Multiprocessor Computer Systems Research Articles

Organization of parallel-pipeline calculations for modeling the pollutant distribution process in a reservoir

One of the most acute problems for today is the water pollution. For rapid decision in emergency situations, it is necessary to develop effective software and algorithmic tools that allow us to make accurate forecasts of the environmental situation changing of coastal systems. Water pollution of the Azov and Black Seas by storm drains and human waste products leads to an increase of toxic substances concentrations that significantly exceed the maximum permissible values. The pollution transport problem is solved on the basis of the Navier-Stokes and the diffusion-convection-reaction equations. As a result of discretization of the continuous problem of transport of pollutants using the finite-difference approach for a rectangular grid, we obtain a system of linear algebraic equations (SLAE) of large dimension, which require significant time costs. To increase the efficiency of calculations (to reduce the time) on a multiprocessor computer system (MCS), there is a need to develop effective parallel algorithms for solving SLAE. The decomposition method for a two-dimensional computational domain is proposed in this paper, which allows organizing a parallel-pipeline process of calculations as follows: at each stage of calculations, each processor core simultaneously processes fragments of the computational domain that are offset from each other. This process is described in the form of a graph, in which each node corresponds to fragments of the computational domain, and the edges – a sign of the adjacency of fragments.

Parallel algorithms for solving diffusion-convection problems on a multiprocessor computer system using hybrid MPI / OpenMP technology

The paper deals with the mathematical models, algorithms and software for mathematical modeling of coastal systems’ water pollution spreading dynamics under various unfavorable phenomena of natural and artificial genesis, developed for high-performance cluster systems. Methods for partitioning the computational domain for solving diffusion-convection problems have been developed, which allow for efficient parallelization of a computationally complex modeling problem, taking into account the architecture of the multiprocessor system used. The developed mathematical models are based on high-precision models of hydrophysics and hydrobiology and take into account the peculiarities of water systems in the south of the Rostov region, as well as factors of hydrobiological dynamics such as microturbulent diffusion and advective transport in various directions, mechanisms of primary and secondary pollution of coastal systems, taking into account currents. The paper presents algorithms for solving a simulated problem based on MPI parallelization technology, as well as based on mixed MPI + OpenMP technology. Numerical experiments have been carried out and the two technologies efficiency comparison has been made in the conditions of computing cluster used.

Real-time simulation of accidental passive transport of radioactive pollutant from a proposed nuclear power plant

Power plant’s site selection is a complex task and involves through analyses of multi-disciplinary processes which are interlinked with each other. The site selection for nuclear power plants additionally requires an assessment of radiation doses to the environment and public during normal operation and in the case of an accident. This demands the problem of radioactive particles’ dispersion in atmosphere to be analysed in real time for a comprehensive set of radioactive release scenarios in prevailing meteorological conditions in the plant surroundings. In this study, a local scale atmospheric dispersion problem, considering a hypothetical accidental release (1 Bq s−1 of I-131) from a nuclear power plant is simulated with a combination of weather forecasting and particle dispersion codes on a multiprocessor computer system. The meteorological parameters are predicted with a weather research and forecasting (WRF) model and used in Lagrangian particle dispersion model based code FLEXPART to calculate the trajectory of released particles, and thereby, the estimation of spatial I-131 dose distribution. The concentration of particles and radiation doses were calculated for release heights of 10, 57, and 107 m and found in a reasonable agreement with the observed data and better than an earlier investigation done with regional atmospheric modelling system (RAMS) code. A comparison between the results of WRF and RAMS for various meteorological parameters revealed that better space–time predictions of wind speeds and directions by WRF had a profound effect on tracing the trajectories of particles and thereby the spatial dose distribution. The particles followed the changes in wind direction predicted by WRF that were known to prevail in the region.

Load - aware intelligent multiprocessor scheduler for time-critical cyber-physical system applications

The cyber-physical system involves domains like physical, communication, and cyber system. The time-critical tasks in the physical domain demand a quick computational response from the cyber system. To meet such critical response times, the cyber system uses a multiprocessor computing system to execute the task. But the performance of such a computing system depends on the multiprocessor task scheduler and load-state of the multiprocessor. If the scheduler is not aware of the load-state of the multiprocessor, the scheduling overloads the particular set of processors, and the time-critical task-sets are not able to meet their deadline. To overcome this issue, this paper presents a load-aware multiprocessor scheduling mechanism. The proposed scheduler uses a machine-learning algorithm to efficiently predict the load to make prior awareness of the load and schedule. The simulation results prove that the algorithm outperforms in comparison with the classical multiprocessor scheduler.

Parallel Implementation of the Algorithm to Compute Forest Fire Impact on Infrastructure Facilities of JSC Russian Railways

Forest fires have a negative impact on the economy in a number of regions, especially in Wildland Urban Interface (WUI) areas. An important link in the fight against fires in WUI areas is the development of information and computer systems for predicting the fire safety of infrastructural facilities of Russian Railways. In this work, a numerical study of heat transfer processes in the enclosing structure of a wooden building near the forest fire front was carried out using the technology of parallel computing. The novelty of the development is explained by the creation of its own program code, which is planned to be put into operation either in the Information System for Remote Monitoring of Forest Fires ISDM-Rosleskhoz, or in the information and computing system of JSC Russian Railways. In the Russian Federation, it is forbidden to use foreign systems in the security services of industrial facilities. The implementation of the deterministic model of heat transfer in the enclosing structure with the complexity of the algorithm O (2N2 + 2K) is presented. The program is implemented in Python 3.x using the NumPy and Concurrent libraries. Calculations were carried out on a multiprocessor cluster in the Sirius University of Science and Technology. The results of calculations and the acceleration coefficient for operating modes for 1, 2, 4, 8, 16, 32, 48 and 64 processes are presented. The developed algorithm can be applied to assess the fire safety of infrastructure facilities of Russian Railways. The main merit of the new development should be noted, which is explained by the ability to use large computational domains with a large number of computational grid nodes in space and time. The use of caching intermediate data in files made it possible to distribute a large number of computational nodes among the processors of a computing multiprocessor system. However, one should also note a drawback; namely, a decrease in the acceleration of computational operations with a large number of involved nodes of a multiprocessor computing system, which is explained by the write and read cycles in cache files.

The problem of choosing the optimal performance in the multiprocessor computer system design

The article presents the formulation of the problem of optimizing the structure of multiprocessor computing systems designed to solve control problems in real time. The features of this problem, which influence the choice of optimization methods, have been studied. It is concluded that this problem can be effectively solved using evolutionary methods of optimization. The considered models for finding performance can be used to optimize the architecture of multiprocessor computing systems. Besides, it should be taken into account that the resources allocated for the development and operation of computing systems are always limited. Therefore, it is advisable to consider the problem of optimizing the structure of a computing system as multi-criterial one: one criterion is the performance, and the other one is the cost of development and operation of the system. Acquired results can be used in development of multiprocessor computing systems for real-time systems, which is going to reduce the cost of development and operation of these control systems.

Subnetwork reliability analysis of bubble-sort graph networks

The bubble-sort graph network Bn is recognized as an attractive interconnection network topology for building multiprocessor computer systems. In this paper, the subnetwork reliability of Bn is analyzed in the presence of node failures. An upper bound and a lower bound on the Bn−1 subnetwork reliability of Bn are established under the probability fault model, and the theoretical results are proved to be in accordance with and near to the simulation results. Moreover, under the node fault model, the mean time to failure to maintain the fault-free status of different number of disjoint Bn−1 subnetworks in Bn is evaluated by using the fixed partitioning and the flexible partitioning, respectively, and it is shown that Bn has better robustness under the flexible partitioning pattern when disjoint Bn−1 subnetworks need to be used.

Reliability of computer structures of integrated modular avionics for hardware configurations

The problem of designing advanced computing systems in the class of structures of integrated modular avionics is considered. The unified topology of the internal network of the computer on the basis of Space Wire exchange channels and variants of its execution for various onboard applications is offered. Equivalent reliability schemes of each of the specific structures are introduced and the probabilities of trouble-free operation of each structure are analyzed. Families of graphic dependencies are given. The analysis of the existing principles and algorithms for testing multiprocessor multimodal onboard digital computer systems is given; the new testing algorithm for the multiprocessor systems which follows the software design standards for products of integrated modular avionics is offered. The structure of the unified automated workplace for checking the functional modules of integrated modular avionics is considered. Specific requirements inherent in the workplaces for testing integrated avionics are identified: an increased level of control of the hardware component of products; the ability to simulate the failure state of individual components of avionics to check the mode of reconfiguration of the computer system; modular construction of software with the division of verification tests into components performed at the level of each CPM and the computer as a whole in single-task and multitasking modes; openness of architecture of a workplace, which provides an ability to change the level of control complexity of a product and control of one class of complexity; intra-project unification of both hardware and software of the workstation of the inspection.

A C++ expression system for partial differential equations enables generic simulations of biological hydrodynamics

We present a user-friendly and intuitive C++ expression system to implement numerical simulations of continuum biological hydrodynamics. The expression system allows writing simulation programs in near-mathematical notation and makes codes more readable, more compact, and less error-prone. It also cleanly separates the implementation of the partial differential equation model from the implementation of the numerical methods used to discretize it. This allows changing either of them with minimal changes to the source code. The presented expression system is implemented in the high-performance computing platform OpenFPM, supporting simulations that transparently parallelize on multi-processor computer systems. We demonstrate that our expression system makes it easier to write scalable codes for simulating biological hydrodynamics in space and time. We showcase the present framework in numerical simulations of active polar fluids, as well as in classic simulations of fluid dynamics from the incompressible Navier–Stokes equations to Stokes flow in a ball. The presented expression system accelerates scalable simulations of spatio-temporal models that encode the physics and material properties of tissues in order to algorithmically study morphogenesis.Graphicabstract

Development and Optimisation of a DNS Solver Using Open-source Library for High-performance Computing

We develope an optimised Direct Numerical Simulation (DNS) solver for laminar, transition, and turbulent flow. It utilises the Crank Nicholson scheme for time advancement, and it is spatially accurate up to fourth-order on uniform grids. The solver's salient features are multi-dimensional parallelisation and modular-based code, which is written in object-oriented C++ language. The DNS solver is developed in the MPI parallel platform using a cubic decomposition technique. The open-source library used for optimisation is MPI, blitz++, YAML, and HDF5 library. The parallel performance analysis of the optimised DNS solver on the multiprocessor computing system (supercomputer) shows 80 efficiency. The solver is verified and benchmarked for wall-bounded flows by simulation of laminar, transition, and turbulent flow in a square duct.

SIMULATION OF MICROBIOLOGICAL DESTRUCTION PROCESS OF OIL POLLUTION IN SHALLOW WATER

The paper covers the research of microbiological destruction processes of petroleum origin pollutants in a shallow water taking into account a number of determining factors that affect the distribution of pollution in the researched water area: the oil fractional composition, the processes of evaporation, dissolution, biological oxidation of petro-hydrocarbons by microorganisms, as well as hydrodynamic and chemical-biological features of the water. A complex of interrelated spatially inhomogeneous mathematical models was proposed that allow researching the dynamics of microbiological destruction processes of petroleum hydrocarbons in shallow water. Schemes with weight taking into account the partial filling of computational cells of the simulated domain were developed at model discretization that made it possible to significantly increase the accuracy of calculations and reduce the calculation time. Based on a multiprocessor computer system, experimental software has been developed for predictive modeling of the ecological situation of shallow water in the event of accidental pollution by oil and other harmful substances during natural and technogenic challenges.

The Efficiency of Discrete Optimization Algorithm Portfolios

Introduction. Solving large-scale discrete optimization problems requires the processing of large-scale data in a reasonable time. Efficient solving is only possible by using multiprocessor computer systems. However, it is a daunting challenge to adapt existing optimization algorithms to get all the benefits of these parallel computing systems. The available computational resources are ineffective without efficient and scalable parallel methods. In this connection, the algorithm unions (portfolios and teams) play a crucial role in the parallel processing of discrete optimization problems. The purpose. The purpose of this paper is to research the efficiency of the algorithm portfolios by solving the weighted max-cut problem. The research is carried out in two stages using stochastic local search algorithms. Results. In this paper, we investigate homogeneous and non-homogeneous algorithm portfolios. We developed the homogeneous portfolios of two stochastic local optimization algorithms for the weighted max-cut problem, which has numerous applications. The results confirm the advantages of the proposed methods. Conclusions. Algorithm portfolios could be used to solve well-known discrete optimization problems of unprecedented scale and significantly improve their solving time. Further, we propose using communication between algorithms, namely teams and portfolios of algorithm teams. The algorithms in a team communicate with each other to boost overall performance. It is supposed that algorithm communication allows enhancing the best features of the developed algorithms and would improve the computational times and solution quality. The underlying algorithms should be able to utilize relevant data that is being communicated effectively to achieve any computational benefit from communication. Keywords: Discrete optimization, algorithm portfolios, computational experiment.

A path relinking enhanced estimation of distribution algorithm for direct acyclic graph task scheduling problem

Superior task scheduling scheme is able to improve the performance in achieving shorter task completion time in multi-processor computing system. Large scale applications are generally modelled as direct acyclic graph (DAG) to be processed efficiently in parallel. To solve DAG task scheduling problem (DAG-SP) with the criterion of minimizing makespan, this paper proposes an estimation of distribution algorithm (EDA) enhanced by the path relinking. An efficient hybrid scheme integrating list scheduling heuristics is designed to take advantage of the knowledge of existing works. In addition, to describe the relative position relationships between the task pairs, a specific probability model is built and the task processing permutations are produced by sampling such a model. To enhance the exploitation of EDA, a path relinking based knowledge is used to design the local search method. Simulation experiments are carried out with both benchmark datasets and real-world graphs, where the comparative results show that the above designs can improve the performance effectively. Moreover, the numerical comparisons show that the proposed algorithm performs significantly better than the existing heuristics and evolutionary algorithms.

EFFICIENT ALGORITHMS FOR PARALLELIZING TRIDIAGONAL SYSTEMS OF EQUATIONS

The article is devoted to the development of the maximal parallel forms of mathematical models with a tridiagonal structure. The example of solving the Dirichlet and Neumann problems by the method of straight lines and the sweep method for the heat equation illustrates the direct fundamental features of constructing parallel algorithms. It is noted that the study of the heat and mass transfer processes is run through their numerical modeling based on modern computer technology.It is shown that with the multiprocessor computing systems’ development, there disappear the problems of increasing their peak performance. On the other hand, building such systems, as a rule, requires standard network technologies, mass-produced processors, and free software. The noted circumstances aim at solving the so-called big problems.It should be borne in mind that the classical approach to solving the tridiagonal structure models based on multiprocessor computing systems is far more time-consuming compared to single-processor computing facilities. That is explained by the recurrence relations that make the basis of classical methods. Therefore, the proposed studies are relevant and aim at the distributed algorithms development for solving applied problems.The proposed research aims to construct the maximal parallel forms of mathematical models with a tridiagonal structure.The paper proposes the schemes to implement parallelization algorithms for applied problems and their mapping to parallel computing systems.Parallelization of tridiagonal mathematical models by the method of straight lines and the sweeping method allows designing absolutely stable algorithms with the maximum parallel form and, therefore, the minimum possible time for their implementation on parallel computing devices. It is noteworthy that in the proposed algorithms, the computational errors of the input data are separated from the round-off errors inherent in a PC.The proposed approach can be used in various branches of metallurgical, thermal physics, economics, and ecology problems in the metallurgical industry.

Mathematical modeling of zooplankton productivity in the Azov Sea in summer on a high-performance computer system

The work is devoted to modeling of eurythermal and stenothermal zooplankton populations productivity in the Azov Sea in summer. The discretization of the problem was carried out using difference schemes of a higher order of accuracy, taking into account the filling of the calculated cells. The grid equations obtained as a result of the proposed mathematical model of biological kinetics discretization were solved by a modified adaptive variable-triangular method, which has a higher convergence rate in comparison with the existing methods of the variational type. A parallel algorithm for solving the problem on a high-performance computing system was developed, which made it possible to significantly reduce the time of the numerical solution of the problem. With the help of mathematical modeling, the analysis of various ways of development of the Azov Sea ecosystem was carried out, the influence of the gelatinous invasive on the production and de-struction processes of phyto- and zooplankton was researched. The developed software module, oriented to a multiprocessor computing system, can be effectively used for predictive modeling of interrelated hydrodynamic and biogeochemical processes in a shallow water body in summer.

Solving the diffusion-convection problem using MPI parallel computing technology

The aim of the work is to build a software package for the distributed solution of the problem of matter transfer in a reservoir. A parallel implementation of the grid domain decomposition methods for computationally complex diffusion-convection problems, taking into account the architecture and parameters of a multiprocessor computing system located on the basis of the infrastructure facility of the University of Science and Technology «Sirius», is considered. An algorithm for parallel solution of the formulated hydrophysics problem using MPI technology has been developed. The analysis of the efficiency of the developed parallel algorithms is carried out and the study of the obtained solutions of computationally complex problems of aquatic ecology is carried out.

RESEARCH OF THE INFLUENCE OF A NETWORK INTERFACE ON THE EFFICIENCY OF MODULAR MULTIPROCESSOR SYSTEMS

The paper is devoted to the approach development related to methodology definition for evaluation of the modular multiprocessor computing systems efficiency. At the same time, the main attention is focused on the impact peculiarities on this network interface value. The formation analysis of the multiprocessor system network interface architecture and the basic modes of its operation have been analyzed. To evaluate the processes occurring in the system during the information flows transmission, the network system bandwidth and the switch throughput were compared; which allowed determining the preconditions for optimal components selection of the multiprocessor computing system network interface. The performed researches also allowed deducing analytical relations for determining the optimal number of system nodes with different functioning modes. The selected processors coherency coefficient, network interface and value of the computing area are deduced. The derived analytical relationships also showed that the optimal number of blades in a multiprocessor computing system, that provide its highest speed, decreases with increasing computing power of the processors included. It is shown that the network data interchange among the multiprocessor computing system nodes the more likely to impede the overall computation process; the less time will be spent directly on solving a specific problem.

NUMERICAL IMPLEMENTATION OF A PARALLEL ALGORITHM FOR SOLVING THE PROBLEM OF POLLUTANT TRANSPORT IN A RESERVOIR ON A HIGH-PERFORMANCE COMPUTER SYSTEM

The purpose of this work is to create a software package for a distributed solution of the problem of transporting a pollutant in a reservoir with complex bathymetry and the presence of technological structures. An algorithm has been developed for the parallel solution of the problem of transporting a pollutant (pollutant) in a reservoir on a graphics accelerator controlled by the CUDA (Compute Unified Device Architecture) system; a comparative analysis of the operation of algorithms on a CPU (Central Processing Unit) and on a graphics accelerator GPU (Graphics Processing Unit) made it possible to evaluate their performance. The software implementation of the modules included in the complex is described, the main classes and implemented methods are documented. The results of numerical experiments showed that solving of pollutant transport’s problem based on the CUDA technology is ineffective for small grids (up to 100 ´ 100 computational nodes). In the case of large grids (1000 ´ 1000 computational nodes), the use of CUDA technology reduces the computation time by an order of magnitude. An analysis of the experiments carried out with the developed components of software showed that the maximum value of the ratio of the algorithm operating time that implements the set task of transferring matter in a shallow water on a GPU to the operating time of a similar algorithm on the CPU was 24.92 times, which is achieved on a grid of 1000 ´ 1000 computational nodes. Implementation of methods for decomposition of grid regions is proposed for solving computationally laborious problems of diffusion-convection, including the problem of transporting pollutants in a reservoir with complex bathymetry with technological objects that take into account the architecture and parameters of a MSC (Multiprocessor Computing System) located on the basis of the infrastructure facility of the STU (Scientific and Technological University) “Sirius” (Sochi, Russia). Consideration was made for such a property of a computing system as the time it takes to transmit and receive floating point data. An algorithm for the parallel solution of the task under the control of MPI (Message Passing Interface) technology has been developed, and its efficiency has been assessed. The acceleration values of the proposed algorithm are obtained depending on the number of involved computers (processors) and the size of the computational grid. The maximum number of computers used is 24, the maximum size of the computational grid was 10 000 ´ 10 000 computational nodes. The developed algorithm showed low efficiency for small computational grids (up to 100 ´ 100 computational nodes). In the case of large computational grids ( from 1000  1000 computational nodes), the use of MPI reduces the computation time by several times.

RESEARCH OF COMPUTING EFFICIENCY IN MODULAR MULTIPROCESSOR SYSTEMS

The article is devoted to the research of efficiency of a multiprocessor computing system in solving problems aimed at expanding the computing area. The basic regularities concerning the time of solving the problem are revealed, depending on the change in the multiprocessor system calculations area. The research is aimed at determining the deceleration factor associated with the increase of the computing area of a multiprocessor system when compared with the computer version with an unlimited computing area. The analytical ratios are derived for determining the calculations deceleration coefficient. A stage of simulation for calculations of the deceleration factor was carried out to determine the regularities of its change, depending on the application of a particular computing platform. The revealed tendencies of such a change point to the need to reconcile the components of the network interface and computing capabilities of the chosen computing platform. The derived analytical relations were aimed at determining the optimal number of nodes of a multiprocessor system which allow the minimum delay of calculations.

Numerical implementation of biological kinetics multi-species model in the Azov Sea

Paper covers the research of biological kinetics processes based on a multi-species model of plankton and fish interaction of the Azov Sea at low and high size considering the V erhulst and Allee effects, competition for resources, taxis, catching, spatial distribution of biogenic matter and detritus. Discrete analogue of developed model problem of water ecology, included in a software complex, were calculated using schemes of increased order of accuracy considering the partial filling of computational cells. The system of grid equations of large dimension, arising at discretization, has been solved on the basis of a two-layer variational type method – the minimum corrections method having the maximum convergence rate. Effective parallel algorithms were developed for numerical implementation of biological kinetics problem and oriented on multiprocessor computer system and NVIDIA Tesla K80 graphics accelerator with the data storage format modification of. Due to it, the reproduction processes of biogeocenose populations have been analysed in real and accelerated time.