No description available.

No description available.

This paper demonstrates an efficient text compressor with parallel Lempel-Ziv-Markov chain algorithm (LZMA) on graphics processing units (GPUs). We divide LZMA into two parts, match finder and range encoder. We parallel both parts and achieve competitive performance with freeArc on AMD 6-core 2.81 GHz CPU. We measure match finder time, range encoder compression time and demonstrate realtime performance on a large dataset: 10 GB web pages crawled by IRLbot. Our parallel range encoder is 15 times faster than sequential algorithm (FastAC) with static model.

The contents of this thesis are concerned with the implementation of parallel algorithms for solving partial differential equations (POEs) by the Alternative Group EXplicit (AGE) method and an investigation into the numerical inversion of the Laplace transform on the Balance 8000 MIMO system. Parallel computer architectures are introduced with different types of existing parallel computers including the Data-Flow computer and VLSI technology which are described from both the hardware and implementation points of view. The main characteristics of the Sequent parallel computer system at Loughborough University is presented, and performance indicators, i.e., the speed-up and efficiency factors are defined for the measurement of parallelism in the system. Basic ideas of programming such computers are also outlined.....

No description available.

No description available.

No description available.

No description available.

Clusters of commodity, "off the shelf" workstations have given developers and users access to scalable and affordable computing resources. However, unlike large, symmetric multi-processing machines, these clusters have an up front cost in complexity, both for the developer and the user. Existing software frameworks have attempted to mitigate this complexity with varied success. In most frameworks, the user is forgotten and left to deal with an unwieldy application. This thesis presents the design, development and testing of a new C++ computer programming library, Scopira Agents Library (SAL). SAL is a message passing interface and implementation suitable for building parallel applications, with a focus on developer ease of use and user application deployment specification. The target developers and users of such a solution would be those who prefer an easy to develop library, with simpler deployment and application integration options with acceptable sacrifices to performance and scalability. The novelty of this parallel programming library is that it is more user-friendly than other existing libraries. This novelty has two major facets: (i) programmer-usability and productivity and (ii) application integration. Together, they permit a wider range of programmers to utilize parallel programming in a wider range of new and existing applications. This goal, user-friendliness, is rare among current parallel programming libraries. The result of the novelty is that parallel programming can be embedded into more applications, especially desktop applications. The user base and use cases for parallel applications can be increased, resulting in more efficient use of resources in a variety of applications. With increased efficiency, work can be performed in less time and larger problems can be tackled.

At the heart of many power system computations lies the solution of a large sparse set of linear equations. These equations arise from the modelling of the network and are the cause of a computational bottleneck in power system analysis applications. Efficient sequential techniques have been developed to solve these equations but the solution is still too slow for applications such as real-time dynamic simulation and on-line security analysis. Parallel computing techniques have been explored in the attempt to find faster solutions but the methods developed to date have not efficiently exploited the full power of parallel processing. This thesis considers the solution of the linear network equations encountered in power system computations. Based on the insight provided by the elimination tree, it is proposed that a novel matrix structure is adopted to allow the exploitation of parallelism which exists within the cutset of a typical parallel solution. Using this matrix structure it is possible to reduce the size of the sequential part of the problem and to increase the speed and efficiency of typical LU-based parallel solution. A method for transforming the admittance matrix into the required form is presented along with network partitioning and load balancing techniques. Sequential solution techniques are considered and existing parallel methods are surveyed to determine their strengths and weaknesses. Combining the benefits of existing solutions with the new matrix structure allows an improved LU-based parallel solution to be derived. A simulation of the improved LU solution is used to show the improvements in performance over a standard LU-based solution that result from the adoption of the new techniques. The results of a multiprocessor implementation of the method are presented and the new method is shown to have a better performance than existing methods for distributed memory multiprocessors.