Single-processor Computer Research Articles

A DIAKOPTICAL TECHNIQUE FOR SHORT CIRCUIT STUDIES OF LARGE SIZE POWER SYSTEM NETWORKS

ABSTRACT The paper presents a diakoptical technique for deriving the impedance matrices required for short circuit studies of large size power system networks by tearing the system into smaller subsystems by cutting appropriate tie lines. The bus admittance matrix is formed for the sequence networks of each subsystem and then modified later by means of diakoptics. The simulation of mutual couplings is taken into account in the zero sequence networks. Only columns of the impedance matrices corresponding to the faulted buses are computed in the proposed technique which is based on factorized bus admittance matrices and full exploitation of sparsity. The proposed technique can either be used in a single–processor computer for sequential solution of torn subsystems or in a multicomputer configuration for a faster solution by parallel processing of torn subsystems. Both symmetrical and unsymmetrical faults are studied by using the proposed technique which produces exact results as the untorn system. No iterati...

Polynomial collocation using a domain decomposition solution to parabolic PDE's via the penalty method and explicit/implicit time marching

A domain decomposition method is examined to solve a time-dependent parabolic equation. The method employs an orthogonal polynomial collocation technique on multiple subdomains. The subdomain interfaces are approximated with the aid of a penalty method. The time discretization is implemented in an explicit/implicit finite difference method. The subdomain interface is approximated using an explicit Dufort-Frankel method, while the interior of each subdomain is approximated using an implicit backwards Euler's method. The principal advantage to the method is the direct implementation on a distributed computing system with a minimum of interprocessor communication. Theoretical results are given for Legendre polynomials, while computational results are given for Chebyshev polynomials. Results are given for both a single processor computer and a distributed computing system.

An Efficient Computational Method of the Jacobian for Robot Manipulators

SUMMARYIn this paper, an efficient method for computing the Jacobian matrix for robot manipulators on a single processor computer is developed. Compared with the existing methods, the number of required numerical operations is considerably smaller, making the proposed technique the fastest, or the least expensive, one for any general N degrees-of-freedom manipulator.

Systolic accelerator for parametric surface modelling

Two classes of systolic architectures are presented that are able to compute bicubical B spline or Bezier polynomial coefficients and carry out polynomial evaluations. Using a pair of full arrays it is possible to compute all the coefficients in parallel, and to evaluate the polynomials for a given surface, as well as provide a speedup factor of more than 1500 compared with the single processor computation. An alternative solution is to partition both tasks into smaller sub-tasks so that a reduced size of the array is required. This allows a reasonable tradeoff between the speed needs and the VLSI implementation requirements to be achieved.

Comparison of autofocus methods for automated microscopy

Traditional autofocus methods were designed for microscopes driven by single processor computers. As computers are developed that exploit massive parallelism when acquiring and analyzing images, parallel cellular logic techniques became available to focus automatically. This paper introduces the reader to both cellular logic techniques for autofocus and a new spectral moment autofocus measure. It then compares these methods with more traditional autofocus methods. It is shown that traditional methods based on measurements of image power-give the best results when tested on one set of real images and two sets of synthetic images. The next best methods are the cellular logic and spectral moment techniques, while the worst are those based on the image probability density function or histogram.

A function oriented architecture for process control systems minimising internal data transfer costs

Process control computer systems are to execute programs, which are associated and temporaly synchronised with the events occurring in external technical processes. Presently, bus oriented, modularly organised single processor computers with the von Neumann structure are generally employed to control industrial processes in real time. They are equipped with appropriate process peripherals and possess interrupt units to enable real time operation. Since peripheral components can independently carry out only rather elementary operations, the usage of processors by the operating system functions is very high. All of them are executed together with the application tasks in a multitasking environment. This concept was suggested mainly for economical reasons. However, with the continuing increase of the integration density of semiconductor devices and the development of microprocessors, the cost relations were overturned: data transmission installations and the wiring of single components are much more expensive than processing and storage elements. Therefore, a new and more powerful architecture of process control computers, which reflects the intrinsic parallelism of the applications and reduces the total costs by the minimisation of the transmission and cabling expenses is developed. Proceeding from the possibility of parallel activities, which is given in real time control applications on the task level, eight types of functional units are defined for the modular configuration of function oriented process control computers. The responsibility for a part of the basic operations and operating system services, which are needed in real time computing, is assigned to each type. The set of functional units allows to compose a suitable asymmetrical distributed multiprocessor system for any kind of application.

Parallel clustering algorithms

Clustering techniques play an important role in exploratory pattern analysis, unsupervised learning and image segmentation applications. Many clustering algorithms, both partitional clustering and hierarchical clustering, require intensive computation, even for a modest number of patterns. This paper presents two parallel clustering algorithms. For a clustering problem with N = 2 n patterns and M = 2 m features, the time complexity of the traditional partitional clustering algorithm on a single processor computer is O( MNK), where K is the number of clusters. The proposed algorithm on anSIMD computer with MN processors has a time complexity O( K( n + m)). The time complexity of the proposed single-link hierarchical clustering algorithm is reduced from O( MN 2) of the uniprocessor algorithm to O( nN) with MN processors.

Parallel algorithms for the finite element method

Existing finite element methods, algorithms and application programs have been developed in the milieu of the serial or single processor computer. Substantial efforts are currently being made to modify their computational aspects in order to take advantage of the power inherent in the newer generations of parallel multiprocessor computers. The parallelizing techniques being explored tend to be derived from or are closely related to serial concepts. Parallelism is difficult to achieve from this direction. The present work develops and implements a new class of parallel algorithms for the finite element method based upon the use of projections on finite-dimensional spaces. Projections provide a formal and general mechanism for decomposing or partitioning a problem into a col;ection of smaller related problems which can be treated in parallel. These algorithms can be programmed in a wholly asynchronous mode and have been implemented and studied on a shared memory multiprocessor computer with up to 24 processors. On large test problems they have been shown to produce almost perfect speedup and to be extremely efficient.

An Analytical Model forUnix® Systems

This paper presents an analytical model for a single-processor interactive computer system running under the UNIX operating system. The model is a closed multichain, multiclass priority-queueing network model. The model solution is an approximation and is based on mean value analysis. The validity of the model has been established by using it with a variety of workloads and workload mixes and with a number of different computer systems. Mean response times predicted by the model are within 10 percent of measured values. The model can be extended to tightly coupled multiprocessor systems and multiple UNIX systems running with network file systems over local-area networks.

Parallelism by hierarchy of GPST inversion algorithm for elastic wave equation

The Generalized Pulse-Spectrum Technique (GPST) is a versatile and efficient iterative numerical algorithm for solving multi-parameter inverse problems of a system of nonlinear partial differential equations. Here parallelism is introduced into the hierarchy of GPST to further improve its efficiency for solving the multi-parameter inverse problems with a large number of unknown parameters in some geophysical applications. Its implementation is quite easy and it achieves large speedup with almost no communication overhead even on a single-processor computer. Effectively, this parallelism in GPST reduces the computational effort for solving any multi-parameter inverse problem approximately to that of the corresponding single-parameter inverse problem. A simple computational complexity analysis is performed for the parallel structured GPST to estimate its speedup. At the present, the feasibility and capability of the parallel structure GPST are only tested in a limited sense by performing numerical simulations for a three-parameter and two-dimensional inverse problem of the isotropic linear elastic wave equation on a single-processor computer sequentially. It is found that the parallel structured GPST is not only as robust as the standard GPST but also shows a speedup per iteration close to its estimated value.

High level parallelism in hierarchy of GPST algorithm for parameter identification in engineering mechanics

The Generalized Pulse-Spectrum Technique (GPST) is a versatile and efficient iterative numerical algorithm for solving multi-parameter idenitification problems in engineering which can be formulated mathematically as multi-parameter inverse problems of a system of partial differential equations. Here high level parallelism is introduced into the hierarchy of GPST. By using a simple computational complexity analysis, it is shown that very large speedup with no communication overhead can be achieved even without the introduction of low level parallelism into the hierarchy of GPST. Moreover, the parallel structured GPST can achieve considerable speedup over the standard GPST even on a single-processor computer.

Numerical micromagnetics by the finite element method

We developed a new numerical procedure to study dynamical behavior in micromagnetic systems. This procedure solves the damped Gilbert equation for a continuous magnetic medium, including all interactions in standard micromagnetic theory in three-dimensional regions of arbitrary geometry and physical properties. The magnetization is linearly interpolated in each tetrahedral element in a finite element mesh from its value on the nodes, and the Galerkin method is used to discretize the dynamic equation. We compute the demagnetizing field by solution of Poisson's equation and treat the external region by means of an asymptotic boundary condition. The procedure is implemented in the general purpose dynamical micromagnetic code (GDM). GDM uses a backward differential formula to solve the stiff ordinary differential equations system and the generalized minimum residual method with an incomplete Cholesky conjugate gradient preconditioner to solve the linear equations. GDM is fully parallelized using MPI and runs on massively parallel processor supercomputers, clusters of workstations, and single processor computers. We have successfully applied GDM to studies of the switching processes in isolated prolate ellipsoidal particles and in a system of multiple particles.

Nonlinear substructuring method for concurrent processing computers

This paper proposes a method, based on substructuring, to solve nonlinear structural analysis problems on multiprocessor computers. Background information is given on the use of substructuring in large-scale finite element programs and computational time distributions for the major components for an example nonlinear finite element analysis are discussed. Implementation of the substructuring method on a typical multiprocessor computer is described and estimates are made of expected reductions in computation times based on nonlinear substructuring results obtained on a single processor computer.

Parallel algorithms for quantum chemistry. I. Integral transformations on a hypercube multiprocessor

For many years it has been recognized that fundamental physical constraints such as the speed of light will limit the ultimate speed of single processor computers to less than about three billion floating point operations per second (3 GFLOPS). This limitation is becoming increasingly restrictive as commercially available machines are now within an order of magnitude of this asymptotic limit. A natural way to avoid this limit is to harness together many processors to work on a single computational problem. In principle, these parallel processing computers have speeds limited only by the number of processors one chooses to acquire. The usefulness of potentially unlimited processing speed to a computationally intensive field such as quantum chemistry is obvious. If these methods are to be applied to significantly larger chemical systems, parallel schemes will have to be employed. For this reason we have developed distributed-memory algorithms for a number of standard quantum chemical methods. We are currently implementing these on a 32 processor Intel hypercube. In this paper we present our algorithm and benchmark results for one of the bottleneck steps in quantum chemical calculations: the four index integral transformation.

Massively Parallel Computers: The Connection Machine and NON-VON

The demand for high-speed computers is increasing, and as the limits on single-processor computers are approached, researchers are turning their attention to parallel computers. Parallel computers have more than one processing element; massively parallel computers contain many processing elements. Constructing computers on this scale and learning how to program them effectively will be major challenges in the next decade. Several such computers, for example the Connection Machine and the NON-VON, are under development.

Towards a taxonomy of computer architecture based on the machine data type view

Existing taxonomies of computer architecture lack the descriptive tools to deal with the large variety of existing principles, features, and mechanisms of the existing spectrum of single processor, multi processor, and multi computer architectures. Consequently, they lack the discriminating power to be able to taxonomize computer architecture. In the paper, a new approach toward a complete taxonomy is presented. The key to the taxonomy is to start with the dichotomy of 'operational principle' and 'hardware structure' as the foundation of a computer architecture and describe the constituents of the operational principle in terms of 'machine data types' consisting of 'machine data objects', their representations, and the functions applicable on the objects. The resulting taxonomy provides a systematic approach to the design of innovative computer architectures.

A problem of organizing data processing on a computer

A method for obtaining all feasible schedules for solving, on a single-processor computer and allowing interruptions, a system of problems with known solution times and specified times for the start and finish of the solution is described.

Why sort-merge gives the best implementation of the natural join

To join two relations efficiently, they must not only be clustered but mutually clustered . Sorting is the only known way to achieve mutual clustering. Once the relations are sorted, merging is the obvious way to implement the join. If the relations are known to be sorted appropriately, the most costly part of the process can be omitted. To know that a relation is sorted already, it is best to remember that we sorted it. Otherwise detecting that the relation is sorted requires inspection of each tuple, a cost comparable in many cases to the cost of sorting. Detecting any other form of mutual clustering is bound to be a combinatorial process of much greater cost than sorting. Cheaper forms of detection may be possible using sampling, but these are heuristics at best and there is no reliable foundation for them.It is assumed that the sizes of the relations are large compared with the RAM memory available or, equivalently, compared with the number of processors available. Our conclusion that sort-merge is best is thus applicable to single-processor computers or to multiple-processor database machines. Discussing the cost of joining relations without making this assumption is outside the subject of databases.

Frontal optimization algorithms for multiprocessor computers

The authors describe one of the approaches to the construction of locally optimal optimization algorithms on multiprocessor computers. Algorithms of this type, called frontal, have been realized previously on single-processor computers, although this configuration does not fully exploit the specific features of their computational scheme. Experience with a number of practical discrete optimization problems confirms that the frontal algorithms are highly successful even with single-processor computers. 9 references.

Parallel Generation of Binary Search Trees

A method of constructing binary search trees in a multiprocessor computer system is proposed. Asymptotically, this method achieves the maximum possible increase in speed as compared with a single processor computer system. To make better use of this method, a parametrized restructuring of binary search trees is also discussed.