Multiprocessor Computer Research Articles

Reduction process based on proper orthogonal decomposition for dual formulation of dynamic substructures

Novel reduction processes utilizing the proper orthogonal decomposition method have been proposed to facilitate dual formulation of dynamic substructures. Dual formulation is significant owing to its capability of effectively solving ill-conditioned problems related to parallel-processing computers. Dual-formulation techniques employ the Lagrange multiplier method to couple substructures. In the proposed approach, substructures are reduced under constraint conditions via use of the Lagrange multiplier method, and each domain is divided into internal and interface parts—each handled separately during the reduction process. The interface to which the coupling condition is imposed is preserved without further reduction (using the internal reduction method) or reduced to a different ratio from the internal part (via internal and interface reduction technique). Once all substructures are reduced, Lagrange multipliers are applied to preserved or reduced interface parts. Since Boolean matrices that preserve constraint conditions are not reduced but constructed instead, the stability and accuracy of the reduced system are highly enhanced.

PLANNING OF DOWNLOAD PROCESSORS IN MULTIPROCESSOR SYSTEMS OF CRITICAL PURPOSE

At present, multiprocessor systems of a critical nature are widely used. Such systems are used for tracking, aiming, observing, etc. Such tasks, as a rule, require maximizing productivity and reducing the time to solve a problem. For these purposes, the initial selection of non-dependent linear, conditional and cyclic sections of sequential programs is used [1]. This is done to release fragments of programs that can be assigned to execution on processors in such a way that during execution they exchange data with neighboring processors as little as possible. Due to this, it is possible to partially improve the performance of a multiprocessor computing system, together with a decrease in the overall execution time of the entire task as a whole. For systems of the considered nature of the processor of the entire system, it is desirable to assign program fragments so that they are constantly loaded throughout the solution of the entire problem. This is another way to improve the performance of a multiprocessor system. It is obvious that the use of software for this purpose is not real due to the criticality of the time parameter. Therefore, it is relevant to use methods and corresponding hardwareoriented algorithms for scheduling processor loads, which is the subject of research in this article. The article shows the relevance of the constant loading of processors of multiprocessor systems with a high availability factor. The necessity of drawing up a plan for loading processors to support this coefficient is substantiated. An appropriate method and algorithm for multiprocessor systems for critical purposes (tracking systems, surveillance, aiming, etc.) are proposed.

Generalized Spatial Differentiation from the Spin Hall Effect of Light and Its Application in Image Processing of Edge Detection

Optical spatial differentiation is important for applications in massively parallel computing and real-time image processing. This study makes use of the generalized spin Hall effect of light to perform optical spatial differentiation. Importantly, this scheme generally accompanies oblique light reflection and refraction at $a\phantom{\rule{0}{0ex}}n\phantom{\rule{0}{0ex}}y$ planar interface, regardless of material compositions or angles of incidence. Moreover, this spin-optical approach offers robust image processing to extract the boundaries of objects, where two distinct images can be stored in two different polarization states.

Wall roughness influence on the NPSH characteristics of centrifugal pumps

Pumps are very important and very widespread energetic machines. They are used in all branches of industry. Pumps are also a major consumer of electricity and therefore their development is even more important. From theoretical through experimental development in the last decades, a quantity of research has focused on numerical analysis. These are used to optimize all kinds of characteristics, from energy and cavitation, to dynamic characteristics. At the beginning of Computational Fluid Dynamics (CFD) analyses were related to the physical model (potential flow, Euler equation, Navier-Stokes equations). Subsequently, a lot of research was carried out in the field of various turbulent models. With the launch of high performance multi-processor computers, more and more time-dependent problems have been analysed. All the above mentioned research was carried out with numerical calculations on smooth surfaces. The fact is that in reality there is always some roughness on wet surfaces, depending on the material, the technology of production of certain parts of machines and finishing surface treatment. In the article, we give preliminary results of numerical analyses of centrifugal pumps, where the roughness of the walls was taken into account. The analyses were completed for different types of pumps, with different roughnesses and for different characteristics. First he effect of the roughness on the efficiency is shown, further the cavitation characteristics were analysed. The basic idea for research comes from the fact that there are still many cases where the matching of the results of the measurements and the results of numerical analyses are not perfect. One of the possible reasons can be the roughness of the walls.

High-resolution hydrologic forecasting for very large urban areas

Abstract With continuing growth of urban populations worldwide, high-resolution hydrologic forecasting is an increasingly important hydroinformatics service for large urban areas. In the Dallas-Fort Worth (DFW) area, the Collaborative Adapting Sensing of Atmosphere (CASA) WX program has been providing real-time hydrologic products, such as rainfall and streamflow, at 1 min–500 m resolution using the NWS Research Hydrologic Distributed Model forced by the Quantitative Precipitation Estimate from a network of X-band weather radars. There is an increasing demand, however, for even higher-spatial resolution hydrologic products. In this paper, we assess the ability of the current streamflow product to capture the hydrologic response of urban catchments in the DFW area, the utility of ultrasonic distance sensors for real-time sensing of water level in urban streams, and the feasibility of higher-resolution operation using parallel processing and cloud computing. We show that the CASA WX streamflow product skillfully captures the stage and streamflow response from rainfall for the majority of the nine catchments studied, but that timing errors significantly deteriorate the quality of streamflow prediction for certain basins. Comparative evaluation of different computing models shows that a reduction in runtime of up to 34% is possible with parallel processing at 1 min–250 m resolution.

Ferroelectric Analog Synaptic Transistors.

Neuromorphic computing is a promising alternative to conventional computing systems as it could enable parallel computation and adaptive learning process. However, the development of energy efficient neuromorphic hardware systems has been hindered by the limited performance of analog synaptic devices. Here, we demonstrate the analog conductance modulation behavior in the ferroelectric thin-film transistors (FeTFT) that have the nanoscale ferroelectric material and oxide semiconductors. Accurate control of polarization changes in the nanoscale ferroelectric layer induces conductance modulation to demonstrate linear potentiation and depression characteristics of FeTFTs. Our devices show potentiation and depression properties, including high linearity, multiple states, and small cycle-to-cycle/device-to-device variations. In simulations with measured properties, a neuromorphic system with FeTFT achieves 91.1% recognition accuracy of handwritten digits. This work may provide a way to realize the neuromorphic hardware systems that use FeTFTs as the synaptic devices.

�������� ��������������� ������� �������� ��������� .NET

The majority of modern software applications make use of several execution threads. For example, in the text editor, one thread is responsible for the spell cheking, another for autosaving, one more thread for handling the user input. The mechanism of multiple threads provides the software developer with a convenient way of structuring the functional possibilities: the software application architecture will be more stable, if it's various possibilities are programmed separately from each other. The ability to use multiple threads will be useful even for a specialist of numerical computing. The iteration methods usually need to execute a large amount of computations in order to receive an approximate solution of a certain problem. This circumstance should be taken into account at the stage of software design architecture so, that a user could follow the computational process and not worry about the responsiveness of the application. In order to launch a separate execution thread, the .Net framework provides a developer with an eective template of task-based asynchronous programming [3] that uses some new keywords async, await and classes Task, Task<TResult> from the namespace System.Threading.Tasks. The instances of this class are created very fast, they need only a little amount of system resources and they eectively use the processors time. That's why it is convenient to move the computations to a separate thread, which on its turn can also launch dierent parallel computational threads. This approach signicantly accelerates a software performance on a multiprocessor (or multi-core) computer. The keyword async in the name of an asynchronous method changes the way of how the result of this method is handled, and gives a possibility to use the instruction await for launching any awaitable object. For this purpose, in the .Net framework there are usually used the instances of the classes Task and Task<Result>. The task, launched with the instruction await, does not block the user interface thread during the computation. Before launching an synchronous task, there are dierent preparation steps: create a stop-marker CancellationToken token, an object Progress<T> progress in order to be able to send the data to the main thread, and nally a set of local variables to correctly catch the execution context. Then the asynchronous method in the right moment calls progress.Report(data), for a "soft" termination of an asynchronous algorithm. In a convenient point in time, this algorithm should check the property token.IsCancellationRequested and terminate the execution on request. The cancellation request can be initiated from the main thread by calling the method token.Cancel(). In order to synchronize the nishing of the array of asynchronous tasks, the static method Task.WhenAll(tasks).Wait() is used. For the demonstration of the described approaches an application has been created. It demonstrates how to apply the genetic algorithm in order to solve the traveling sales-man problem [6] for a large number of cities. It is experimentally analyzed how the number of threads inuences the performance of the application. The complete code base can be downloaded from the repository [4].

Modern computer technologies for modeling the life cycle of buildings based on parallel calculations

The article deals with high-performance information technology (HPC) for the problems of stress-strain analysis at all stages of the life cycle of buildings and structures: construction, operation and reconstruction. The results of numerical simulation of high buildings using software as a processor component based on a new hybrid algorithm for solving systems of linear algebraic equations [1] with a symmetric positive-definite matrix that combines computation on multi-core processors and graphs. It has been found that to accelerate the calculations, hybrid systems that combine multi-core CPUs with accelerator coprocessors, including GPUs, are promising [5]. To test the effectiveness of the proposed parallel algorithm for solving systems of linear algebraic equations [1], numerical experiments were carried out at the most dangerous loads of a 27-story building. Results of numerical researches with use for preprocessor (input of initial data) and postprocessor (output of results of calculations) of processing of the LIRA-SAPR software complex are presented [2, 4, 6]. The results of numerical studies of the behavior of structures of high buildings have shown a multiple reduction in the time of solving systems of linear algebraic equations with symmetric matrices on multiprocessor (multi-core) computers with graphical accelerators using the proposed hybrid algorithms [1]. High-performance technologies based on parallel calculations give more effect than more complex processes: modeling of life cycle of high buildings, bridges, especially complex structures of NPPs, etc. for static and dynamic loads, including emergencies in normal and difficult geological conditions, which make up 70% of Ukraine's territories.

Economic explicit-implicit schemes for solving multidimensional diffusion-convection problems

This work concerns the development of an efficient parallel algorithm for numerical solution of nonstationary diffusion-convection problems by means of a multiprocessor computer system with distributed memory. Economically explicit-implicit difference schemes and the method of splitting into physical processes are used as a basis. The original problem is replaced by a sequence of one-dimensional and two-dimensional difference problems using complex schemes that approximate the original problem in the general sense. Explicit-implicit difference schemes involve explicit approximation in horizontal directions and implicit approximation with weights in a vertical direction and require less time for solving diffusion-convection problems compared to explicit schemes while maintaining acceptable accuracy of solutions. The algorithm is proposed to find the optimal weight value and it yields the lowest approximation error in the solution of the diffusion-convection problem in the vertical direction for given time grid steps. The three-dimensional model problem of transport of suspensions in the water environment is considered. The model takes into account the following processes: advective transport caused by the movement of the water medium, microturbulent diffusion and gravitational deposition of suspended particles, as well as changes in the geometry of the bottom caused by the deposition of suspended particles or the rise of sediment particles. Application of the parallel algorithm developed for numerical modeling of suspension transport makes it possible to significantly improve the real-time forecast accuracy and the validity of the accepted engineering decisions used in designing the objects of coastal infrastructure.

Direct Numerical Simulation of Homogeneous Isotropic Helical Turbulence with the TARANG Code

The problem of taking into account the influence of turbulence comes up while solving both fundamental questions of geo- and astrophysics and applied problems arising in the development of new engineering solutions. Difficulties in applying the standard propositions of the theory appear when considering flows with a special spatial structure, for example, helical flows. The flow helicity determines the topology of vortices and is conserved in the process of energy transfer in a turbulent flow. In this paper we suggest an approach for numerical simulation of homogeneous isotropic helical turbulence aimed at detecting characteristic signatures of the inertial range and finding the distributions of the spectral energy and helicity densities. In this approach we use the TARANG code designed to numerically solve various problems of fluid dynamics in the regime of a developed turbulent flow and to study hydrodynamic instability phenomena of a different physical nature (thermal convection, advection of passive and active scalars, magnetohydrodynamics, and the influence of Coriolis forces). TARANG is an open source code written in the object-oriented C++ language with a high efficiency of computation on multiprocessor computers. Particular attention in the paper is given to the application of the tool kit from the package to analyze the solutions obtained. The spectral distributions and fluxes of energy and helicity have been computed for Reynolds numbers of 5700 and 14 000 on 5123 and 10243 grids, respectively. We have checked whether the –5/3 spectral law is realizable and estimated the universal Kolmogorov and Batchelor constants in the inertial range. An analysis of the energy and helicity transfer functions between the selected scales (shell-to-shell transfer) shows a significant contribution of nonlocal interactions to the cascade process.

Characterizations of Minimum Structure- and Substructure-Cuts of Hypercubes

Abstract For a graph G and a connected subgraph H of G, denote by ℋ(G;H) the set of subgraphs of G which are isomorphic to H and denote by ℋS(G;H) the union of sets of subgraphs of T, where T ranges over all the elements in ℋ(G;H). An H-structure-cut (respectively, H-substructure-cut) of G is a subset of ℋ(G;H) (respectively, ℋS(G;H)), if any, whose removal disconnects G. The H-structure connectivity (respectively, H-substructure connectivity) is the cardinality of a minimum H-structure-cut (respectively, H-substructure-cut) of G. The hypercube is one of the most attractive interconnection networks for large-scale multiprocessor computer systems. In this paper, we will characterize the minimum H-structure-cuts and the minimum H-substructure-cuts of hypercubes for H∈{K1,1,K1,2,K1,3,C4}.

Two Direct Low Thrust Trajectory Optimization Techniques

Direct low thrust trajectory optimization techniques that reduce the optimization problem in the class of continuous functions to a nonlinear programming problem are described. Two direct low thrust trajectory optimization methods are implemented; one of them is based on the optimization of thrust acceleration and the other optimizes the impulses approximating the thrust arcs. These methods are compared using a sample problem of transfer from Earth to Mars in terms of runtime, the number of iterations, and the size of the convergence domains. The computations are performed on a personal computer using MATLAB and on a multiprocessor computer.

Determination of radiation pattern for linear antenna array with multistage adaptive compensator of interferences

Introduction. Adaptive compensation algorithms of interference in digital antenna array (DAA) are characterized by high computational complexity. As result, it makes it difficult to implement them in real time. A parallelization of computational processes and simultaneous processing in all parallel channels is one of the main directions of increasing the speed of processing. The approach based on block Gram-Schmidt orthogonalization makes it possible to synthesize multistage adaptive compensators of interference with parallel-sequential signal processing that similar to the architecture of multiprocessor computer systems. The signal processing in DAA based on multistage adaptive compensator of interference is more complex than in single-stage ones. Each stage has own set of weight coefficients. As result, a technique for calculating radiation pattern (RP) of a linear DAA with a multistage adaptive compensator of interference and certain architecture and parameters is required. Theoretical results. The block diagram of the adaptive compensator of interference by using block signal orthogonalization of compensation channels and consistent determination of the compensation error has a ladder form and includes some stages. Each stage consists groups of modules. Each module in the group is a multi-input weight adder. The complex signal envelope from the DAA output applies to the main channel of an adaptive compensator of interference. The technique for calculating RP of DAA with a multistage adaptive compensator of interference by using block orthogonalization of compensation channels signals includes the following steps: 1. Forming of the matrix impulse characteristics for the adaptive compensator of interference stages. 2. Calculation of the matrix impulse response for a multistage filter realizing an interference compensation procedure based on its factorized representation. 3. Determination of the weight coefficients of an equivalent single-stage adaptive compensator of interference in the form of multi-input weighted adder. 4. Calculation of the RP of DAA with an equivalent single-stage adaptive compensator of interference. Experimental results. RP analysis based on the proposed technique is carried out for six-element linear DAA with one-, two-, three- and six-stage adaptive compensators of interference and three jammers. The analysis confirms that considered adaptive compensators of interference have the same effectiveness of interference elimination. In the same time, they are characterized by the same accuracy of "zeros" formation on interference sources and the width of rejection zone for different values of correlation matrix conditioning of the interference. Conclusions. We propose the technique for calculating the RP of DAA with multi-stage adaptive compensator of interference by using the factorized representation of the matrix impulse response of a multistage filter. This compensator is based on block signals orthogonalization of compensating channels and provides parallel-sequence signal processing.

Modeling of Heat Mass Transfer in High-Temperature Reacting Flows with Combustion

A multiprocessor computer system suitable for physical, mathematical, and chemical models, as well as an exact method for the solution of a system of differential equations that describe the actual combustion of a pulverized coal flare, are necessary to study the numerically complex, physicochemical processes occurring in the combustion chambers of power plants. The results of numerical simulation can provide quite a high accuracy. However, the task of setting up a physical and mathematical model with the correct initial and boundary conditions has yet to be completed. In this paper, we studied heat and mass transfer in high-temperature reacting flows during the burning of Karaganda coal in the combustion chamber of an actual power boiler of a thermal power plant in Kazakhstan. The optimal conditions for computational experiments that correspond to real combustion processes are determined.

Practical aspects of implementation of the parallel algorithm for solving problem of ctenophore population interaction in the Azov sea

The paper covers the development and researching mathematical model of interaction processes between plankton and ctenophore populations based on the modern information technologies and computational methods, which leads to increase of the accuracy of predictive modeling of the ecology situation in shallow water in summer. The model takes into account the following: the transport of water environment; microturbulent diffusion; nonlinear interaction of plankton and ctenophore populations; biogenic, temperature and oxygen regimes; influence of salinity. The computational accuracy is significantly increased, and computational time is decreased at using the calculation method based on partially filled cells for discretization of model. The practical significance is the software implementation of the proposed model, the limits and prospects of its practical use are defined. Experimental software was developed based on multiprocessor computer system, which is intended for mathematical modeling of possible progress scenarios in shallow waters ecosystems on the example of the Azov Sea in summer. We used decomposition methods of grid domains in parallel implementation for computationally laborious convection-diffusion problems, taking into account the architecture and parameters of multiprocessor computer system.

Experimental study of energy and time constrained task scheduling with irregular speed and power levels

We consider energy and time constrained scheduling of independent sequential tasks on a multiprocessor computer with bounded and discrete and irregular clock frequency and supply voltage and execution speed and power consumption levels. This is a very realistic power consumption model. However, it is very difficult to find useful information about the optimal solutions which are critical in evaluating the performance of heuristic algorithms. Our approach in this paper has two unique features. First, we develop algorithms that are applicable to all multiprocessor computers with bounded and discrete and irregular clock frequency and supply voltage and execution speed and power consumption levels. Second, we evaluate the performance of these algorithms on multiprocessors with a regular or close-to-regular power consumption model, for which, we have lower bounds for the optimal solutions. By using these lower bounds, we can perform both analytical study and experimental evaluation of the performance of our algorithms. We propose nine pre-power-determination algorithms for energy constrained scheduling; nine post-power-determination algorithms for energy constrained scheduling; nine pre-power-determination algorithms for time constrained scheduling; and nine post-power-determination algorithms for time constrained scheduling. The performance of these algorithms are evaluated analytically and also experimentally by comparing their solutions with optimal solutions, where the optimal solutions are obtained for regular or approximately regular power consumption models. We find that the combination of the largest execution requirement first method for task selection and the longest time first method for list scheduling yields the best performance.

High-performance computer-aided optimization of viscous dampers for improving the seismic performance of a tall building

Using fluid viscous dampers (FVDs) has been demonstrated to be an effective method to improve seismic performance of new and existing buildings. In engineering applications, designs of these dampers mainly rely on trial and error, which is repetitive and labor intensive. To improve on this tedious manual process, it is beneficial to explore more formal and automated approaches that rely on recent advances in software applications for nonlinear dynamic analysis, performance-based evaluation, workflow management and the computational power of high-performance, parallel processing computers. The optimization design procedure follows the framework of Performance Based Earthquake Engineering (PBEE) and uses an automatic tool that incorporates an optimization engine and structural analysis software: Open System for Earthquake Engineering Simulation (OpenSEES). An existing 35-story steel moment frame is selected as a case-study building for verification of this procedure. The goal of the retrofit design of FVDs is to improve the building's seismic behavior to avoid collapse under a Level 2, basic safety earthquake event (BSE-2E). An objective function of building's total loss under a BSE-2E event is used and its optimal damper patterns will be proposed. The efficiency of the optimization procedure will be demonstrated and compared with the manual procedure.

A parallel computing thin‐sheet inversion algorithm for airborne time‐domain data utilising a variable overburden

ABSTRACTAccurate modelling of the conductivity structure of mineralisations can often be difficult. In order to remedy this, a parametric approach is often used. We have developed a parametric thin‐sheet code, with a variable overburden. The code is capable of performing inversions of time‐domain airborne electromagnetic data, and it has been tested successfully on both synthetic data and field data. The code implements an integral solution containing one or more conductive sheets, buried in a half‐space with a laterally varying conductive overburden. This implementation increases the area of applicability compared to, for example, codes operating in free space, but it comes with a significant increase in computational cost. To minimise the cost, the code is parallelised using OpenMP and heavily optimised, which means that inversions of field data can be performed in hours on multiprocessor desktop computers. The code models the full system transfer function of the electromagnetic system, including variable flight height. The code is demonstrated with a synthetic example imitating a mineralisation buried underneath a conductive meadow. As a field example, the Valen mineral deposit, which is a graphite mineral deposit located in a variable overburden, is successfully inverted. Our results match well with previous models of the deposit; however, our predicted sheet remains inconclusive. These examples collectively demonstrate the effectiveness of our thin‐sheet code.

SOLUTION OF DIFFERENTIAL EQUATIONS AND SYSTEMS OF DIFFERENTIAL EQUATIONS WITH THE USE OF MULTIPROCESSOR COMPUTERS

The paper proposes computational methods for solving differential equations and systems of two differential equations with initial conditions. At the beginning of the article the statement of the problem is given. Then we propose an essential complement to the classical methods for solving the Cauchy problem, which gives a new indication of the correctness of the estimation of the functionsolution. This feature is based on the geometric construction of the solution of several Cauchy problems that differ only in the initial conditions. The article describes an algorithm for constructing a solution on the investigated interval. After that, the possibility of generalizing a feature to systems of two differential equations is described, when the solution is not just a function but a function in space. The methods themselves are easily parallelized, therefore, they can effectively use multiprocessor computers.

METHODS OF DEVELOPING TEST SOFTWARE OF MULTIPROCESSOR COMPUTER SYSTEMS

The article is devoted to various aspects of the creation of test software for multiprocessor systems (platforms). All the more urgent becomes the task of developing an efficient test checks to determine the operability and functional characteristics of heterogeneous computing systems. For test development was done using various tools: a cross-platform compiler QT 5.2, boost library, CUDA technology, the software package MPICH2, a library QWT, the command processor Bash. The design of the test software was performed using templates developed with the aim of building the desired software architecture. The article describes methods of constructing the control program Test Manager, which provides the ability to run tests in multi-threaded mode. Discuss methods of constructing test programs to test different modules of a heterogeneous multiprocessor computing platform. The results of the acceptance tests confirm the high efficiency of the proposed test methods in a multiprocessor computing system.