High-speed Parallel Computation Research Articles

HIGH-PERFORMANCE MODELING, IDENTIFICATION AND ANALYSIS OF HETEROGENEOUS ABNORMAL NEUROLOGICAL MOVEMENT’S PARAMETERS BASED ON COGNITIVE NEURO FEEDBACK-INFLUENCES

The technique is based on a hybrid model of the neuro-system (Brain cortex nodes and tremor-object), which describes on the basis of wave signal propagation the state and behavior of tremor-objects T, namely the segmental de-scription of 3D elements of trajectories of anormal neurological movements of the studied tremor-objects (limb of the hand) taking into account the matrix of cogni-tive influences of groups of cortex neuro-nodes. The rapid analytical solution of the model as a vector function that describes the 3D elements of the trajectories at each movements segment are constructed using the hybrid integral Fourier’s transformations and hybrid spectral function. The main element of the solution is the adaptive infuences matrix that determines the state parameters of the action of certain groups of brain neuro- cortex. Models and methods of multivariable identification are being developed to investigate their neuro-feedback, which suggest high-speed parallel computations on multicore computers. This model-ing technology consider as a scientific basis for designing inelidgence information systems of the quality medical diagnostic i of critical neurological diseases.Key words: Computer simulation, Software system, High-performance compu-ting, Tremor diseases, Modeling of objects and processes, Multi-parameter iden-tification.

High-Performance Supercomputer Technologies of Simulation of Nanoporous Feedback Systems for Adsorption Gas Purification

High-performance methods and computer technologies for modeling of non-isothermal gas adsorption in nanoporous cyber-feedback systems based on the Heaviside operating method and Landau’s linearization approach to Langmuir’s nonlinear adsorption equilibrium using decomposition in a convergent series at the temperature transition point with subsequent efficient parallelization of the model over a small parameter are developed. The results of numerical experiments based on high-speed parallel computations on multicore computers are presented.

Reconfigurable Architecture for Multi-lead ECG Signal Compression with High-frequency Noise Reduction

Electrocardiogram (ECG) is a record of the heart’s electrical activity over a specified period, and it is the most popular noninvasive diagnostic test to identify several cardiac diseases. It is an integral part of a typical eHealth system, where the ECG signals are often needed to be compressed for long term data recording and remote transmission. Reconfigurable architecture offers high-speed parallel computation unit, particularly the Field Programmable Gate Array (FPGA) along with adaptable software features. Hence, this type of design is suitable for multi-channel signal processing units like ECGs, which usually require precise real-time computation. This paper presents a reconfigurable signal processing unit which is implemented in ZedBoard- a development board for Xilinx Zynq −7000 SoC. The compression algorithm is based on Fast Fourier Transformation. The implemented system can work in real-time and achieve a maximum 90% compression rate without any significant signal distortion (i.e., less than 9% normalized percentage of root-mean-square deviation). This compression rate is 5% higher than the state-of-the-art hardware implementation. Additionally, this algorithm has an inherent capability of high-frequency noise reduction, which makes it unique in this field. The confirmatory analysis is done using six databases from the PhysioNet databank to compare and validate the effectiveness of the proposed system.

ВИКОРИСТАННЯ ТЕХНОЛОГІЇ ІНТЕРАКТИВНОЇ КЛАСИФІКАЦІЇ ГЕОЛОГІЧНИХ ТІЛ ДЛЯ ПРОГНОЗУВАННЯ ПОКЛАДІВ ГАЗУ НА СХОДІ УКРАЇНИ

The possibilities of allocation of geological bodies with certain physical and filtration-capacitive properties on the basis of classification by a set of seismic attributes are considered. This classification is based on the use of computer technology with parallel computing on graphic processors. High-speed parallel computing provides the ability to interactively classify and get results in real-time. At the same time, application of graphic processors allows to realize technologies of interactive classification not only on computing clusters, but also on personal computers. Geological efficiency of technologies of interactive classification is shown. Their application allows to detect geological bodies with certain physical properties on the basis of computer analysis of three-dimensional arrays of seismic data, in particular seismic images and arrays of seismic attributes. It is important to note that computer technologies of the interactive classification of three-dimensional seismic data not only provide a high speed of determination of the spatial position and properties of geological bodies, but also allow to realize the fundamental possibility of such a definition. Traditional approaches to the identification and classification of geological bodies are based on a sequential analysis of sections of three-dimensional seismic data. In this case, the solution of the problem of the identification and classification of geological bodies often encounter the fundamental problems due to the complexity of the visual assessment of three-dimensional geological objects based on the analysis of the sequence of two-dimensional sections of the arrays of seismic data. The authors propose a convenient approach to systematizing methods of interactive classification of geological bodies by one and several seismic attributes. The known technologies of bright spots and AVO are considered as elements of a sequence of methods of interactive classification using different numbers of seismic attributes. According to the results of 3-D seismic survey carried out on the areas of the east of Ukraine, the classification of geological bodies using one and two seismic attributes was performed. Some objects with perspectives concerning presence of hydrocarbon were found and analyzed. The conclusion of classification expediency using several parameters is done. It creates preconditions for realization of more perfect and versatile approaches to the detection of geological bodies with given physical properties.

Interactive 3D simulation for fluid–structure interactions using dual coupled GPUs

The scope of this work involves the integration of high-speed parallel computation with interactive, 3D visualization of the lattice-Boltzmann-based immersed boundary method for fluid–structure interaction. An NVIDIA Tesla K40c is used for the computations, while an NVIDIA Quadro K5000 is used for 3D vector field visualization. The simulation can be paused at any time step so that the vector field can be explored. The density and placement of streamlines and glyphs are adjustable by the user, while panning and zooming is controlled by the mouse. The simulation can then be resumed. Unlike most scientific applications in computational fluid dynamics where visualization is performed after the computations, our software allows for real-time visualizations of the flow fields while the computations take place. To the best of our knowledge, such a tool on GPUs for FSI does not exist. Our software can facilitate debugging, enable observation of detailed local fields of flow and deformation while computing, and expedite identification of ‘correct’ parameter combinations in parametric studies for new phenomenon. Therefore, our software is expected to shorten the ‘time to solution’ process and expedite the scientific discoveries via scientific computing.

Numerical Modeling and High-Speed Parallel Computing: New Perspectives on Tomographic Microwave Imaging for Brain Stroke Detection and Monitoring

This article deals with microwave tomography for brain stroke imaging using state-of-the-art numerical modeling and massively parallel computing. Iterative microwave tomographic imaging requires the solution of an inverse problem based on a minimization algorithm (e.g., gradient based) with successive solutions of a direct problem such as the accurate modeling of a whole-microwave measurement system. Moreover, a sufficiently high number of unknowns is required to accurately represent the solution. As the system will be used for detecting a brain stroke (ischemic or hemorrhagic) as well as for monitoring during the treatment, the running times for the reconstructions should be reasonable. The method used is based on high-order finite elements, parallel preconditioners from the domain decomposition method and domain-specific language with the opensource FreeFEM++ solver.

Efficient Load Balancing Algorithm for the Arrangement-Star Network

The Arrangement-Star is a well-known network in the literature and it is one of the promising interconnection networks in the area of super computing, it is expected to be one of the attractive alternatives in the future for High Speed Parallel Computers. The Arrangement-Star network has many attractive topological properties such as small diameter, low degree, good connectivity, low broadcasting cost and flexibility in choosing the desired network size. Although, some of the research work has been done on Arrangement-Star network, it still needs more investigation and research efforts to explore these attractive topologies and utilize it to solve some real life applications. In this paper we attempt to fill this gap by proposing an efficient algorithm for load balancing among different processors of the Arrangement-Star network. The proposed algorithm is named as Arrangement Star Clustered Dimension Exchange Method ASCDEM presented and implemented on the Arrangement-Star network. The algorithm is based on the Clustered Dimension Exchange Method (CDEM). The ASCDEM algorithm is shown to be efficient in redistributing the load balancing among all different processors of the network as evenly as possible. A complete detail of this algorithm in addition to examples and discussions to explore the benefits of applying this distributed algorithm is presented in this paper. Furthermore an analytical study on the algorithm is presented and discussed to explore the attractive performance of the proposed algorithm.

A Hybrid Optimization Based on GPU Parallel Computing Method and its Application of Three Dimensional Large Eddy Simulations

Using compute Unified device architecture (CUDA), a traditional computational fluid dynamics (CFD) program is paralleled and optimized based on graphic processing unit (GPU). The calculation process is divided into two parts as serial and parallel. Their main characteristics are analyzed and different optimization schemes are given. CPU (central processing unit) and GPU work respectively as flow control and high-speed parallel computation. Bandwidth between devices is applied effectively. Data transfer between devices is moderately improved to simplify algorithm. Finally, the method is verified by simulating a three-dimensional isotropic homogeneous turbulence flow field. The calculation uses large eddy simulation (LES) method with secondary filter and solves the three-dimensional N-S equations. The maximum grid number achieves 8,000,000 and takes 33 seconds each step. All calculations are using ordinary single desktop computer, optimized acceleration ratio can reach 9.

Radiation dosimetry using three-dimensional optical random access memories.

The ability to determine particle type and energy plays an important role in the dosimetry of heavy charged particles (HCP) and neutrons. A new approach to radiation dosimetry is presented, which is shown to be capable of particle type and energy discrimination. This method is based on utilising radiation induced changes in the digital information stored on three-dimensional optical random access memories (3D ORAM). 3D ORAM is a small cube (a few mm3) composed of poly(methyl methacrylate) doped with a photochromic dye. and it was originally proposed as a memory device in high speed parallel computers. A Nd:YAG laser system is used to write and read binary information (bits) on the ORAM, which functions as a charged particle detector. Both the read and the write processes use two laser beams that simultaneously strike the material to cause a colour change at their intersection (similar to the darkening of light-sensitive sunglasses when exposed to sunlight.) The laser produces colour changes in the ORAM, which then reverts to the original colour ('bit-flips') at sites where energy is deposited from interaction with incident HCP or neutron-recoil protons. The feasibility of this approach was demonstrated both theoretically and experimentally. Calculations based on track structure theory (TST) predict that when HCP interact with the ORAM material, the local energy deposition is capable of inducing measurable 'bit-flips'. These predictions were recently confirmed experimentally using two types of ORAM systems, one based on spirobenzopyran and the other on anthracene, as the photochromic dyes.

High Speed Parallel Solving Method for Stiffness Equation by Skewed-Tree Data Transfer (Parallelization and Acceleration of Rigid-Plastic Finite Element Analysis).

Most of the calculation time for large scale numerical simulation which uses rigid-plastic finite element method is occupied by the solver part of simultaneous linear equation called stiffness equation.We incorporated the parallelized conjugated gradient method onto our cluster of workstations to shorten the computation time of them.It is important for high speed parallel computation in our cluster system of many workstations to reduce the data communication time through TPDDI network.In this paper, we propose a new method which enables to reduce data communication time by using the skewed tree multiple data transfer through concurrent or parallel paths from plural slave processors to that of others like systolic pulsations. Then, we achieve about eightfold speedup of execution time by twenty-four slave processors.And, we infer that the method shown in this paper is also effective to elastic finite element method.

Design, fabrication and characterization of high-speed asymmetric Fabry-Perot modulators for optical interconnect applications

Free-space ‘smart-pixel’ optical interconnect architectures promise to relieve the interconnect ‘bottleneck’ in high-speed parallel computers and switching systems. One of the most promising output devices that has been advanced for use in these systems is the asymmetric Fabry-Perot modulator, or AFPM, which offers high on-off contrast, low insertion loss, and low operating voltage swing, among other advantages. In this paper we summarize our work on optimizing the AFPM for high-speed operation, including analysis of the material structure design considerations, fabrication of small (16×20 μm) devices, and high-speed electrical and optical characterization of the finished modulators. We conclude that at relatively high incident optical intensities the modulators' speed appears to be limited by transit effects to about 18 GHz, but that at lower optical intensities their frequency response outstrips that of our 20 GHz measurement apparatus — that is, these AFPMs are still capable of large signal modulation (20 dB contrast, 1.5 dB insertion loss) at low AC voltage swings (±3 V) for operating frequencies up to 20 GHz. We presume that further investigation will prove them to be RC-limited in this low-intensity regime to speeds of about 35 GHz.

The MIMDIX Environment for Parallel Simulation

MIMDIX supports simulation and evaluation of dynamical systems on high speed parallel computers. Version 1.0 provides synchronization facilities at the kernel level for parallel computation. Probabilistic synchronization, proposed by Madisetti, provides the basic mechanism for synchronization that is provably fast, efficient in memory utilization, flexible, and capable of adapting to the dynamic event-driven nature of the computation. MIMDIX has been realized on a 32-processor BBN Butterfly system. This paper describes the design of MIMDIX, its implementation, and initial performance measurements. Initial results show a significant improvement in speedup over current first generation operating systems for parallel simulation.

Geometry-defining processors for engineering design and analysis

Critical issues of interactive three-dimensional geometry definition and high-speed parallel computation are addressed in a unified fashion by Geometry-Defining Processors (GDPs). GDPs are microprocessors housed in three-dimensional physical polyhedral packages which can be easily manually assembled or reconfigured to construct approximate scale models of physical objects or domains. An individual GDP communicates with neighboring GDPs in an assembly through optical ports associated with the faces of its package. An assembly of communicating GDPs is able to bothdefine a system geometry and, operating as an optimally connected parallel processor,solve the associated continuum partial differential equations required for design evaluation. Combining simplicity-of-use with efficient computational capabilities, the GDP design system should prove useful in numberous engineering applications.

Topological Analysis of Multibody Systems for Recursive Dynamics Formulations

Abstract Graph theory is used to characterize the topology of multibody systems for high speed parallel computer dynamic simulation. Decoupled loops are defined for use in the recursive dynamics formulations, wherein Langrange multipliers can be eliminated concurrently within each decoupled loop. A row-scanning process for the nondirected adjacency matrix of a graph is described, to generate an identification array that facilitates compact storage and defines system connectivities. A decision tree search algorithm is employed to generate all spanning trees associated with a closed loop graph. For each spanning tree generated from the decision tree search algorithm, a precedence array is defined. A path matrix for the spanning tree and an outdegree of junction nodes are defined to assist in parallel implementation of recursive dynamics algorithms and are then constructed from the precedence array. Finally, an optimal tree search algorithm that identifies a spanning tree that leads to best solution sequences for parallel implementation of recursive dynamics formulations is presented.

Timing information in transient behavior of neuropopulations

A computational model of neuronal populations based on neurophysiological data is presented. The neuronal population consists of model neurons that simulate electrochemical processes in real neurons. Input and output of the population are defined in terms of the stochastic behavior of the single neurons. Analysis of the populational behavior shows that neuronal populations can approximately function as delay elements that transform step input to step output, and the delay intervals can specify timing information in the several-millisecond range, which is essential to high speed parallel computation. An example network composed of neuronal populations is simulated to show how the timing information can serve to achieve parallel processing in neural networks. >

Intel offers faster line of supercomputers

Intel Scientific Computers' hypercube-based family of concurrent supercomputers entered its second generation last week. The company announced new hardware and software developments that make the iPSC/2 family much simpler to program and up to 10 times faster than the first generation. The high-speed parallel computer systems are used primarily for large-scale scientific and engineering applications. Some of the possible applications are of direct chemical interest. Since introducing the first iPSC generation two years ago (C&EN, Feb. 18, 1985, page 31), Intel has shipped about 70 of the systems worldwide. About 70% are used in research at universities, 20% in defense-related applications at government and defense contractorsites, and 10% in production use. During that period, Intel says it determined that there definitely is a market for cost-effective supercomputing featuring high performance for real-world applications. The iPSC family architecture is based on the hypercube topology developed at Califo...

Transistor applications in a high-speed parallel computer

Transistors are at present the most suitable power-gain devices for use in parallel digital computers on account of their small size, low power consumption and high presumptive reliability. However, the design of the building-block circuits, which are logically interconnected in large numbers to form the computer, must guarantee adequate performance in spite of contemporary transistor disadvantages—their characteristic spreads, restricted frequency response and limited power dissipation.The paper describes how transistors can be used with diode logic to satisfy some fundamental requirements peculiar to binary amplifiers and 2-state elements. Feedback saturation control is generally adopted; building-block circuits incorporating this feature are described. Satisfactory operation at 1 megapulse/sec is obtained using 5Mc/s alloy-junction transistors. Some specific arithmetic and counting applications are discussed.When binary amplifiers fail to give adequate frequency response, or adequate power-handling capacity, because of transistor limitations, master-pulse techniques may provide a solution. These techniques are discussed and their application to the ferrite-core-store drive problem is described.