Multiprocessor Computer Systems Research Articles

On the reliability of alternating group graph-based networks

The probability of having faults in a multiprocessor computer system increases as the size of system grows. One way to quantify the reliability of a system is using the probability that a fault-free subsystem of a certain size still exists with the presence of individual faults. The higher the probability is, the more reliable the system is. In this paper, we establish the reliability for networks based on AGn, the n-dimensional alternating group graph. More specifically, we calculate the probability of a subnetwork (or subgraph) AGnn−1 being fault-free, when given a single node's fault probability. Since subnetworks of AGn intersect in highly complex manners, our scheme is to use the Principle of Inclusion–Exclusion to obtain a lower-bound of the probability, by considering intersections of up to four subgraphs. We show that the lower-bound derived this way is very close to the upper-bound obtained in a previous result, which means the lower-bound we get is a very tight one. Therefore, both lower-bound and upper-bound are close approximations of the accurate probability.

Програмне реалізування математичних моделей, методів та алгоритмів оціннювання часу виконання складаних програмних комплексів у багатопроцесорних комп'ютерних системах

Програмне реалізування математичних моделей, методів та алгоритмів оціннювання часу виконання складаних програмних комплексів у багатопроцесорних комп'ютерних системах

Modeling of production and destruction processes in coastal systems on a supercomputer

The paper covers the development and research mathematical model of production and destruction processes in coastal system, based on the modern information technologies and computational methods to improve the accuracy of predictive modelling the ecological situations in waters during the summer. The model takes into account following: the movement of water flow; microturbulent diffusion; gravitational settling of pollutants; nonlinear interaction of plankton populations; the impact of salinity, temperature, as well as the pollution entrance and deposition from river flows and air. The scheme with weights was proposed for discretization of the developed model, that allowed significantly reduce the error and the computational time. Practical significance of the work is that the proposed model was software implemented; the limits and prospects of its use were defined. On the basis of a supercomputer, the experimental software was designed for mathematical modelling of possible development scenarios of the ecosystems of coastal waters on the example of the Azov Sea in the summer period, taking into account the influence of environment. Parallel implementation was performed using the decomposition methods for computationally labours diffusion-convection problems taking into account the architecture and parameters of multiprocessor computer system.

Some aspects of the organization of information security of functioning of multiprocessor computing systems

The subject of the research is to improve the features of the organization of information security of multiprocessor computing systems. The goal of this work is to identify and implement measures to secure information that can be effective when using multiprocessor modular computing systems or in parallel calculations on multithreaded systems. To achieve this goal, the following tasks are solved: comparison of data protection methods in multiprocessor and sequential systems, as well as study of the specifics of applying various cryptographic methods to the methods of implementing protection; in accordance with some aspects of constructing multiprocessor systems, consideration and identification of key elements that require special attention in the development of a security system; to show that the main choice of data protection methods in multiprocessor systems is determined by differences from sequential systems in the theoretical and hardware implementation. The methods used are: main provisions of the theory of computing systems, theory of parallel computing, theory of the construction of operating systems, methods and algorithms for data protection. The following results were obtained. The main aspects in definition and use of information protection measures that can be effective when using multiprocessor modular computing systems or in parallel calculations on multithreaded systems are identified. Comparison of methods of data protection with sequential systems is carried out, and also the influence of application of various cryptographic methods on methods of data protection implementation is considered. It is shown that the main choice of data protection methods in multiprocessor systems is determined by differences from sequential systems in theoretical and hardware implementation. The conclusions . In accordance with some aspects of the construction of multiprocessor systems, the key elements that require special attention in the development of a security system are examined and identified. The paper shows that a parallel system requires both more hardware and software, and with each additional calculation module the system becomes more complex. And this, in turn, further complicates the protection system as a whole, which may entail some slowdown in the execution of application programs with the help of multiprocessor computing systems. However, in the long term, the proposed approach makes it possible to provide increased security for the operation of multiprocessor systems. To protect data in such systems, a number of methods are considered and analyzed.

Increasing the accuracy of evaluation of the performance period of software complex components in multiprocessor computer systems under noise stochastic modeling

Increasing the accuracy of evaluation of the performance period of software complex components in multiprocessor computer systems under noise stochastic modeling

Solution of problems of oil reservoirs flooded with water using polymer-dispersed systems on a multiprocessor computer system

The paper provides the results of the numerical simulation of the impact of polymer-dispersed systems on oil reservoirs in order to increase the flow resistance of flooded areas and to involve nonmoving or slow-moving oil reserves in active development. The mathematical model of this process was built based on the known laws of the conservation of mass and momentum. The capillary model of the porous medium was used to construct closure relations that allowed us to describe the formation of polymer-dispersed components and their influence on a porous medium. The complexity of solving the problem of the impact of polymer-dispersed systems on the process of oil displacement by water from a porous medium is related to the three-dimensionality and nonstationarity processes of filtration, as well as the need to calculate the fields of polymer concentration, pressure and saturation fields, and the structural changes in porosity and permeability at each time step and at each point in space. All these factors require significant computer time and resources. The paper presents the algorithm of parallel calculations using a multiprocessor computer system with the MPI library. Calculations on a real field demonstrated the efficiency of the algorithm.

Aircraft Control During Cruise Flight in Windshear Conditions: Viability Approach

This paper addresses the analysis of aircraft control capabilities during the cruise phase (flying at the established level with practically constant configuration and speed) in the presence of windshears. The study uses a point-mass aircraft model describing flight in a vertical plane. The problem is formulated as a differential game against wind disturbances. The first player, autopilot, controls the angle of attack and the power setting, whereas the second player, wind, produces dangerous gusts. The state variables of the model are subjected to constraints expressing aircraft safety conditions. Namely, the altitude, path inclination, and velocity are constrained. Viability theory is used to find the so-called viability kernel, the maximal subset of the state constraint where the aircraft trajectories can remain arbitrary long if the first player utilizes an appropriate feedback control, and the second player generates any admissible disturbances. The computations are based on grid methods developed by the authors and implemented on a multiprocessor computer system.

Complex of parallel programs for modeling oil products transport in coastal systems

This paper covers the model of oil products transport in view the evaporation of light, neutral and no-evaporating pseudofractions of oil slick, dissolution of oil slick and biodegradation. Approximation of diffusion-convection problem was performed on the basis of schemes of high-order of accuracy. Experimental software was designed for mathematical modelling of possible scenarios of development of ecosystems of shallow waters for oil spills on the example of the Azov-Black Sea basin, based on multiprocessor computer systems. Decomposition methods of grid domains have been used for computationally laborious diffusion-convection tasks in parallel implementation. Maximum acceleration was equal to 228.36 times on 512 computational nodes.

Features of development and analysis of the simulation model of a multiprocessor computer system

Over the past decade, multiprocessor systems have been applied in computer technology. At present,multi-core processors are equipped not only with supercomputers, but also with the vast majority of mobile devices. This creates the need for students to learn the basic principles of their construction and functioning.One of the possible methods for analyzing the operation of multiprocessor systems is simulation modeling.Its use contributes to a better understanding of the effect of workload and structure parameters on performance. The article considers the features of the development of the simulation model for estimating the time characteristics of a multiprocessor computer system, as well as the use of the regenerative method of model analysis. The characteristics of the software implementation of the inverse kinematics solution of the robot are adopted as a workload. The given task consists in definition of turns in joints of the manipulator on known angular and linear position of its grasp. An analytical algorithm for solving the problem was chosen, namely, the method of simple kinematic relations. The work of the program is characterized by the presence of parallel calculations, during which resource conflicts arise between the processor cores, involved in simultaneous access to the memory via a common bus. In connection with the high information connectivity between parallel running programs, it is assumed that all processing cores use shared memory. The simulation model takes into account probabilistic memory accesses and tracks emerging queues to shared resources. The collected statistics reveal the productive and overhead time costs for the program implementation for each processor core involved. The simulation results show the unevenness of kernel utilization, downtime in queues to shared resources and temporary losses while waiting for other cores due to information dependencies. The results of the simulation are estimated by the regenerative method, which allows determining the average time spent searching for memory access in queues and the confidence intervals of these values for various degrees of trust. The given approach to the construction of the simulation model of a multiprocessor computer system and its analysis can be used to analyze the functioning of parallel computing systemsand for educational purposes for teaching students at the courses “Computer Systems” and “Simulation Modeling”.

Methods of Assessing the Characteristics of the Multiprocessor Computer System Adaptation Unit

Currently, a growing number of satellite communication systems developers and manufacturers pay attention to the adaptive multibeam antenna systems. These systems allow you to implement flexible traffic management by forming a multi-beam patterns within the service area. In addition, they have such an important property, as the high noise immunity when exposed as a deliberate and industrial noise provided by the use of spatial filtering methods for interference. The adaptive multibeam antenna can be built on the basis of a phased antenna arrays with scanning beams and multibeam antennas are constructed on the basis of hybrid reflector antennas [1]. The main control element of such antennas is a block adaptation that implements control algorithms and individual beams beamforming. A block adaptation is a processor based on the adaptive signal processor, and the adaptation is made on the program level. In [2] made

Математическое моделирование процессов эвтрофикации в мелководных водоемах на многопроцессорной вычислительной системе

А.И. Сухинов1, А.В. Никитина2, А.Е. Чистяков2, И.С. Семенов3, А.А. Семенякина3, Д.С. Хачунц2 1Донской государственный технический университет (344000 Ростов-на-Дону, пл. Гагарина, д. 1), 2Научно-исследовательский институт многопроцессорных вычислительных систем имени академика А.В. Каляева Южного федерального университета (347928 Таганрог, ул. Чехова, д. 2), 3ОАО «Научно-исследовательский центр суперЭВМ и нейрокомпьютеров» (347900 Таганрог, пер. Итальянский, д. 106) E-mail: sukhin@gmail.com, nikitina.vm@gmail.com, cheese_05@mail.ru, flanker555@yandex.ru, j.a.s.s.y@mail.ru, diana-hachunts@mail.ru. A.I. Sukhinov1, A.V. Nikitina2, A.E. Chistyakov2, I.S. Semenov3, A.A. Semenyakina3, D.S. Khachunts2 1Don State Technical University (Gagarina Sq. 1, Taganrog, 344000 Russia), 2Kalyaev Scientific Research Institute of Multiprocessor Computer Systems at Southern Federal University (Chekhova 2, Taganrog, 347928 Russia), 3Scientific Research Center of Supercomputers and Neurocomputers (per. Italyansky 106, Taganrog, 347900 Russia), E-mail: sukhin@gmail.com, nikitina.vm@gmail.com, cheese_05@mail.ru, flanker555@ yandex.ru, j.a.s.s.y@mail.ru, diana-hachunts@mail.ru

On the partial synchronization of iterative methods

The rapidly growing field of parallel computing systems promotes the study of parallel algorithms, with the Monte Carlo method and asynchronous iterations being among the most valuable types. These algorithms have a number of advantages. There is no need for a global time in a parallel system (no need for synchronization), and all computational resources are efficiently loaded (the minimum processor idle time). The method of partial synchronization of iterations for systems of equations was proposed by the authors earlier. In this article, this method is generalized to include the case of nonlinear equations of the form x = F(x), where x is an unknown column vector of length n, and F is an operator from ℝ n into ℝ n . We consider operators that do not satisfy conditions that are sufficient for the convergence of asynchronous iterations, with simple iterations still converging. In this case, one can specify such an incidence of the operator and such properties of the parallel system that asynchronous iterations fail to converge. Partial synchronization is one of the effective ways to solve this problem. An algorithm is proposed that guarantees the convergence of asynchronous iterations and the Monte Carlo method for the above class of operators. The rate of convergence of the algorithm is estimated. The results can prove useful for solving high-dimensional problems on multiprocessor computational systems.

Problems of Parallel Solution of Large Systems of Linear Algebraic Equations

The paper considers some modern problems arising in developing parallel algorithms for solving large systems of linear algebraic equations with sparse matrices occurring in mathematical modeling of real-life processes and phenomena on a multiprocessor computer system (MCS). Two main requirements to methods and technologies under consideration are fast convergence of iterations and scalable parallelism, which are intrinsically contradictory and need a special investigation. The paper analyzes main trends is developing preconditioned iterative methods in Krylov’s subspaces based on algebraic domain decomposition and principles of their program implementation on a heterogeneous MCS with hierarchical memory structure.

Necessary and sufficient conditions for t-diagnosability of multiprocessor computer systems for various models of nonreliable testing established using the system graph-theoretical model

Consideration was given to the graph-theoretical model of self-diagnosis at the system level under multiple permanent faults. A group of testing models based on complete unreliable tests was analyzed. For the models of this group, the necessary and sufficient conditions for t-diagnosability without repair were established. A critical review of the results obtained in the publications of various authors was given, and the noticed errors concerning the proof of the conditions for t-diagnosability without repair were corrected.

Self-Configurable FPGA-Based Computer Systems: Basics and Proof of Concept

Computer systems performance is today improved with two major approaches: general-purpose computer computing power increase (creation of multicore processors, multiprocessor computer systems, supercomputers), and adaptation of the computer hardware to the executed algorithm (class of algorithms). The last approach often provides application of hardware accelerators – ASIC-based and FPGA-based, also named reconfigurable, and is characterized by better performance / power consumption ratio and lower cost as compared to the general-purpose computers of equivalent performance. However, such systems have typical problems. The ASIC-based accelerators: 1) they are effective only for certain classes of algorithms; 2) for effective application there is a need to adapt algorithms and software. The FPGA-based accelerators and reconfigurable computer systems (that use FPGAs as a processing units): 1) the need in the process of writing a program to perform computing tasks balancing among the general-purpose computer and FPGA; 2) the need of designing applicationspecific processors soft-cores; and 3) they are effective only for certain classes of problems, for which applicationspecific processors soft-cores were previously developed. This paper covers the scope of questions regarding concept of design, architecture, and proof of concept of the Self-Configurable FPGA-Based Computer Systems – an emerging type of high-performance computer systems, which are deprived of specified challenges. The method of information processing in reconfigurable computer systems and its improvements that allow an information processing efficiency to increase are shown. These improvements are used as a base for creating a new type of high-performance computer systems with reconfigurable logic, which are named self-configurable ones, and a new method of information processing in these systems. The structure of self-configurable FPGA-based computer system, the rules of application of computer software and hardware means necessary for these systems implementation are described. Major processes on the stages of program loading and execution in the self-configurable computer system are studied, and their durational characteristics are determined. On the basis of these characteristics, the expressions for evaluating the program execution duration in the self-configurable computer system are obtained. The directions for further works are discussed.

Organization basics and timing characteristics of the multiprocessor self-configurable computer systems

Organization basics and timing characteristics of the multiprocessor self-configurable computer systems

Численное моделирование экологического состояния Азовского моря с применением схем повышенного порядка точности на многопроцессорной вычислительной системе

Численное моделирование экологического состояния Азовского моря с применением схем повышенного порядка точности на многопроцессорной вычислительной системе

On partial synchronization of iterative methods

The rapidly growing field of parallel computing systems promotes the study of parallel algorithms, with the Monte Carlo method and asynchronous iterations being among the most valuable types. These algorithms have a number of advantages. There is no need for a global time in a parallel system (no need for synchronization), and all computational resources are efficiently loaded (the minimum processor idle time). The method of partial synchronization of iterations for systems of equations was proposed by the authors earlier. In this article, this method is generalized to include the case of nonlinear equations of the form x = F(x), where x is an unknown column vector of length n, and F is an operator from ℝn into ℝn. We consider operators that do not satisfy conditions that are sufficient for the convergence of asynchronous iterations, with simple iterations still converging. In this case, one can specify such an incidence of the operator and such properties of the parallel system that asynchronous iterations fail to converge. Partial synchronization is one of the effective ways to solve this problem. An algorithm is proposed that guarantees the convergence of asynchronous iterations and the Monte Carlo method for the above class of operators. The rate of convergence of the algorithm is estimated. The results can prove useful for solving high-dimensional problems on multiprocessor computational systems.

Parameter-free method for computing the turbulent flow in a plane channel in a wide range of Reynolds numbers

Near-wall turbulence in a plane channel at high Reynolds numbers is studied by applying direct numerical simulation. A new numerical algorithm is proposed in which convective fluxes are explicitly approximated using the CABARET scheme and two grid elliptic equations are solved in order to satisfy the incompressibility condition. The equations are of high dimension and are solved by employing a fast direct method that can be efficiently parallelized. In contrast to most methods, including spectral ones, the CABARET scheme does not involve any tuning parameters. It has a compact stencil, which simplifies setting boundary conditions and improves the efficiency of parallelization in computations on multiprocessor computer systems. Numerical results are obtained in a wide range of Reynolds numbers and are compared with experimental data and numerical results of other authors. The error in the computed drag coefficient for turbulent flow is examined as a function of grid characteristics.

The Reliability of Subgraphs in the Arrangement Graph

As the size of a multiprocessor computer system grows, the probability of having faulty (i.e., malfunctioning or failing) processors in the system increases. It is then important to quantify how the faults collectively affect the entire system. The reliability of subsystems in a system, defined as the probability that a fault-free subsystem of a certain size still exists when the system has faults, is a measure for the faults' effect on the whole system. It can be used as an indicator of system health. In this paper, we will present two schemes to calculate the reliability of an <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex Notation="TeX">$(n-1,k-1)$</tex></formula> -subgraph in the <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$(n,k)$</tex></formula> -Arrangement Graph <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$A_{n,k}$</tex> </formula> , an extensively studied interconnection network proposed for multiprocessor computers. The first scheme will use a probability fault model and the Principle of Inclusion-Exclusion to establish an upper-bound of the reliability, by taking into account the intersection of not more than three subgraphs. The second scheme uses basically the same idea, but completely neglects the intersection among subgraphs to calculate an approximate reliability. The results of the two schemes are compared, and are shown to be in good agreement, especially as the single-node reliability <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$p$</tex></formula> goes low.