Software architecture based on remote components was a component based framework to rearrange the invoking order of local function and remote component for parallel running. This paper presents a construct concurrent ...
详细信息
ISBN:
(纸本)0780385608
Software architecture based on remote components was a component based framework to rearrange the invoking order of local function and remote component for parallel running. This paper presents a construct concurrent code algorithm for software architectures based on remote components. Using the formal description of the algorithm, it impacts upon the system performance are studied in detail. The experiment results showed that algorithm scheme method that can greatly improve system performance. The concurrent algorithm provides a cooperating method of local framework and remote service for distributed software architecture.
We propose a general-purposed parallel algorithm for volume segmentation, which does not require any prior knowledge on volume nor region. The algorithm provides binary tree structured split-and-merge mechanism to sea...
详细信息
We propose a general-purposed parallel algorithm for volume segmentation, which does not require any prior knowledge on volume nor region. The algorithm provides binary tree structured split-and-merge mechanism to search and localize boundaries along discontinuities and adapts the partition of volume to those detected discontinuities. This algorithm is independent from order of processing or seed selection. And, even though overlapping only one voxel wide boundary between process blocks, by adopting the smoothness-based local feature as homogeneity criteria, consistencies are maintained without overhead of communication between adjacent process blocks. Our efficient hierarchical step-wised mechanism in merging target evaluation makes merge process so simple and efficient that only two brother blocks are considered at each merge step in binary fashion. Experimental results on an artificial and a CT scan volume data are shown.
Summary form only given. This paper describes a parallel solution of the sequential dynamic programming method for solving a NP class, 2D knapsack (or cutting-stock) problem which is the optimal packing of multiples o...
详细信息
Summary form only given. This paper describes a parallel solution of the sequential dynamic programming method for solving a NP class, 2D knapsack (or cutting-stock) problem which is the optimal packing of multiples of n rectangular objects into a knapsack of size L/spl times/W and are only obtainable with guillotine-type (side to side) cuts. Here, we describe and analyze this problem for the associative model. Since the introduction of associative SIMD computers over a quarter of a century ago, associative computing and the data-parallel paradigm remain popular. The MASC (multiple instruction stream associative computer) parallel model supports a generalized version of an associative style of computing. This model supports data parallelism, constant time maximum and minimum operations, one or more instruction streams (ISs) which are sent to an equal number of partition sets of processors, and assignment of tasks to ISs using control parallelism. We solve this NP class problem with a parallel algorithm that runs in O(W(n+L+W)) time using L processors, where L>W for a 2D knapsack problem with a capacity of L/spl times/W. The new multiple IS version using LW processors and max{L,M} ISs runs in O(n+L+W) given practical hardware considerations. Both of these results are cost optimal with respect to the best sequential implementation. Moreover, an efficient MASC algorithm for this well-known problem should give insight to how the associative model compares to other parallel models such as PRAM.
Due to some drawbacks, such as the instability of the analog receiver in wideband conditions, an all-digital parallel receiver (APRX) has become more and more attractive. Based on the structure of an APRX proposed by ...
详细信息
ISBN:
(纸本)0780386477
Due to some drawbacks, such as the instability of the analog receiver in wideband conditions, an all-digital parallel receiver (APRX) has become more and more attractive. Based on the structure of an APRX proposed by the Jet Propulsion Laboratory (Gray, A. et al., JPL TDA Progress Report 42-131, NASA Code 314-30-11-20-06, 1997), a novel parallel algorithm for fractionally spaced equalization (PFSE) is proposed. Utilizing the constant modulus algorithm (CMA), PFSE takes best advantage of its data overlapping characteristic and achieves perfect combination with the APRX's structure. Simulation results show that the performance of PFSE obviously predominates that of the T-spaced equalizer, with hardly any loss of performance compared with the corresponding serial FSE.
This work presents a study of parallel evolutionary programming (EP). The paper is divided into two parts. The first part proposes a concept of parallel EP. Four numerical functions are used to compare the performance...
详细信息
ISBN:
(纸本)0780385152
This work presents a study of parallel evolutionary programming (EP). The paper is divided into two parts. The first part proposes a concept of parallel EP. Four numerical functions are used to compare the performance between the serial algorithm and the parallel algorithm. In the second part, we apply parallel EP to a more complicated problem - an evolving neural networks problem. The results from this problem show that the parallel version is not only faster than the serial version, but the parallel version also more reliably finds optimal solutions.
This work describes the authors' approach at their university in the last three years for computer science second grade undergraduate students to experience parallel and distributed computing. The goal is to give ...
详细信息
This work describes the authors' approach at their university in the last three years for computer science second grade undergraduate students to experience parallel and distributed computing. The goal is to give a solid understanding of parallel and distributed processing technologies and to build up basic skills in the field, such as parallel algorithms, multi-thread/network programming, IP/socket communication, MVC paradigm, RPC/remote method invocation (RMI), Database/SQL, and Java/JDBC. The course features a combination of active experimental learning and N to N networking approach. Unlike typical laboratories where central parallel servers or parallel machines are used (N users to one system networking), our laboratories do without them and instead organize groups of student PCs to form virtual parallel/distributed systems (N users to N systems networking). All PCs work as servers as well as clients. parallel bucket sorting and virtual shopping mall implementations are employed for the course projects. The course consists of 14 ninety-minutes sessions within a semester, including introductory Java network programming and two projects. As the time is limited, homework and pre-laboratory experiments are encouraged. Web based course material distribution and the virtual laboratory environment contributed to student success.
In this paper, new parallel algorithm, based on the conventional FDTD algorithm, is presented for solving the Maxwell's curl equations on a local area network using the MPI system. The performance of the proposed ...
详细信息
In this paper, new parallel algorithm, based on the conventional FDTD algorithm, is presented for solving the Maxwell's curl equations on a local area network using the MPI system. The performance of the proposed parallel algorithm is studied by using a line current source radiating in a two-dimensional domain. It is observed that, for large computational domains, the new parallel algorithm provides a significant reduction in the simulation time as compared with the sequential solution. On the other hand, when partitioning the computational domains over many processors, especially for the small domains, the efficiency of the parallelization reaches a limitation. This is because of the computation time required to update the fields will be reduced to a point where it has the same order as the communication time needed to exchange the fields between the processors.
An edge is a bisector of a simple path if it contains the middle point of the path. Let T = (V, E) be a tree. Given a source vertex s is an element of V, the single-source tree bisector problem is to find, for every v...
详细信息
An edge is a bisector of a simple path if it contains the middle point of the path. Let T = (V, E) be a tree. Given a source vertex s is an element of V, the single-source tree bisector problem is to find, for every vertex v is an element of V, a bisector of the simple path from s to v. The all-pairs tree bisector problem is to find for, every pair of vertices u, v is an element of V, a bisector of the simple path from u to v. In this paper, it is first shown that solving the single-source tree bisector problem of a weighted tree has a time lower bound Omega (n log n) in the sequential case. Then, efficient parallel algorithms are proposed on the EREW PRAM for the single-source and all-pairs tree bisector problems. Two O(log n) time single-source algorithms are proposed. One uses O(n) work and is for unweighted trees. The other uses O(n log n) work and is for weighted trees. Previous algorithms for the single-source problem could achieve the same time O(log n) and the same optimal work, O(n) for unweighted trees and O(n log n) for weighted trees, on the CRCW PRAM. The contribution of our single-source algorithms is the improvement from CRCW to EREW. One all-pairs parallel algorithm is proposed. It requires O(log n) time using O(n(2)) work. All the proposed algorithms are cost-optimal. Efficient tree bisector algorithms have practical applications to several location problems on trees. Using the proposed algorithms, efficient parallel solutions for those problems are also presented.
In this paper: we present rounding error bounds of recent parallel versions of Forsythe's and Clenshaw's algorithms for the evaluation of finite series uf Chebyshev polynomials of the first and second kind. Th...
详细信息
In this paper: we present rounding error bounds of recent parallel versions of Forsythe's and Clenshaw's algorithms for the evaluation of finite series uf Chebyshev polynomials of the first and second kind. The backward errors are studied by using the matrix formulation of the algorithm, whereas the forward error is also Studied by means of a more direct approach that permits us to obtain sharper bounds. The bounds show all almost stable behavior as ill the sequential algorithms. This fact is: illustrated with several numerical tests. (C) 2001 Elsevier Science Ltd. All rights reserved.
暂无评论