High-performance Cluster System Research Articles

COMPUTATIONAL EXPERIMENTS ON SOLVING GRID ELLIPTIC EQUATIONS BY SOME PARALLEL ITERATIVE METHODS ON THE K60 CLUSTER

The paper deals with the problem of preserving natural water systems, as well as maintaining their integrity, not only through the enterprise of organizational, engineering and technical solutions, but also through the development of highly effective mathematical modeling techniques that make it possible quickly and efficiently, based on interconnected high-precision models of hydrophysics and hydrobiology, predict the processes of pollution spreading and the occurrence of hazardous phenomena in coastal systems. The article considers algorithms for solving grid equations developed for high-performance cluster systems. A model of parallel computations is proposed, which makes it possible, when choosing the appropriate method for solving the problem of aquatic ecology, to estimate the cost of calculations, which is defined as the product of the time of parallel solution of the problem and the number of processors used. An estimate of the optimal amount of information packet for exchange between processors is obtained. The adaptive modified alternating-triangular method of minimum corrections is described, the results of numerical experiments for the parallel variant of this one are presented.

Augmentation-Based Edge Differentially Private Path Publishing in Networks

Paths in a given network represent the occurrence sequences of nodes in many real world applications, such as disease transmission chains, object trajectories and data access sequences. In this paper, we address the problem of publishing edge-privacy preserved path information for a single path such that legitimate users with the full knowledge of the network can reconstruct the path with the published information, but not adversaries, even if they have the maximum background knowledge of all the vertices and all edges but one (on the path) of the network. Existing studies on edge privacy against inference attacks focus on publishing either differential privacy (DP) noise injected graph statistics or DP edge perturbed graph topology to achieve edge differential privacy preservation. However, none of them provides an assurance on both edge privacy and data utility. To effectively protect edge privacy and maintain data utility, we propose a novel scheme of DP augmentation instead of DP perturbation as did in existing work, that publishes a simple-topology graph containing an augmented path with fake edges and vertices applying differential privacy to protect the actual path, such that only the legitimate users are able to reconstruct the actual path with high probability. We theoretically analyse the performance of our algorithm in terms of output quality on differential privacy and utility, and execution efficiency. We also conduct extensive experimental evaluations on a high-performance cluster system to validate our analytical results.

Using a Workflow Management Platform in Textual Data Management

Abstract The paper gives a brief introduction about the workflow management platform, Flowable, and how it is used for textual-data management. It is relatively new with its first release on 13 October, 2016. Despite the short time on the market, it seems to be quickly well-noticed with 4.6 thousand stars on GitHub at the moment. The focus of our project is to build a platform for text analysis on a large scale by including many different text resources. Currently, we have successfully connected to four different text resources and obtained more than one million works. Some resources are dynamic, which means that they might add more data or modify their current data. Therefore, it is necessary to keep data, both the metadata and the raw data, from our side up to date with the resources. In addition, to comply with FAIR principles, each work is assigned a persistent identifier (PID) and indexed for searching purposes. In the last step, we perform some standard analyses on the data to enhance our search engine and to generate a knowledge graph. End-users can utilize our platform to search on our data or get access to the knowledge graph. Furthermore, they can submit their code for their analyses to the system. The code will be executed on a High-Performance Cluster (HPC) and users can receive the results later on. In this case, Flowable can take advantage of PIDs for digital objects identification and management to facilitate the communication with the HPC system. As one may already notice, the whole process can be expressed as a workflow. A workflow, including error handling and notification, has been created and deployed. Workflow execution can be triggered manually or after predefined time intervals. According to our evaluation, the Flowable platform proves to be powerful and flexible. Further usage of the platform is already planned or implemented for many of our projects.

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.

Modeling of program-target organization and management of competitiveness of territorial reproduction systems in construction

.This study examines modern theoretical and methodological approaches to solving the problem of program-target organization and management of meso-competitiveness in construction as a prerequisite for effective practical implementation of transformational development according to key monitoring indicators such as productivity, organizational and technical reliability. Taking into account the inevitability of the involvement of territorial construction complexes in the reproduction processes as a meso-economic competitive territorial-reproduction construction system, the article substantiates the need to develop special program-target methods for regulating these processes. It shows the expediency of combining the competitiveness, productivity of territorial construction complexes and reproduction processes in construction by the program-target method. The adopted programs should essentially become mortgage and investment programs for long-term management of the reproduction of territorial-portfolio land and property complexes in various industries and infrastructures. The study presents the efficiency of using consortium-type structures to implement the idea of forming high-performance cluster systems in construction at regional and inter-regional levels. Mortgage and investment programs for sustainable development of residential construction should become a tool for managing the process of transformational regulation of competitiveness from its local level to the interregional and international levels with a clear system organization.

Routing and Fault Tolerance in Z-Fat Tree

Fat tree topologies have been extensively used as interconnection networks for high performance computing, cluster and data center systems, with their most recent variants able to fairly extend and scale to accommodate higher processing power. While each progressive and evolved fat-tree topology includes some extra advancements, these networks do not fully address all the issues of large scale HPC. We propose a topology called Zoned-Fat tree ( Z -Fat tree,) which is a further extension to the fat trees. The extension relates to the provision of extra degree of connectivity to utilize the extra ports per switches (routing nodes), that are, in some cases, not utilized by the architectural constraints of other variants of fat trees, and hence increases the bisection bandwidth, reduces the latency and supplies additional paths for fault tolerance. To support and profit from the extra links, we propose an adaptive low latency routing for up traffic which is based on a series of leading direction bits predefined at the source; furthermore we suggest a deterministic routing by implementing a dynamic round robin algorithm that overtakes D-mod-K in same cases and guarantees the utilization of all the extra links. We also propose a fault tolerance algorithm, named recoil-and-reroute which makes use of the extra links to ensure higher message delivery even in the presence of faulty links and switches.

Enabling Parallel Simulation of Large-Scale HPC Network Systems

With the increasing complexity of today's high-performance computing (HPC) architectures, simulation has become an indispensable tool for exploring the design space of HPC systems-in particular, networks. In order to make effective design decisions, simulations of these systems must possess the following properties: (1) have high accuracy and fidelity, (2) produce results in a timely manner, and (3) be able to analyze a broad range of network workloads. Most state-of-the-art HPC network simulation frameworks, however, are constrained in one or more of these areas. In this work, we present a simulation framework for modeling two important classes of networks used in today's IBM and Cray supercomputers: torus and dragonfly networks. We use the Co-Design of Multi-layer Exascale Storage Architecture (CODES) simulation framework to simulate these network topologies at a flit-level detail using the Rensselaer Optimistic Simulation System (ROSS) for parallel discrete-event simulation. Our simulation framework meets all the requirements of a practical network simulation and can assist network designers in design space exploration. First, it uses validated and detailed flit-level network models to provide an accurate and high-fidelity network simulation. Second, instead of relying on serial time-stepped or traditional conservative discrete-event simulations that limit simulation scalability and efficiency, we use the optimistic event-scheduling capability of ROSS to achieve efficient and scalable HPC network simulations on today's high-performance cluster systems. Third, our models give network designers a choice in simulating a broad range of network workloads, including HPC application workloads using detailed network traces, an ability that is rarely offered in parallel with high-fidelity network simulations.

Secure and efficient high-performance PROOF-based cluster system for high-energy physics

The particle detectors at the LHC produced about 25 PB data in a year, and the total size of data recorded on tape media at the CERN data center reached almost 100 PB in 2013. In order to preserve such a large-scale data safely and process them fast and efficiently, the collaboration of the data centers around the world (WLCG) was founded. The overall performance of the WLCG has strong site-dependency, and the time scale for the result is unpredictable. In this paper, we introduce PROOF developed for providing high performance and reliable environment for the user analysis task to complement the grid computing, and we provide a new Linux cluster (KIAF), by deploying PROOF. The KIAF cluster provides data access in low latency and secure connection with the clients. In order to find the correlation between the processing performance and the scalability of KIAF, we perform a simple analysis task counting the number of particles in the data on KIAF by varying the number of worker nodes from 1 to 96. We show that the linear scalability of performance of KIAF and its limitation as a function of the number of workers participating in processing.

Implementing an affordable high-performance computing for teaching-oriented computer science curriculum

With the advances in computing power, high-performance computing (HPC) platforms have had an impact on not only scientific research in advanced organizations but also computer science curriculum in the educational community. For example, multicore programming and parallel systems are highly desired courses in the computer science major. However, the high cost of HPC equipment and maintenance makes it hard to be adapted into a conventional computer science curriculum. Specifically, teaching-oriented institutions cannot afford an HPC system due to the high cost, lack of experience, and smaller research infrastructure. The main objective of this article is to present an affordable and easy-to-use high-performance cluster system for teaching-oriented computer science curriculums. In order to address this, we have designed and implemented an affordable high-performance cluster system based on the PlayStation 3 (PS3). For the performance evaluation of the PS3 cluster, we conducted a benchmarking test, that is, matrix multiplication, with different numbers of synergistic processing elements (SPEs) and nodes. As a result, it was concluded that the PS3Cluster provides enough computing power as an HPC for computer science courses, while the total cost is less than 10% of an existing cluster system on the market that has similar performance. In addition, the implemented PS3Cluster system has been used for computer science courses, such as Parallel and Distributed Databases and Parallel Programming.

Probability of receptor lethality in blowout of sour gas wells

Abstract By the Fluent software on the high-performance cluster system of parallel computing, the computational fluid dynamics method was applied to simulate the temporal and spatial evolvement of hydrogen sulfide's concentration during the blowout of a typical natural gas well with hydrogen sulfide in Sichuan and Chongqing. The difference and integral calculations were conducted to obtain the distribution of toxicity load for hydrogen sulfide. Then, the probability of lethality for receptors around the well was analyzed. The topographic condition of the gas well had a great impact on the calculation results. The topographic condition of the selected gas well was flat on the east, west, and north sides, dominantly low mounds, and there was a mountain of 1230 m altitude on the south side. Because of the influence of the mountain, the position of second peak value for toxicity load and lethality probability can reach 1000 m from the well under northerly wind. So it must be fully considered when performing the consequence simulation and quantitative risk assessment of gas wells. This method is preferable in the quantitative risk assessment and safety planning of sour gas wells.

Service-Oriented Atmospheric Radiances (SOAR): Gridding and Analysis Services for Multisensor Aqua IR Radiance Data for Climate Studies

The Aqua spacecraft, launched on May 4, 2002, carries two well-calibrated independent infrared (IR) grating spectrometers Atmospheric Infrared Sounder (AIRS) and Moderate Resolution Imaging Spectrometer (MODIS), which have been continuously returning upwelling IR spectral radiance measurements for over five years. Based on an Aqua Sr. Project Review, estimates of available flight fuel, power, and orbital projections assess the life span of the Aqua satellite, and these two instruments, to be reliable to 2013. Since launch, these instruments have generated petabytes of data, which are managed and made available by the Goddard Space Flight Center (GSFC) Earth Science Data and Information Services Center and GSFC MODAPS. Agencies such as NOAA, DOD, EPA, and USGS use the AIRS data mostly for weather-related applications, whereas MODIS data are used, in addition to some climate-related studies, for studies of weather, oceans, and land processes, aerosols, natural and man-made disasters, and earth ecology. The Science Investigator-led Processing Systems (SIPS) teams have made many of the desired products derived from these data sets available either as level 2 products and/or level 3 gridded product fields. However, no gridded level 3 data products of radiances, either averaged for a grid element, max, min, or as brightness temperatures (BTs), are provided directly by the SIPS. Thus, one impediment that the general community faces in accessing these MODIS produced petabytes of data is storing such large data sets, interpreting the multiformatted data, and transforming it into helpful data sets for climate-research needs. The Service-Oriented Atmospheric Radiance (SOAR) system has been designed to bridge these gaps and overcome the challenges of bringing this rich data source to the science community, by delivering applications that process these valuable radiance data into standard spatial-temporal grids as well as user-defined criteria on demand. SOAR can serve this community with aggregated, enriched, and thinned gridded data sets provided with access to the data on demand, with query and subsetting capabilities across many dimensions. In addition, SOAR provides online user-specified visualization and analysis requests, all accessible via a Web browser. The utility of SOAR is exposed via Web-service routines, using the Simple Object Access Protocol. The Web-service library and supporting technologies (Axis, PostgreSQL, and Tomcat) reside on a University of Maryland Baltimore Campus client server, which interfaces to and invokes algorithms on the process server, a high-performance computer cluster and storage system. These servers are connected to the sensor data stores at the GSFC via a high-speed fiber-optic network connection [10 Gb/s], providing reliable and fast on-demand access to a vast online library of AIRS and current monthly MODIS source data.

이중 연결 구조 CC-NUMA 시스템의 효율적인 상호 연결망 구성 기법

반도체 미세 공정의 개발과 더불어, 높아진 집적도 및 동작 클럭의 고속화로 단일 프로세서 시스템 성능은 지속적으로 개선되고 있다. 이 결과 기가헬즈 이상의 클럭 속도를 가지는 개인용 컴퓨터가 보편적인 데스크 탑 시스템으로 자리잡게 되었으며, 불과 수년 전의 고가 대형 시스템은 점차 이러한 작은 시스템들을 상호 연결망으로 연결한 형태로 급속히 대체되어가고 있다. 이러한 구조의 클러스터 컴퓨터는 높은 확장성과 고성능을 얻을 수 있으므로, 점차 그 영역을 확대해나가고 있으나, 상호 연결망의 대역폭 및 지연에 따라 성능 제한 요소는 여전히 존재하고 있으며, 이러한 이유로 SCI, Myrinet, Gigabit Ethernet 등 고속의 상호 연결망이 클러스터 시스템의 연결 구조로 사용되고 있다. 프로세서 속도의 개발과 더불어 상호 연결망의 속도 또한 개선되어 왔는데, 상호 연결망은 그 대역폭을 늘리는 것과, 상호 연결망을 이용한 경우의 통신 시간지연의 축소로 볼 수 있다. 대역폭의 확장 및 지연시간의 단축은 상호 연결망의 고속화를 통하여 이루어질 수 있으나, 작은 면적에 집적되어 있는 프로세서와는 달리, 보다 넓은 면적에 펼쳐져 있는 상호 연결망의 동작 속도는, 물리적 거리에 의한 지연으로 인하여 개선의 난이도가 높으며, 따라서 클러스터 시스템의 확장 규모는 상호 연결망의 병목 현상에 의하여 제한된다고 할 수 있다. 이러한 이유로 보다 높은 대역폭의 상호 연결망을 구현하려는 노력은 복수개의 연결 구조를 이용한 형태로 개선되어 왔으며, 고속으로 동작하는 SCI 점 대 점 연결구조론 이용한 다중연결 형태의 시스템이 활발히 연구되어 왔다. 본 논문은 이러한 이중 점 대 점 연결 구조 시스템의 성능 제한 요소인 접근 시간 및 효율을 개선하기 위하여, 두개 중 하나의 점 대 점 연결을 링 형태로, 나머지 하나는 링을 몇 개의 노드의 묶음으로 분할하여 연결하는 구성을 제시하였으며, 방송 및 일 대 일 전송에 적합한, 간단하고 효율적인 경로 설정 방법과 적절한 묶음의 수를 제시하였다. 본 논문에 제시한 구조의 시스템의 성능 측정의 비교 대상으로, 최신 시스템에 채용되어 있는 반대방향 이중 링 구조를 비교 대상으로 하였으며, 반대방향 이중 연결 구조에 비하여 단 논문에 제시한 상호연결망 구성 및 트랜잭션 경로 설정 방법이 상대적으로 우수함을 시뮬레이션을 통하여 검증하였다. 실험 결과, 본 논문에서 제안한 상호연결망 구조 및 트랜잭션 경고 설정 방법을 이용한 경우, 반대방향 이중 링 구조의 시스템 구조에 비하여 단위 트랜잭션의 처리 시간이 1.05∼l.11배 향상되었으며, 시스템의 성능은 1.42∼2.1배 향상되었다. The performance of the multiprocessor systems is limited by the several factors. The system performance is affected by the processor speed, memory delay, and interconnection network bandwidth/latency. By the evolution of semiconductor technology, off the shelf microprocessor speed breaks beyond GHz, and the processors can be scalable up to multiprocessor system by connecting through the interconnection networks. In this situation, the system performances are bound by the latencies and the bandwidth of the interconnection networks. SCI, Myrinet, and Gigabit Ethernet are widely adopted as a high-speed interconnection network links for the high performance cluster systems. Performance improvement of the interconnection network can be achieved by the bandwidth extension and the latency minimization. Speed up of the operation clock speed is a simple way to accomplish the bandwidth and latency betterment, while its physical distance makes the difficulties to attain the high frequency clock. Hence the system performance and scalability suffered from the interconnection network limitation. Duplicating the link of the interconnection network is one of the solutions to resolve the bottleneck of the scalable systems. Dual-ring SCI link structure is an example of the interconnection network improvement. In this paper, I propose a network topology and a transaction path algorism, which optimize the latency and the efficiency under the duplicated links. By the simulation results, the proposed structure shows 1.05 to 1.11 times better latency, and exhibits 1.42 to 2.1 times faster execution compared to the dual ring systems.

고성능 클러스터 시스템을 위한 인피니밴드 시스템 연결망의 설계 및 구현

InfiniBand technology is being accepted as the future system interconnect to serve as the high-end enterprise fabric for cluster computing. This paper presents the design and implementation of the InfiniBand system interconnect, focusing on an InfiniBand host channel adapter (HCA) based on dual ARM9 processor cores The HCA is an SoC tailed KinCA which connects a host node onto the InfiniBand network both in hardware and in software. Since the ARM9 processor core does not provide necessary features for multiprocessor configuration, novel inter-processor communication and interrupt mechanisms between the two processors were designed and embedded within the KinCA chip. Kinch was fabricated as a 564-pin enhanced BGA (Bail Grid Array) device using 0.18 CMOS technology Mounted on host nodes, it provides 10 Gbps outbound and inbound channels for transmit and receive, respectively, resulting in a high-performance cluster system.

An SCI-Based PC Cluster Utilizing Coherent Network Cache

It is extremely important to minimize network access time in constructing a high-performance PC cluster system. For an SCI-based PC cluster, it is possible to reduce the network access time by maintaining network cache in each cluster node. This paper presents a Network-Cache-Coherent-NUMA (NCC-NUMA) card that utilizes network cache for SCI-based PC clustering. The NCC-NUMA card is directly plugged into the PCI slot of each node, and contains shared memory, network cache, and interconnection modules. The network cache is maintained for the shared memory on the PCI bus of cluster nodes. The coherency mechanism between the network cache and the shared memory is based on the IEEE SCI standard. Both a simulator and an NCC-NUMA prototype card are developed to evaluate the performance of the system. According to the experiments, the cluster system with the NCC-NUMA card showed considerable improvements compared with an SCI-based cluster without network cache.

Performance of Workstation Cluster with High-Speed Network for Parallel Finite Element Analysis.

High-speed networks such as the Fast Ethernet are currently available for workstation clusters. They have the potential for solving communication capacity problems for large-scale parallel finite element analysis on a workstation cluster. Parallel performance of the domain decomposition method, the domain partitioned conjugate gradient method and the parallel Gaussian elimination method is evaluated on a high-performance cluster system with Fast Ethernet and compared with different communication and computing performance. The optimization of parallel parameters for the Fast Ethernet environment is also performed based on the parallel performance evaluation. A large-scale problem with 350000 degrees of freedom has been solved in about one hour using ten workstations.