Network Of Workstations Research Articles

Migrating Engineering Windows HPC applications to Linux HTCondor and Slurm Clusters

The CERN IT department has been maintaining different High Performance Computing (HPC) services over the past five years. While the bulk of computing facilities at CERN are running under Linux, a Windows cluster was dedicated for engineering simulations and analysis related to accelerator technology development. The Windows cluster consisted of machines with powerful CPUs, big memory, and a low-latency interconnect. The Linux cluster resources are accessible through HTCondor, and are used for general purpose parallel but single-node type jobs, providing computing power to the CERN experiments and departments for tasks such as physics event reconstruction, data analysis, and simulation. For HPC workloads that require multi-node parallel environments for Message Passing Interface (MPI) based programs, there is another Linux-based HPC service that is comprised of several clusters running under the Slurm batch system, and consist of powerful hardware with low-latency interconnects.In 2018, it was decided to consolidate compute intensive jobs in Linux to make a better use of the existing resources. Moreover, this was also in line with CERN IT strategy to reduce its dependencies on Microsoft products. This paper focuses on the migration of Ansys [1], COMSOL [2] and CST [3] users from Windows HPC to Linux clusters. Ansys, COMSOL and CST are three engineering applications used at CERN for different domains, like multiphysics simulations and electromagnetic field problems. Users of these applications are in different departments, with different needs and levels of expertise. In most cases, the users have no prior knowledge of Linux. The paper will present the technical strategy to allow the engineering users to submit their simulations to the appropriate Linux cluster, depending on their simulation requirements. We also describe the technical solution to integrate their Windows workstations in order from them to be able to submit to Linux clusters. Finally, we discuss the challenges and lessons learnt during the migration.

A New Approach to Parallel Processing

The Application Space Architecture (ASA) defines parallel processor design as an adjunct to the Instruction Set Architectures (ISA) defined by Von Neumann for single processors. Before addressing hardware, one must understand applications requiring parallel processors, and software approaches required to meet speed constraints. Parallel processor applications use a single task to run N (number-of-processors) times faster than could be done on a single processor, where Processor Utilization Efficiency (PUE) is the ratio of single processor time to parallel processor time divided by N. Such applications require processing to be split into independent elements running in parallel on a large number of processors. This requires that processors do not sit idle (current PUEs are typically 2-10%.) This sets the requirement for “independent” elements to communicate with each other without stopping. In special cases, clusters of PCs are used to perform parametric analyses running many copies of a simulation on separate computers, typically using different random numbers. Except for initial and terminal processes, these separate simulations hardly communicate while running. Thus they are easily divided into separate tasks that typically exchange information between computers passing data or messages through a top level management task. Such applications (e.g., LINPACK) are labeled “embarrassingly parallel”.

Performance of Parallel Distributed Bat Algorithm using MPI on a PC Cluster

Optimization algorithms are often used to obtain optimal solutions to complex nonlinear problems and appear in many areas such as control, communication, computation, and others. Bat algorithm is a heuristic optimization algorithm and efficient in obtaining approximate best solutions to non-linear problems. In many situations complex problems involve large amount of computations that may require simulations to run for days or weeks or even years for an algorithm to converge to a solution. In this research, a Parallel Distributed Bat Algorithm (PDBA) is formulated using Message Passing Interface (MPI) in C language code for a PC Cluster. The time complexity of PDBA is determined and presented. The performance in terms of speed-up, efficiency, elapsed time, and number of times fitness function is executed is also presented.

Message Passing Interface - An Overview

The Message Passing Interface (MPI) is a communication library specification for parallel computers as well as for the workstation network. This paper signifies the role of MPI in parallel Computing. To give the general understanding of MPI Specification, this paper also presents the detailed view of the primary subject including “Evolution of MPI, Modes of MPI, MPI features, Communicator” etc. This paper ideal for the learner, new to parallel programming with MPI.

Priority Based Classical Data Encapsulated Scheduling For New Era in Computing Environment

In computational environment the data processing and data transactions are major issue, to overcome this problem of processing and reduce the band width in distributed models. The abstract mentioned here clearly state the novelty of the work regarding to data processing and data utilization. This paper shows that the data provided between nodes that are involved in clusters and that is useful to the data utilization schematic in all kinds of clusters. The proposed approach arranges data into data tables and distributes them transversely nodes in a cluster. It recommends jobs to be processed by the cluster then transfers packaged and encapsulated data into nodes to process the data in parallel. This technique employs an incremental data scheduled computation way that keep away from the costly enumeration of data pattern matching necessary by cluster methods. It improves the data locality by forwarding data to the job supporting cluster. This abstraction is enthused by the data processing and data reduce primitives presented in the various job processing environment and many other functional applications used in data grid, data intensive approaches. The Architectural interface is provided in between clusters and is used to achieve high performance on large clusters of commodity PCs. This approach reduces the data request and its processing when it is signed into data clusters. The proposed approach arranges data into data tables and distributes them across nodes in a cluster. It recommend jobs to be processed by the cluster then transfers packaged and encapsulated data into nodes to progression the data in parallel. It improves the data locality by forwarding data to the job supporting cluster. The proposed approach in this paper is helpful where number of nodes involve in cluster is always increasing due to the high end computations that are involved in distributed environment.

Parallel solvers for virtual element discretizations of elliptic equations in mixed form

The aim of this paper is twofold. On the one hand, we numerically test the performance of mixed virtual elements in three dimensions to solve the mixed formulation of three-dimensional elliptic equations on polyhedral meshes. On the other hand, we focus on the parallel solution of the linear system arising from such discretization, considering both direct and iterative parallel solvers. In the latter case, we develop two block preconditioners, one based on the approximate Schur complement and one on a regularization technique. Both these topics are numerically validated by several parallel tests performed on a Linux cluster. More specifically, we show that the proposed virtual element discretization recovers the expected theoretical convergence rates and we analyze the performance of the direct and iterative parallel solvers taken into account.

AtmoSwing: Analog Technique Model for Statistical Weather forecastING and downscalING (v2.1.0)

Abstract. Analog methods (AMs) use synoptic-scale predictors to search in the past for similar days to a target day in order to infer the predictand of interest, such as daily precipitation. They can rely on outputs of numerical weather prediction (NWP) models in the context of operational forecasting or outputs of climate models in the context of climate impact studies. AMs require low computing capacity and have demonstrated useful potential for application in several contexts. AtmoSwing is open-source software written in C++ that implements AMs in a flexible way so that different variants can be handled dynamically. It comprises four tools: a Forecaster for use in operational forecasting, a Viewer to display the results, a Downscaler for climate studies, and an Optimizer to establish the relationship between predictands and predictors. The Forecaster handles every required processing internally, such as NWP output downloading (when possible) and reading as well as grid interpolation, without external scripts or file conversion. The processing of a forecast requires low computing efforts and can even run on a Raspberry Pi computer. It provides valuable results, as revealed by a 3-year-long operational forecast in the Swiss Alps. The Viewer displays the forecasts in an interactive GIS environment with several levels of synthesis and detail. This allows for the provision of a quick overview of the potential critical situations in the upcoming days, as well as the possibility for the user to delve into the details of the forecasted predictand and criteria distributions. The Downscaler allows for the use of AMs in a climatic context, either for climate reconstruction or for climate change impact studies. When used for future climate studies, it is necessary to pay close attention to the selected predictors so that they contain the climate change signal. The Optimizer implements different optimization techniques, such as a semiautomatic sequential approach, Monte Carlo simulations, and a global optimization technique, using genetic algorithms. Establishing a statistical relationship between predictors and predictands is computationally intensive because it requires numerous assessments over decades. To this end, the code was highly optimized for computing efficiency, is parallelized (using multiple threads), and scales well on a Linux cluster. This procedure is only required to establish the statistical relationship, which can then be used for forecasting or downscaling at a low computing cost.

Wave prediction in a port using a fully nonlinear Boussinesq wave model

A wave forecasting system using FUNWAVE-TVD which is based on the fully nonlinear Boussinesq equations by Chen (2006) was developed to provide an accurate wave prediction in the Port of Busan, South Korea. This system is linked to the Korea Operational Oceanographic System (KOOS) developed by Park et al. (2015). The computational domain covers a region of 9.6 km×7.0 km with a grid size of 2 m in both directions, which is sufficient to resolve short waves and dominant sea states. The total number of grid points exceeds 16 millions, making the model computational expensive. To provide real-time forecasting, an interpolation method, which is based on pre-calculated results of FUNWAVE-TVD and SWAN forecasting results at the FUNWAVE-TVD offshore boundary, was used. A total of 45 cases were pre-calculated, which took 71 days on 924 computational cores of a Linux cluster system. Wind wave generation and propagation from the deep water were computed using the SWAN in KOOS. SWAN results provided a boundary condition for the FUNWAVE-TVD forecasting system. To verify the model, wave observations were conducted at three locations inside the port in a time period of more than 7 months. A model/model comparison between FUNWAVE-TVD and SWAN was also carried out. It is found that, FUNWAVE-TVD improves the forecasting results significantly compared to SWAN which underestimates wave heights in sheltered areas due to incorrect physical mechanism of wave diffraction, as well as large wave heights caused by wave reflections inside the port.

DelPhi Suite: New Developments and Review of Functionalities.

Electrostatic potential, energies, and forces affect virtually any process in molecular biology, however, computing these quantities is a difficult task due to irregularly shaped macromolecules and the presence of water. Here, we report a new edition of the popular software package DelPhi along with describing its functionalities. The new DelPhi is a C++ object‐oriented package supporting various levels of multiprocessing and memory distribution. It is demonstrated that multiprocessing results in significant improvement of computational time. Furthermore, for computations requiring large grid size (large macromolecular assemblages), the approach of memory distribution is shown to reduce the requirement of RAM and thus permitting large‐scale modeling to be done on Linux clusters with moderate architecture. The new release comes with new features, whose functionalities and applications are described as well. © 2019 The Authors. Journal of Computational Chemistry published by Wiley Periodicals, Inc.

Systematic computational identification of promiscuity cliff pathways formed by inhibitors of the human kinome.

The ability of a small molecule to interact with multiple target proteins provides the molecular basis of polypharmacology. So-defined compound promiscuity is intensely investigated in drug discovery. For example, for kinase inhibitors, the interplay between target selectivity and promiscuity plays a decisive role for different therapeutic applications. The "promiscuity cliff" (PC) concept was introduced previously to aid in promiscuity analysis. A PC is defined as a pair of structurally similar compounds with a large difference in promiscuity. Accordingly, PCs can reveal small structural modifications that might be responsible for selectivity or multi-target activity. In network representations, PCs form clusters of varying size and complexity that are difficult to analyze interactively. Herein, we introduce a computational method to systematically identify PC pathways, which are particularly rich in structure-promiscuity information, and extract them from PC clusters. PC pathways provide informative templates for experimental design. In a proof-of-concept investigation, we have applied the new computational approach to systematically identify pathways in more than 600 PC clusters formed by inhibitors of the human kinome, demonstrating the utility of the method and revealing many interesting promiscuity patterns.

Decentralized iterative approaches for community clustering in the networks

In this era of big data, as the data size is scaling up, the need for computing power is exponentially increasing. However, most of the community detection algorithms in the literature are classified as global algorithms, which require access to the entire information of the network. These algorithms designed to work on a single machine cannot be directly parallelized. Hence, it is impossible for such algorithms working in stand-alone machines to find communities in large-scale networks and also the required processing power far exceeds the processing capabilities of single machines. In this paper, a set of novel Decentralized Iterative Community Clustering Approaches to extract an efficient community structure for large networks are proposed and devalued using the LFR benchmark model. The approaches have the ability to identify the community clusters from the entire network without global knowledge of the network topology and will work with a range of computer architecture platforms (e.g., cluster of PCs, multi-core distributed memory servers, GPUs). Detecting and characterizing such community structures is one of the fundamental topics in network systems’ analysis, and it has many important applications in different branches of science including computer science, physics, mathematics and biology ranging from visualization, exploratory and data mining to building prediction models.

Status of Marine Seismic Exploration Technology

In this study, we examined the current status and future trends of the Korean and overseas marine seismic exploration technology. The marine seismic exploration technology is particularly useful to understand subsurface structures. It has been widely developed to find oil and gas resources in subsurface layers. The seismic exploration technology has been developed from a two-dimensional line survey to three- and four-dimensional surveys of the same area, several times, to identify changes over time. That approach enables an understanding not only of two-dimensional structures along the survey line but also of three-dimensional structures and deformation. In addition to that, attribute analysis and multicomponent seismic exploration technologies have been developed to identify the physical properties of pore fluids, such as oil and gas, and the broadband survey to obtain high-resolution seismic images to prevent specific frequency loss and the various azimuth acquisitions which do not depend on a particular shooting direction. Moreover, multiple attenuation, depth migration, and full waveform inversion have been performed using a supercomputer composed of PC clusters or GPUs for data processing. Various attribute analysis technologies that use computers were considerably developed for data interpretation. Finally, a high-resolution shallow exploration technology, which uses an oil exploration technology, has been developed for precise engineering exploration and geological shallow water survey.

Parallel Processing of k-Means Clustering Algorithm for Unsupervised Classification of Large Satellite Images: A Hybrid Method Using Multicores and a PC-Cluster

Parallel Processing of k-Means Clustering Algorithm for Unsupervised Classification of Large Satellite Images: A Hybrid Method Using Multicores and a PC-Cluster

A Step-by-Step Guide to Using BioNetFit.

BioNetFit is a software tool designed for solving parameter identification problems that arise in the development of rule-based models. It solves these problems through curve fitting (i.e., nonlinear regression). BioNetFit is compatible with deterministic and stochastic simulators that accept BioNetGen language (BNGL)-formatted files as inputs, such as those available within the BioNetGen framework. BioNetFit can be used on a laptop or stand-alone multicore workstation as well as on many Linux clusters, such as those that use the Slurm Workload Manager to schedule jobs. BioNetFit implements a metaheuristic population-based global optimization procedure, an evolutionary algorithm (EA), to minimize a user-defined objective function, such as a residual sum of squares (RSS) function. BioNetFit also implements a bootstrapping procedure for determining confidence intervals for parameter estimates. Here, we provide step-by-step instructions for using BioNetFit to estimate the values of parameters of a BNGL-encoded model and to define bootstrap confidence intervals. The process entails the use of several plain-text files, which are processed by BioNetFit and BioNetGen. In general, these files include (1) one or more EXP files, which each contains (experimental) data to be used in parameter identification/bootstrapping; (2) a BNGL file containing a model section, which defines a (rule-based) model, and an actions section, which defines simulation protocols that generate GDAT and/or SCAN files with model predictions corresponding to the data in the EXP file(s); and (3) a CONF file that configures the fitting/bootstrapping job and that defines algorithmic parameter settings.

Integration of Remote Sensing Evapotranspiration into Multi-Objective Calibration of Distributed Hydrology–Soil–Vegetation Model (DHSVM) in a Humid Region of China

This study presents an approach that integrates remote sensing evapotranspiration into multi-objective calibration (i.e., runoff and evapotranspiration) of a fully distributed hydrological model, namely a distributed hydrology–soil–vegetation model (DHSVM). Because of the lack of a calibration module in the DHSVM, a multi-objective calibration module using ε-dominance non-dominated sorted genetic algorithm II (ε-NSGAII) and based on parallel computing of a Linux cluster for the DHSVM (εP-DHSVM) is developed. The module with DHSVM is applied to a humid river basin located in the mid-west of Zhejiang Province, east China. The results show that runoff is simulated well in single objective calibration, whereas evapotranspiration is not. By considering more variables in multi-objective calibration, DHSVM provides more reasonable simulation for both runoff (NS: 0.74% and PBIAS: 10.5%) and evapotranspiration (NS: 0.76% and PBIAS: 8.6%) and great reduction of equifinality, which illustrates the effect of remote sensing evapotranspiration integration in the calibration of hydrological models.

A novel parallel and distributed magnetotelluric inversion algorithm on multi-threads workloads cluster

Different domains of research are moving to cloud computing whether to carry out compute intensive experiments or to store large datasets. Large-scale computation in geophysical exploration is often inefficient, especially in the Just-in-time (JIT) environment. To alleviate this, we devised a new parallel magnetotelluric inversion method on high performance computing (HPC) multi-threads workloads cluster. This parallel algorithm adapted to single CPU or PC clusters with multi-threads workloads allocates different waves to each thread in a coarse-gained mode. In all multi-threads, the master thread deals with all parallel tasks, and other slave threads compute the electromagnetic field values of each wave in a parallel fork-join model. Experiments show that the proposed parallel algorithm not only achieves effective data accuracy, but is more efficient than the serial version.

Applications of A Hyper–Graph Grammar System in Adaptive Finite–Element Computations

Abstract This paper describes application of a hyper-graph grammar system for modeling a three-dimensional adaptive finite element method. The hyper-graph grammar approach allows obtaining a linear computational cost of adaptive mesh transformations and computations performed over refined meshes. The computations are done by a hyper-graph grammar driven algorithm applicable to three-dimensional problems. For the case of typical refinements performed towards a point or an edge, the algorithm yields linear computational cost with respect to the mesh nodes for its sequential execution and logarithmic cost for its parallel execution. Such hyper-graph grammar productions are the mathematical formalism used to describe the computational algorithm implementing the finite element method. Each production indicates the smallest atomic task that can be executed concurrently. The mesh transformations and computations by using the hyper-graph grammar-based approach have been tested in the GALOIS environment. We conclude the paper with some numerical results performed on a shared-memory Linux cluster node, for the case of three-dimensional computational meshes refined towards a point, an edge and a face.

Parallel fast isogeometric L2 projection solver with GALOIS system for 3D tumor growth simulations

We present a parallel solver for isogeometric L2 projections implemented in GALOIS system. We use the solver for isogeometric finite element method simulations of tumor growth. First, we describe the dependencies between the quantities describing the melanoma tumor growth in three-dimensions. Later, we derive the system of partial differential equations modeling the tumor dynamics, including tumor cell density, tumor angiogenic factor, flux, pressure, extracellular and degraded extracellular matrices. The continuous model is coupled every ten-time steps with graph grammar model describing the vasculature network. The resulting system of PDEs is solved with our parallel solver over the shared memory Linux cluster node. We show high computational efficiency and accuracy of our model and we demonstrate its perfect parallel scalability with the number of CPU cores. We discuss the application of our solver in predictive oncology.

Distribution of Serpens South protostars revealed with ALMA

Context. Clusters are common sites of star formation, whose members display varying degrees of mass segregation. The cause may be primordial or dynamical, or a combination both. If mass segregation were to be observed in a very young protostellar cluster, then the primordial case can be assumed more likely for that region. Aims. We investigated the masses and spatial distributions of pre-stellar and protostellar candidates in the young, low-mass star forming region Serpens South, where active star formation is known to occur along a predominant filamentary structure. Previous observations used to study these distributions have been limited by two important observational factors: (1) sensitivity limits that leave the lowest-mass sources undetected or (2) resolution limits that cannot distinguish binaries and/or cluster members in close proximity. Methods. Recent millimeter-wavelength interferometry observations can now uncover faint and/or compact sources in order to study a more complete population of protostars, especially in nearby (D < 500 pc) clusters. Here we present ALMA observations of 1 mm (Band 6) continuum in a 3 × 2 arcmin region at the center of Serpens South. Our angular resolution of ~1′′ is equivalent to ~400 au, corresponding to scales of envelopes and/or disks of protostellar sources. Results. We detect 52 sources with 1 mm continuum, and we measure masses of 0.002–0.9 solar masses corresponding to gas and dust in the disk and/or envelope of the protostellar system. For the deeply embedded (youngest) sources with no IR counterparts, we find evidence of mass segregation and clustering according to: the minimum spanning tree method, distribution of projected separations between unique sources, and concentration of higher-mass sources near to the dense gas at the cluster center. Conclusions. The mass segregation of the mm sources is likely primordial rather than dynamical given the young age of this cluster, compared with segregation time. This is the first case to show this for mm sources in a low-mass protostellar cluster environment.

Optimizing 10-Gigabit Ethernet for Networks of Workstations, Clusters, and Grids: A Case Study

Optimizing 10-Gigabit Ethernet for Networks of Workstations, Clusters, and Grids: A Case Study