Science (66)
- ZIB Allgemein (31)
- Numerical Mathematics (22)
- Vergleichende Visualisierung (12)
- Computational Medicine (5)
- Geometric Data Analysis and Processing (5)
- Distributed Algorithms and Supercomputing (3)
- Computational Molecular Design (1)
- Computational Systems Biology (1)
- Mathematical Optimization (1)
- Mathematics of Transportation and Logistics (1)
The FORTRAN preprocessor fpp in the newly introduced Autotasking System of CRAY Research allows automatic vectorization and parallelization on basis of a data dependence analysis. An introduction into data dependence analysis is given, showing how data dependence graphs unveil opportunities for program transformations like vectorization and concurrentization. The report contains a complete description of the preprocessors functionality, its options and directives for increasing the effectiveness of the dependence analyzer and steering the code transformations. Finally, some advice is given for the practical use of fpp on CRAY computers.
Efficient implementations of irregular problems on vector and parallel architectures are generally hard to realize. An important class of problems are Gauß-Seidel iteration schemes applied to irregular data sets. The unstructured data dependences arising there prevent restructuring compilers from generating efficient code for vector or parallel machines. It is shown, how to structure the data dependences by decomposing the underlying data set using graph coloring techniques and by specifying a particular execution order already on the algorithm level. Methods to master the irregularities originating from different types of tasks are proposed. An application is given and some open issues and future developments are discussed.
In this paper various algorithms for rendering gaseous phenomena are reviewed. In computer graphics such algorithms are used to model natural scenes containing clouds, fog, flames and so on. On the other hand it has become an important technique in scientific visualization to display three dimensional scalar datasets as cloudy objects. Our emphasis is on this latter subject of so-called {\em direct volume rendering}. All algorithms will be discussed within the framework of linear transport theory. The equation of transfer is derived. This equation is suitable to describe the radiation field in a participating medium where absorption, emission, and scattering of light can occur. Almost all volume rendering algorithms can be shown to solve special cases of the equation of transfer. Related problems like the mapping from data values to model parameters or possible parallelization strategies will be discussed as well.
Sparse LU factorization offers some potential for parallelism, but at a level of very fine granularity. However, most current distributed memory MIMD architectures have too high communication latencies for exploiting all parallelism available. To cope with this, latencies must be avoided by coarsening the granularity and by message fusion. However, both techniques limit the concurrency, thereby reducing the scalability. In this paper, an implementation of a parallel LU decomposition algorithm for linear programming bases is presented for distributed memory parallel computers with noticable communication latencies. Several design decisions due to latencies, including data distribution and load balancing techniques, are discussed. An approximate performance model is set up for the algorithm, which allows to quantify the impact of latencies on its performance. Finally, experimental results for an Intel iPSC/860 parallel computer are reported and discussed.