检索结果-内蒙古大学图书馆

Task allocation in organic computing systems: Networks with reconfigurable helper units

International Journal of Autonomous and Adaptive Communications systems 2015年第1期8卷 60-80页

作者： Merkle, Daniel Middendorf, Martin Scheidler, Alexander Department of Mathematics and Computer Science University of Southern Denmark Campusvej 55 Odense M5230 Denmark Parallel Computing and Complex Systems Group Department of Computer Science University of Leipzig Johannisgasse 26 Leipzig04103 Germany Université Libre de Bruxelles Av F. D. Roosevelt 50 CP 194/6 Bruxelles1050 Belgium

In this paper, computing systems that have no central control and consist of many (partially) autonomous components are studied. Two types of components are distinguished which are called workers and helpers respectively. The components are connected via a network and the helper components perform service tasks for the worker components. It is assumed that the helper components exhibit reconfigurable hardware to be able to execute different service tasks efficiently. The problem addressed in this paper is how to organise such a decentralised system in a way that requests of the workers are executed by suitable helpers and the total reconfiguration costs of the helpers are small. Several decentralised task allocation methods are proposed. One of them uses a combination of a fully decentralised dynamic clustering algorithm and a self-organised task allocation system. The clustering algorithm is used to classify the service requests that are sent through the network in order to give the helpers hints which requests are suitable to be executed by them. Simulations are done for static and dynamic scenarios to investigate the reconfiguration costs and the number of dropped requests, i.e., requests that could not be satisfied. The results show that the clustering-based system has a strong adaptive behaviour and that the decentralised clustering is able to reduce the reconfiguration costs significantly. Copyright © 2015 Inderscience Enterprises Ltd.

关键词： Reconfigurable hardware

来源：评论

学校读者我要写书评

暂无评论

Domain-specific optimisation for the high-level synthesis of CellML-based simulation accelerators

Domain-specific optimisation for the high-level synthesis of...

引用

International Conference on Field Programmable Logic and Applications

作者： Julian Oppermann Andreas Koch Ting Yu Oliver Sinnen Embedded Systems and Applications Group Technische Universität Darmstadt Germany Parallel and Reconfigurable Computing Group University of Auckland New Zealand

The simulation of biomedical models often requires the numerical integration of ordinary differential equation systems, a computationally intensive task that can be accelerated well by deeply-pipelined FPGA-based accelerators. Since the main design target is throughput, larger FPGA devices can easily be exploited by scaling-up the number of parallel datapath instances on a chip. To this end, reducing the area of each datapath becomes a key optimisation. High-level synthesis can be employed to generate custom simulation accelerators from standardised cell descriptions in CellML. In this work, we improve this process by inserting LLVM into the flow to pre-optimise the simulation models generated from CellML for hardware synthesis. This is achieved not only by the selective application of general-purpose optimisation passes, but also by adding new domain-specific optimisations, including unsafe floating-point transformations, to the optimisation flow. We investigate their effect on the quality-of-results and show that a novel strategy using our optimisations outperforms standard strategies, such as LLVM's -Oz (aggressive size reduction), when applied for hardware synthesis in 99 out of 146 example models. Our approach, which reduces area by up to 25%, leads to the smallest implementations for four models examined in detail, and allows a particularly complex cell model to fit on the target FPGA device for the first time.

关键词： Mathematical model Optimization Computational modeling Biological system modeling Numerical models Hardware Field programmable gate arrays

来源：评论

学校读者我要写书评

暂无评论

The influence of correlated objectives on different types of P-ACO algorithms 14

The influence of correlated objectives on different types of...

引用

14th European Conference on Evolutionary Computation in Combinatorial Optimization, EvoCOP 2014

作者： Moritz, Ruby L. V. Reich, Enrico Bernt, Matthias Middendorf, Martin Parallel Computing and Complex Systems Group Institute of Computer Science University of Leipzig Germany

ISBN: (纸本)9783662443194

The influence of correlated objectives on different types of P-ACO algorithms for solutions of multi objective optimization problems is investigated. Therefore, a simple method to create multi objective optimization problems with correlated objectives is proposed. Theoretical results show how certain correlations between the objectives can be obtained. The method is applied to the Traveling Salesperson problem. The influence of the correlation type and strength on the optimization behavior of different P-ACO algorithms is analyzed empirically. A particular focus is given on P-ACOs with ranking methods. © Springer-Verlag Berlin Heidelberg 2014.

关键词： Multiobjective optimization

来源：评论

学校读者我要写书评

暂无评论

Zero-copy, hierarchical gather is not possible with MPI datatypes and collectives 14

Zero-copy, hierarchical gather is not possible with MPI data...

引用

21st European MPI Users' group Meeting, EuroMPI/ASIA 2014

作者： Träff, Jesper Larsson Rougier, Antoine Faculty of Informatics Institute of Information Systems Research Group Parallel Computing Favoritenstrasse 16/184-5 Vienna1040 Austria

ISBN: (纸本)9781450328753

It is a desirable feature of MPI that application specific collective operations can be implemented efficiently in terms of other, more primitive operations of MPI. Recent MPI 3.0 functionality makes it possible to build portable libraries that are sensitive to and can exploit the hierarchical structure of, e.g., current shared-memory clusters, and thus in principle to implement efficient, hierarchy-aware collective operations effectively. In this paper, which complements investigations on all-to-all communication, we show that current MPI 3.0 functionality is insufficient to implement simple, collective gather operations as found in the MPI standard in a hierarchical manner in terms of simpler gather operations, if we pose the extra constraint that no explicit data reorganization apart from what can be expressed with MPI derived datatypes is allowed in any stage of the implementation. Thus, we establish that zero-copy, hierarchical implementations of MPI-Gather are not possible in current MPI. We discuss ways to alleviate the problems, and sketch a possible solution. © ACM 2014.

关键词： Message passing

来源：评论

学校读者我要写书评

暂无评论

MPI collectives and datatypes for hierarchical all-to-all communication 14

MPI collectives and datatypes for hierarchical all-to-all co...

引用

21st European MPI Users' group Meeting, EuroMPI/ASIA 2014

作者： Träff, Jesper Larsson Rougier, Antoine Faculty of Informatics Institute of Information Systems Research Group Parallel Computing Favoritenstrasse 16/184-5 Vienna1040 Austria

ISBN: (纸本)9781450328753

With recent MPI 3.0 functionality for creating communicators that partly reflects the hierarchy of standard clusters of shared-memory nodes, hierarchical, collective algorithms can more conveniently be implemented by combinations of other collective MPI operations. On systems that support MPI 3.0, with MPI-Alltoall as a concrete example, we show that performance on par with or better than native MPI implementations is possible, thus illustrating that the provided hierarchy awareness can be an effective and portable means for applications to implement their own, efficient (non-MPI) collective operations. Parts of the MPI-Alltoall implementation relies on MPI derived datatypes;however, the MPI collective interfaces lack expressivity to take full advantage of the performance benefits offered by the derived datatype mechanism. © ACM 2014.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Optimal MPI datatype normalization for vector and index-block types 14

Optimal MPI datatype normalization for vector and index-bloc...

引用

21st European MPI Users' group Meeting, EuroMPI/ASIA 2014

作者： Träff, Jesper Larsson Faculty of Informatics Institute of Information Systems Research Group Parallel Computing Favoritenstrasse 16/184-5 Vienna1040 Austria

ISBN: (纸本)9781450328753

The derived datatypes of MPI is an extremely powerful mechanism for specifying the layout of data in communication operations. It is desirable that MPI libraries internally simplify complex datatype descriptions into representations that are efficient for the communication operations in which they are used. This process is called datatype normalization, and MPI libraries typically employ simple heuristics for this task. In this paper we embark on a study of the inherent complexity of datatype normalization, and show that the problem with a specific, but flexible cost model can be solved cost-optimally for MPI vector and index-block types in polynomial time. For a type map consisting of n displacement-basetype pairs, we first give a type reconstruction algorithm running in O(n√n) time steps. We then use this algorithm for datatype normalization of given derived datatypes. © ACM 2014.

关键词： Libraries

来源：评论

学校读者我要写书评

暂无评论

Stepping stones to reproducible research: A study of current practices in parallel computing

Stepping stones to reproducible research: A study of current...

引用

Lecture Notes in Computer Science

作者： Carpen-Amarie, Alexandra Rougier, Antoine Lübbe, Felix D. Vienna University of Technology Institute of Information Systems Research Group Parallel Computing Favoritenstrasse 16/184-5 Vienna1040 Austria

ISBN: (纸本)9783319143248

Experimental research plays an important role in parallel computing, as in this field scientific discovery often relies on experimental findings, which complement and validate theoretical models. However, parallel hardware and applications have become extremely complex to study, due to their diversity and rapid evolution. Furthermore, applications are designed to run on thousands of nodes, often spanning across several programming models and generating large amounts of data. In this context, reproducibility is essential to foster reliable scientific results. In this paper we aim at studying the requirements and pitfalls of each stage of experimental research, from data acquisition to data analysis, with respect to achieving reproducible results. We investigate state-ofthe- art experimental practices in parallel computing by conducting a survey on the papers published in EuroMPI 2013, a major conference targeting the MPI community. Our findings show that while there is a clear concern for reproducibility in the parallel computing community, a better understanding of the criteria for achieving it is necessary. © Springer International Publishing Switzerland 2014.

关键词： Data acquisition

来源：评论

学校读者我要写书评

暂无评论

Reproducible MPI micro-benchmarking isn't as easy as you think 14

Reproducible MPI micro-benchmarking isn't as easy as you thi...

引用

21st European MPI Users' group Meeting, EuroMPI/ASIA 2014

作者： Hunold, Sascha Carpen-Amarie, Alexandra Träff, Jesper Larsson Vienna University of Technology Faculty of Informatics Institute of Information Systems Research Group Parallel Computing Favoritenstrasse 16/184-5 Vienna1040 Austria

ISBN: (纸本)9781450328753

The Message Passing Interface (MPI) is the prevalent programming model for supercomputers. Optimizing the performance of individual MPI functions is therefore of great interest for the HPC community. However, a fair comparison of different algorithms and implementations requires a statistically sound analysis. It is often overlooked that the time to complete an MPI communication function does not only depend on internal factors such as the algorithm but also on external factors such as the system noise. Most noise produced by the system is uncontrollable without changing the software stack, e.g., the memory allocation method used by the operating system. Possibly controllable factors have not yet been identified as such in this context. We investigate several possible factors-which have been discovered in other microbenchmarks-whether they have a significant effect on the execution time of MPI functions. We experimentally and statistically show that results obtained with other common benchmarking methods for MPI functions can be misleading when comparing alternatives. To overcome these issues, we explain how to carefully design MPI micro-benchmarking experiments and how to make a fair, statistically sound comparison of MPI implementations. © ACM 2014.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Density estimation with adaptive sparse grids for large data sets 14

Density estimation with adaptive sparse grids for large data...

引用

14th SIAM International Conference on Data Mining, SDM 2014

作者： Peherstorfer, Benjamin Pflüger, Dirk Bungartz, Hans-Joachim Aerospace Computational Design Laboratory AeroAstro Massachusetts Institute of Technology United States Simulation of Large Systems Group Institute for Parallel and Distributed Systems University of Stuttgart Germany Scientific Computing Group Department of Computer Science Technische Universität München Germany

ISBN: (纸本)9781510811515

Nonparametric density estimation is a fundamental problem of statistics and data mining. Even though kernel density estimation is the most widely used method, its performance highly depends on the choice of the kernel bandwidth, and it can become computationally expensive for large data sets. WTe present an adaptive sparse-grid-based density estimation method which discretizes the estimated density function on basis functions centered at grid points rather than on kernels centered at the data points. Thus, the costs of evaluating the estimated density function are independent from the number of data points. We give details on how to estimate density functions on sparse grids and develop a cross validation technique for the parameter selection. We show numerical results to confirm that our sparse-grid-based method is well-suited for large data sets, and, finally, employ our method for the classification of astronomical objects to demonstrate that it is competitive to current kernel-based density estimation approaches with respect to classification accuracy and runtime. Copyright © SIAM.

关键词： Data mining

来源：评论

学校读者我要写书评

暂无评论

On weighting schemes for gene order analysis

On weighting schemes for gene order analysis

引用

2013 German Conference on Bioinformatics, GCB 2013

作者： Bernt, Matthias Wieseke, Nicolas Middendorf, Martin Parallel Computing and Complex Systems Group Institute of Computer Science University Leipzig Germany

ISBN: (纸本)9783939897590

Gene order analysis aims at extracting phylogenetic information from the comparison of the order and orientation of the genes on the genomes of different species. This can be achieved by computing parsimonious rearrangement scenarios, i.e. to determine a sequence of rearrangements events that transforms one given gene order into another such that the sum of weights of the included rearrangement events is minimal. In this sequence only certain types of rearrangements, given by the rearrangement model, are admissible and weights are assigned with respect to the rearrangement type. The choice of a suitable rearrangement model and corresponding weights for the included rearrangement types is important for the meaningful reconstruction. So far the analysis of weighting schemes for gene order analysis has not been considered sufficiently. In this paper weighting schemes for gene order analysis are considered for two rearrangement models: 1) inversions, transpositions, and inverse transpositions;2) inversions, block interchanges, and inverse transpositions. For both rearrangement models we determined properties of the weighting functions that exclude certain types of rearrangements from parsimonious rearrangement scenarios.

关键词： Genes

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：