Parallel Computing Environment Research Articles

Turing machines with two-level memory: New computational models for analyzing the input/output complexity

The input/output complexity, which is the complexity of data exchange between the main memory and the external memory, has been elaborately studied by a lot of former researchers. However, the existing works failed to consider the input/output complexity in a computational model point of view. In this paper we remedy this by proposing four variants of Turing machine that include external memory and the mechanism of exchanging data between main memory and external memory. Based on these new models, the input/output complexity is deeply studied. We discuss the relationship between input/output complexity and the other complexity measures such as time complexity and parameterized complexity, which is not considered by former researchers. We also define the external access trace complexity, which reflects the physical behavior of magnetic disks and gives a theoretical evidence of IO-efficient algorithms. Finally, we generalize the model into parallel computation environments, and discuss the relationship between input/output complexity and communication complexity.

Enhancing GPU Performance for SHA-3 Algorithm: Optimizing Hashing Operations in a Parallel Computing Environment

Software implementation of Hash function have not been able to offer satisfactory performances for various application thus far. Additionally, SHA-3 and SHAKE, which utilize SHA-3, are extensively utilized in many Post Quantum Cryptosystem (PQC) . Therefore, there is a need for research to optimize SHA-3 in the software environments. Our proposal involves an optimized software implementation of SHA-3 on a GPU environment. To improve performance efficiency, we suggest various techniques such as optimizing the internal processes of SHA-3, inline PTX optimization, efficient memory usage, and asynchronous CUDA stream application. After implementing these optimization methods, our SHA-3(512) (and SHA-3(256)) algorithm provides a maximum throughput of 88.51 Gb/s (and 171.62 Gb/s) on the RTX2080Ti GPU without CUDA stream. The proposal aims to optimize the software implementation of SHA-3 in a GPU environment to enhance performance efficiency. The suggested techniques include internal process optimization of SHA-3, inline PTX optimization, efficient memory usage, and the application of an asynchronous CUDA stream. After applying these optimization methods, our SHA-3(512) and SHA-3(256) algorithms provide a maximum throughput of 88.51 Gb/s and 171.62 Gb/s, respectively, on the RTX2080Ti GPU without CUDA stream.

Implementation and performance evaluation of a hierarchical parallel solver for saddle point problems on a GPU cluster

Recently, we have proposed an efficient method for solving saddle point problems using their block structure. Instead of solving an original problem directly, this method solves a linear system with multiple right-hand sides. Then it computes solution vectors of the original problem using a solution matrix of this linear system. Since solving a linear system with multiple right-hand sides is the most time-consuming part of our method, it is desirable to solve that as fast as possible. In this paper, a parallelization strategy of our method is proposed, and our method is implemented on a parallel computation environment. Moreover, to achieve further speed-up of our method, this paper also proposes an implementation strategy using GPUs. Numerical experiments illustrate the parallel performance of our method on a GPU cluster.

Twitter Data Analysis Using Hadoop and ‘R’ and Emotional Analysis Using Optimized SVNN

Standalone systems cannot handle the giant traffic loads generated by Twitter due to memory constraints. A parallel computational environment provided by Apache Hadoop can distribute and process the data over different destination systems. In this paper, the Hadoop cluster with four nodes integrated with RHadoop, Flume, and Hive is created to analyze the tweets gathered from the Twitter stream. Twitter stream data is collected relevant to an event/topic like IPL- 2015, cricket, Royal Challengers Bangalore, Kohli, Modi, from May 24 to 30, 2016 using Flume. Hive is used as a data warehouse to store the streamed tweets. Twitter analytics like maximum number of tweets by users, the average number of followers, and maximum number of friends are obtained using Hive. The network graph is constructed with the user’s unique screen name and mentions using ‘R’. A timeline graph of individual users is generated using ‘R’. Also, the proposed solution analyses the emotions of cricket fans by classifying their Twitter messages into appropriate emotional categories using the optimized support vector neural network (OSVNN) classification model. To attain better classification accuracy, the performance of SVNN is enhanced using a chimp optimization algorithm (ChOA). Extracting the users’ emotions toward an event is beneficial for prediction, but when coupled with visualizations, it becomes more powerful. Bar-chart and wordcloud are generated to visualize the emotional analysis results.

Baryon pasting algorithm: halo-based and particle-based pasting methods

ABSTRACT We present a fast methodology to produce mock observations of the thermal and kinetic Sunyaev–Zel’dovich (SZ) effects based on the dark matter only N-body simulations coupled with the analytic intracluster medium model. The methods employ two different approaches: halo-based pasting (HP) and particle-based pasting (PP). The former pastes gas density and pressure on to haloes and requires only a halo catalogue, and the latter considers the contribution from field particles as well, i.e. particles that do not belong to any haloes and thus utilize the full particle information. Therefore, the PP algorithm incorporates secondary effects beyond the HP algorithm: asphericity of haloes and contribution from diffuse gas. In particular, such a diffuse component is the dominant source of the kinetic SZ effect. As validation of our methods, we have produced 108 all-sky maps with HP and 108 flat-sky maps, which cover $5 \times 5 \, \mathrm{deg}^2$ with both HP and PP, and measured power spectra of the maps. Our method can produce a mock map within a few hours, even for all-sky coverage with a parallel computational environment. The power spectra of HP maps are consistent with the halo model prediction of the thermal SZ effect. On the other hand, the power spectra of PP maps are suppressed due to the halo asphericity but can reproduce better the theoretical prediction for the kinetic SZ effect. We discuss the utility of baryon-pasted mock SZ maps for estimating the covariance matrix of SZ statistics and modelling the selection and projection effects for cluster cosmology.

SyncSignature

This paper introduces SyncSignature, the first fully parallelizable algorithmic framework for tree similarity joins under edit distance. SyncSignature makes use of implicit-synchronized signature generation schemes, which allow for an efficient and parallelizable candidate-generation procedure via hash join. Our experiments on large real-world datasets show that the proposed algorithms under the SyncSignature framework significantly outperform the state-of-the-art algorithm in the parallel computation environment. For datasets with big trees, they also exceed the state-of-the-art algorithms by a notable margin in the centralized/single-thread computation environment. To complement and guide the experimental study, we also provide a thorough theoretical analysis for all proposed signature generation schemes.

Output error behavior for discretizations of ergodic, chaotic systems of ordinary differential equations

The use of numerical simulation for prediction of characteristics of chaotic dynamical systems inherently involves unpredictable processes. In this work, we develop a model for the expected error in the simulation of ergodic, chaotic ordinary differential equation (ODE) systems, which allows for discretization and statistical effects due to unpredictability. Using this model, we then generate a framework for understanding the relationship between the sampling cost of a simulation and the expected error in the result and explore the implications of the various parameters of simulations. Finally, we generalize the framework to consider the total cost—including unsampled spin-up timesteps—of simulations and consider the implications of parallel computational environments to give a realistic model of the relationship between wall-clock time and the expected error in simulation of a chaotic ODE system.

Consensus clustering for Bayesian mixture models

BackgroundCluster analysis is an integral part of precision medicine and systems biology, used to define groups of patients or biomolecules. Consensus clustering is an ensemble approach that is widely used in these areas, which combines the output from multiple runs of a non-deterministic clustering algorithm. Here we consider the application of consensus clustering to a broad class of heuristic clustering algorithms that can be derived from Bayesian mixture models (and extensions thereof) by adopting an early stopping criterion when performing sampling-based inference for these models. While the resulting approach is non-Bayesian, it inherits the usual benefits of consensus clustering, particularly in terms of computational scalability and providing assessments of clustering stability/robustness.ResultsIn simulation studies, we show that our approach can successfully uncover the target clustering structure, while also exploring different plausible clusterings of the data. We show that, when a parallel computation environment is available, our approach offers significant reductions in runtime compared to performing sampling-based Bayesian inference for the underlying model, while retaining many of the practical benefits of the Bayesian approach, such as exploring different numbers of clusters. We propose a heuristic to decide upon ensemble size and the early stopping criterion, and then apply consensus clustering to a clustering algorithm derived from a Bayesian integrative clustering method. We use the resulting approach to perform an integrative analysis of three ’omics datasets for budding yeast and find clusters of co-expressed genes with shared regulatory proteins. We validate these clusters using data external to the analysis.ConclustionsOur approach can be used as a wrapper for essentially any existing sampling-based Bayesian clustering implementation, and enables meaningful clustering analyses to be performed using such implementations, even when computational Bayesian inference is not feasible, e.g. due to poor exploration of the target density (often as a result of increasing numbers of features) or a limited computational budget that does not along sufficient samples to drawn from a single chain. This enables researchers to straightforwardly extend the applicability of existing software to much larger datasets, including implementations of sophisticated models such as those that jointly model multiple datasets.

A multiobjective surrogate-assisted optimisation and exploration of low-boom supersonic transport planforms

Optimal non-planar wing planform geometries for a Mach 2.2 45-seat supersonic transport are generated using Kriging-based Bayesian optimisation techniques on a 22-variable, 3-objective problem for two wing-loading conditions. Sonic boom loudness, drag, and root bending moment are simultaneously minimised at a fixed lift. A Python-based framework is developed for rapid evaluation of designs in a parallel computation environment, automating conceptual geometry generation, optimised mesh griding, Euler CFD evaluation, and thermoviscous atmospheric boom propagation. An empirical turbulent skin friction coefficient is used to correct the Euler drag and an estimate of wave drag is performed. Self-organising maps and polynomial chaos expansion Shapley effects are used to explore the design variables and their associated sensitivities with respect to the design objectives. A geometric centroid-based proxy for trim is used to filter the results; feasible extreme designs and a compromise design are compared and discussed. The results indicate a trade-off between each of the three objectives due to competing design requirements, however an improvement in two of the objective functions is possible by sacrificing the remaining objective. One possible compromise solution reduces the root bending moment coefficient by 25% and the sonic boom loudness by 24% at the cost of a 5% increase in drag.

Numerical Modeling of Motion of Near-Earth Objects in a Parallel Computing Environment

Numerical Modeling of Motion of Near-Earth Objects in a Parallel Computing Environment

Knockout-Tournament Procedures for Large-Scale Ranking and Selection in Parallel Computing Environments

On one hand, large-scale ranking and selection (R&S) problems require a large amount of computation. On the other hand, parallel computing environments that provide a large capacity for computation are becoming prevalent today, and they are accessible by ordinary users. Therefore, solving large-scale R&S problems in parallel computing environments has emerged as an important research topic in recent years. However, directly implementing traditional stagewise procedures and fully sequential procedures in parallel computing environments may encounter problems because either the procedures require too many simulation observations or the procedures’ selection structures induce too many comparisons and too frequent communications among the processors. In this paper, inspired by the knockout-tournament arrangement of tennis Grand Slam tournaments, we develop new R&S procedures to solve large-scale problems in parallel computing environments. We show that no matter whether the variances of the alternatives are known or not, our procedures can theoretically achieve the lowest growth rate on the expected total sample size with respect to the number of alternatives and thus, are optimal in rate. Moreover, common random numbers can be easily adopted in our procedures to further reduce the total sample size. Meanwhile, the comparison time in our procedures is negligible compared with the simulation time, and our procedures barely request for communications among the processors.

Efficient stress‐constrained topology optimization using inexact design sensitivities

AbstractAn efficient computational approach to stress‐constrained topology optimization is presented. Using a multigrid‐preconditioned Krylov solver for the state and adjoint problems, the number of Krylov iterations is reduced significantly by enforcing early termination of the iterative solves. The criterion for early termination is based on the convergence of the design sensitivities with respect to Krylov iterations. Consequently, the progress of optimization is not affected by the inexact resolution of the state and adjoint problems. The proposed approach is demonstrated on several design problems that possess different characteristics of the maximum stress. Optimization results obtained with early termination show very good agreement with those obtained with an accurate direct solver—in terms of objective value, constrained maximum stress and overall number of design cycles. Savings of 80% in the number of Krylov iterations are achieved, compared to the number of iterations required to satisfy the force residual convergence criterion to a common tolerance. The approach is directly applicable to high‐resolution problems that are solved on parallel computational environments. A MATLAB code for reproducing the results is freely available at https://github.com/odedamir/topopt‐stress‐inexact‐sensitivities

COSIPY v1.3 – an open-source coupled snowpack and ice surface energy and mass balance model

Abstract. Glacier changes are a vivid example of how environmental systems react to a changing climate. Distributed surface mass balance models, which translate the meteorological conditions on glaciers into local melting rates, help to attribute and detect glacier mass and volume responses to changes in the climate drivers. A well-calibrated model is a suitable test bed for sensitivity, detection, and attribution analyses for many scientific applications and often serves as a tool for quantifying the inherent uncertainties. Here, we present the open-source COupled Snowpack and Ice surface energy and mass balance model in PYthon (COSIPY), which provides a flexible and user-friendly framework for modeling distributed snow and glacier mass changes. The model has a modular structure so that the exchange of routines or parameterizations of physical processes is possible with little effort for the user. The framework consists of a computational kernel, which forms the runtime environment and takes care of the initialization, the input–output routines, and the parallelization, as well as the grid and data structures. This structure offers maximum flexibility without having to worry about the internal numerical flow. The adaptive subsurface scheme allows an efficient and fast calculation of the otherwise computationally demanding fundamental equations. The surface energy balance scheme uses established standard parameterizations for radiation as well as for the energy exchange between atmosphere and surface. The schemes are coupled by solving both surface energy balance and subsurface fluxes iteratively such that consistent surface skin temperature is returned at the interface. COSIPY uses a one-dimensional approach limited to the vertical fluxes of energy and matter but neglects any lateral processes. Accordingly, the model can be easily set up in parallel computational environments for calculating both energy balance and climatic surface mass balance of glacier surfaces based on flexible horizontal grids and with varying temporal resolution. The model is made available on a freely accessible site and can be used for non-profit purposes. Scientists are encouraged to actively participate in the extension and improvement of the model code.

Application of the parallel diagonal dominant algorithm for the incompressible Navier-Stokes equations

The accuracy and the applicability of the parallel diagonal dominant (PDD) algorithm are explored for highly scalable computation of the incompressible Navier-Stokes equations which are integrated using a fully-implicit fractional-step method in parallel computational environments. The PDD algorithm is known to be applicable only for an evenly diagonal dominant matrix. In the present study, however, it is shown mathematically that the PDD algorithm is utilizable even for non-diagonal dominant matrices derived from discretization of incompressible momentum equations. The order of accuracy and the error characteristics are investigated in detail in terms of the Courant-Friedrichs-Lewy (CFL) number and the grid spacing by conducting simulations of decaying vortices in both two and three dimensions, flow in a lid-driven cavity, and flow over a circular cylinder. In order to reduce communication cost, which is one of bottlenecks in parallel computation, an aggregative data communication method is combined with the PDD algorithm. Parallel performance of the present PDD-based method is investigated by measuring the speedup, efficiency, overhead, and serial fraction.

A multimodal fusion based framework to reinforce IDS for securing Big Data environment using Spark

Securing Big Data has become one of the major issues of the exponentially pacing computing world, where data analysis plays an integral role, as it helps data analysts to figure out the interests and detailed information of organizational and industrial assets. Acts like cyber espionage and data theft lead to the inappropriate use of data. In order to detect the malicious content, we propose a model that ensures the security of heterogeneous data residing in commodity hardware. To verify the correctness of our model, (Knowledge Data Discovery) NSL KDD Cup 99 dataset is used that has been used by various researchers for working on (Intrusion Detection System) IDS. We incorporate decision-based majority voting multi-modal fusion that combines the results of different classifiers and facilitates better performance in terms of accuracy, detection rate and false alarm rate. Moreover, (Non-dominated Sorting Genetic Algorithm) NSGA-II plays its integral role for the selection of most promising features. Additionally, to reduce the computational complexity which is again a crucial aspect while processing Big Data, we incorporate the concepts of Hadoop MapReduce and Spark to ensure the fast processing of Big Data in a parallel computational environment. Our proposed model is able to achieve 92.03% accuracy, 99.38% detection rate and a testing time of 0.32 seconds. Additionally, we have achieved advantages in terms of accuracy and testing time of data over the existing techniques that use IDS as a security mechanism.

Modelling of Communication in Unified Parallel and Distributed Computing Environment

In modelling of sequential computers and algorithms, communication delay has often been ignored. Later, in modelling of parallel computers and parallel algorithms based on shared memory, it was supposed that the influence of communication is lower compared to computation complexity. It was supposed that, in better cases, communication delay is included in analysed computation complexity. However, a different situation can occur in high performance processors (CPU) or simplified CPUs called cores within the parallel computing node (multiprocessor/ multicore SMP). The similar case could occur in the multiple use of high performing computing nodes as the building computing nodes of modern parallel computers NOW (network of workstations) or Grid module (network of NOW modules). However, if complex problem are intended to be solved in a parallel way (parallel algorithm), they generate additional communication steps which correspond to inter process communication (IPC) among the decomposed parts (parallel processes) of the given solved problem. In general, it can be said that any parallel algorithm (PA) consists of created parallel processes having sequential character and IPC communication. Therefore, the increased role of IPC communication has been analysed in this paper through the analytical modelling of IPC complexity (number of communication steps) in a similar way as the one used to analyse computation complexity in sequential algorithms or parallel processes). Based on the defined communication parameters and the extensions of the complexity theory to PA, a unique modelling approach was developed to model the communication complexity of all existing parallel computers and PA. The secondary problem is to model critical parts of IPC influence for PA performance and that, at first, in minimisation of IPC steps of decomposed parallel processes and secondly at whole performance optimisation of the PA. The achieved results showed, from the user’s point of view, the important influence of modelling IPC complexity in PA on the illustrated real examples.

Multilevel approaches for FSAI preconditioning

SummaryFactorized sparse approximate inverse (FSAI) preconditioners are robust algorithms for symmetric positive matrices, which are particularly attractive in a parallel computational environment because of their inherent and almost perfect scalability. Their parallel degree is even redundant with respect to the actual capabilities of the current computational architectures. In this work, we present two new approaches for FSAI preconditioners with the aim of improving the algorithm effectiveness by adding some sequentiality to the native formulation. The first one, denoted as block tridiagonal FSAI, is based on a block tridiagonal factorization strategy, whereas the second one, domain decomposition FSAI, is built by reordering the matrix graph according to a multilevel k‐way partitioning method followed by a bandwidth minimization algorithm. We test these preconditioners by solving a set of symmetric positive definite problems arising from different engineering applications. The results are evaluated in terms of performance, scalability, and robustness, showing that both strategies lead to faster convergent schemes regarding the number of iterations and total computational time in comparison with the native FSAI with no significant loss in the algorithmic parallel degree.

Solving Traveling Salesman Problems with Ant Colony Optimization Algorithms in Sequential and Parallel Computing Environments: A Normalized Comparison

Solving Traveling Salesman Problems with Ant Colony Optimization Algorithms in Sequential and Parallel Computing Environments: A Normalized Comparison

A robust and cooperative parallel tabu search algorithm for the maximum vertex weight clique problem

Abstract The maximum vertex weight clique problem (MVWCP) is a challenging NP-Hard combinatorial optimization problem that searches for a clique with maximum total sum of vertices’ weights. In this study, we propose a robust and cooperative parallel tabu search algorithm (PTC) for the MVWCP. Our proposed algorithm uses a dedicated tabu search algorithm with a multistart strategy for the diversification of search space on a parallel computation environment. An effective seeding mechanism is developed with respect to the rank of the processors to choose diversified starting points for a better exploration of the search space. Classical add, swap and drop operators of tabu search are improved with parallel computation and a combined neighborhood approach. The PTC algorithm is evaluated on a set of 120 problem instances from DIMACS-W and BHOSLIB-W benchmarks. Computational results show that the PTC algorithm competes with state-of-the-art heuristic algorithms by reporting average best (optimal) result hit ratios up to 99.0%.

An efficient, robust and automatic overlapping grid assembly approach for partitioned multi-block structured grids

An efficient, robust and fully automatic grid assembly method on multi-block cell-centered structured grids for massively parallel computation is proposed in this paper. Compared with the traditional serial algorithm, the new approach eliminates the complex irregular boundaries created during the grid partition and avoids the large load imbalance caused by the large variation of grid-block overlapping. The main task of the overlapping grid assembly is to categorize all grid points into field points, fringe points and hole points. As to the main processes of the overlapping grid assembly, for hole cutting, an improved hole map method is applied to accurately identify the hole points located on the wall boundary with less memory cost. For donor search which is the most complex process on account of the irregular distribution of the partitioned multi-block structured grids in a parallel computation environment, the Alternating Digital Tree (ADT) is utilized to find out the potential donor cells quickly for query points. Besides, to achieve better overlapping quality, the wall distance criterion is implemented for overlapping optimization. In addition, two load balance algorithms are designed to solve the imbalance problem of overlapping grid assembly. Two test cases are applied to test the new overlapping grid assembly algorithm and the results show that the new overlapping grid assembly algorithm can deal with large-scale simulation of vehicles. The comparison of total time and speed-up among three algorithms manifests that the initial load balance algorithm using query point number as load criterion is not reliable while the improved load balance algorithm achieves good speed-up and least runtime. Meanwhile, the maximum proportion the improved load balance algorithm takes in one physical unsteady step in wing-pylon-store separation test case is less than 6.1%.