检索结果-内蒙古大学图书馆

arXiv 2020年

作者： Brown, Nick Mark Bull, J. Bethune, Iain Edinburgh Parallel Computing Centre James Clerk Maxwell Building Kings Buildings Edinburgh United Kingdom

Iterative methods for solving large sparse systems of linear equations are widely used in many HPC applications. Extreme scaling of these methods can be difficult, however, since global communication to form dot products is typically required at every iteration. To try to overcome this limitation we propose a hybrid approach, where the matrix is partitioned into blocks. Within each block, we use a highly optimised (parallel) conventional solver, but we then couple the blocks together using block Jacobi or some other multisplitting technique that can be implemented in either a synchronous or an asynchronous fashion. This allows us to limit the block size to the point where the conventional iterative methods no longer scale, and to avoid global communication (and possibly synchronisation) across all processes. Our block framework has been built to use PETSc, a popular scientific suite for solving sparse linear systems, as the synchronous intra-block solver, and we demonstrate results on up to 32768 cores of a Cray XE6 system. At this scale, the conventional solvers are still more efficient, though trends suggest that the hybrid approach may be beneficial at higher core counts. Copyright © 2020, The Authors. All rights reserved.

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Applying type oriented programming to the PGAS memory model

arXiv

引用

arXiv 2020年

作者： Brown, Nick Edinburgh Parallel Computing Centre James Clerk Maxwell Building Kings Buildings Edinburgh United Kingdom

The Partitioned Global Address Space memory model has been popularised by a number of languages and applications. However this abstraction can often result in the programmer having to rely on some in built choices and with this implicit parallelism, with little assistance by the programmer, the scalability and performance of the code heavily depends on the compiler and choice of application. We propose an approach, type oriented programming, where all aspects of parallelism are encoded via types and the type system. The type information associated by the programmer will determine, for instance, how an array is allocated, partitioned and distributed. With this rich, high level of information the compiler can generate an efficient target executable. If the programmer wishes to omit detailed type information then the compiler will rely on well documented and safe default behaviour which can be tuned at a later date with the addition of types. The type oriented parallel programming language Mesham, which follows the PGAS memory model, is presented. We illustrate how, if so wished, with the use of types one can tune all parameters and options associated with this PGAS model in a clean and consistent manner without rewriting large portions of code. An FFT case study is presented and considered both in terms of programmability and performance - the latter we demonstrate by a comparison with an existing FFT solver. Copyright © 2020, The Authors. All rights reserved.

关键词： Program compilers

来源：评论

学校读者我要写书评

暂无评论

Large-scale lattice-Boltzmann simulations over lambda networks

Large-scale lattice-Boltzmann simulations over lambda networ...

引用

2007 Lighting the Blue Touchpaper for UK e-Science - Closing Conference of ESLEA Project, ESLEA 2007

作者： Saksena, Radhika S. Coveney, Peter V. Pinning, Robin L. Booth, Stephen P. Centre for Computational Science Department of Chemistry University College London 20 Gordon Street LondonWC1H 0AJ United Kingdom Manchester Computing University of Manchester Oxford Road ManchesterM13 9PL United Kingdom Edinburgh Parallel Computing Centre University of Edinburgh EdinburghEH9 3JZ United Kingdom

Amphiphilic molecules are of immense industrial importance, mainly due to their tendency to align at interfaces in a solution of immiscible species, e.g., oil and water, thereby reducing surface tension. Depending on the concentration of amphiphiles in the solution, they may assemble into a variety of morphologies, such as lamellae, micelles, sponge and cubic bicontinuous structures exhibiting non-trivial rheological properties. The main objective of this work is to study the rheological properties of very large, defect-containing gyroidal systems (of up to 10243 lattice sites) using the lattice-Boltzmann method. Memory requirements for the simulation of such large lattices exceed that available to us on most supercomputers and so we use MPICH-G2/MPIg to investigate geographically distributed domain decomposition simulations across HPCx in the UK and TeraGrid in the US. Use of MPICH-G2/MPIg requires the port-forwarder to work with the grid middleware on HPCx. Data from the simulations is streamed to a high performance visualisation resource at UCL (London) for rendering and visualisation. © Copyright owned by the author(s) under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike Licence.

关键词： Supercomputers

来源：评论

学校读者我要写书评

暂无评论

Design and implementation of an efficient multi-cluster GridRPC system 05

Design and implementation of an efficient multi-cluster Grid...

引用

IEEE/ACM International Symposium on Cluster computing and the Grid (CCGRID)

作者： Quoc-Thuan Ho Wentong Cai Yew-Soon Ong Institute of High Performance Computing Agency for Science Technology and Research Singapore Parallel & Distributed Computing Centre School of Computer Engineering Nanyang Technological University Singapore

ISBN: (纸本)9780780390744

Multi-cluster Grids have emerged as the most popular type of grid environments. On multi-cluster grid, resources are distributed across different networks. Hence, scheduling and dispatching jobs are difficult and inefficient due to the restrictions of network and the communication overhead. This paper presents the design and implementation of an efficient GridRPC system for the multi-cluster grid. Metascheduling mechanism is developed to distribute jobs onto different clusters of computers effectively. Further, a dynamic bundling mechanism is provided to reduce overhead of job dispatching. This mechanism bundles similar requests to form a composite request that is dispatched to a grid resource. The above mechanisms have been tested with different types of services and loads, and the results show that they help to achieve more balanced workload and reduce the overhead of job dispatching. This cuts down execution time of applications.

关键词： Dispatching Distributed computing Computer architecture Concurrent computing Grid computing Mesh generation High performance computing Processor scheduling Testing Network servers

来源：评论

学校读者我要写书评

暂无评论

Developing non-loeal iterative parallel algorithms for GIS on cray T3D using MPI 4th

Developing non-loeal iterative parallel algorithms for GIS o...

引用

4th European Conference on parallel Virtual Machine and Message Passing Interface, PVM/MPI 1997

作者： Clematis, A. Coda, A. Spagnuolo, M. Mineter, M. Sloan, T. Istituto per la Matematica Applicata CNR Via De Marim 6 Genova16149 Italy Geography Department Drummond Street Edinburgh United Kingdom Edinburgh Parallel Computing Centre King's Buildings Edinburgh United Kingdom

ISBN: (纸本)3540636978

In the application context of spatial data analysis, several parallelisation strategies are proposed for algorithms which are characterized by the non-locality of computation and by the iterative nature of the data processing. Different parallelisation approaches are presented and the methods developed for Cray T3D using the MPI message passing library are briefly compared to a workstation-network solution. The better results in the two environments are obtained following different strategies. The lesson learned is that MPI and other message passing libraries provide valuable help for portability, but the development of efficient and portable parallel software needs the availability of higher level approaches supporting different parallelisation strategies in a simple and effective way. © Springer-Verlag Berlin Heidelberg 1997.

关键词： Message passing

来源：评论

学校读者我要写书评

暂无评论

PARAMICS-moving vehicles on the connection machine 94

PARAMICS-moving vehicles on the connection machine

引用

Supercomputing Conference

作者： G. Cameron B.J.N. Wylie D. McArthur Edinburgh Parallel Computing Centre (EPCC) University of Edinburgh CSCS Lugano Switzerland SIAS Limited Edinburgh UK

ISBN: (纸本)9780818666056

PARAMICS is a parallel MICroscopic Traffic Simulator which is, to our knowledge, the most powerful of its type in the world. The simulator can model around 200,000 vehicles on around 7,000 roads (taken from real road traffic network data) at faster than 'real-time' rates, making use of 16 K processor TMC Connection Machine CM-200 for the simulation aspect. The project aims to make available to road network planners a new range of tools, and demonstrates that use of high performance computing in real applications is possible and worthwhile, while yielding important and interesting research results.< >

关键词： Traffic control Microscopy Road vehicles Telecommunication traffic Computational modeling parallel processing Computer applications Computer networks Concurrent computing Packaging machines

来源：评论

学校读者我要写书评

暂无评论

Instability in the molecular dynamics step of a hybrid Monte Carlo algorithm in dynamical fermion lattice QCD simulations

引用

Physical Review D 2000年第11期62卷 114501-114501页

作者： Bálint Joó Brian Pendleton Anthony D. Kennedy Alan C. Irving James C. Sexton Stephen M. Pickles Stephen P. Booth Edinburgh Parallel Computing Centre (EPCC) The University of Edinburgh Edinburgh EH9 3JZ Scotland United Kingdom

We investigate instability and reversibility within hybrid Monte Carlo simulations using a nonperturbatively improved Wilson action. We demonstrate the onset of instability as tolerance parameters and molecular dynamics step sizes are varied. We compare these findings with theoretical expectations and present limits on simulation parameters within which a stable and reversible algorithm is obtained for physically relevant simulations. Results of optimization experiments with respect to tolerance parameters are also presented.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Fault-Tolerance on regular decomposition grid applications

Fault-Tolerance on regular decomposition grid applications

引用

1995 Euromicro Workshop on parallel and Distributed Processing

作者： Silva, Luis Moura Silva, João Gabriel Chappie, Simon Clarke, Lyndon Departamento de Engenharia Informatica Universidade de Coimbra POLO II Coimbra3000 Portugal Edinburgh Parallel Computing Centre University of Edinburgh JCMB Mayfield Road EdinburghEH9 3JZ United Kingdom

ISBN: (纸本)0818670312

Writing parallel applications is considerably more complex due to additional problems not found in the sequential environment. The main problems include communication, synchronization, data partitioning and distribution, mapping of processes, heterogeneity and fault tolerance. Fault tolerance is a very important feature in parallel/distributed systems since the mean time between failures of the system decreases with the number of processors, and the failure of just one process(or) can lead to the crash of the entire application. This paper presents an example of a parallel library (PUL-RD) that solves most of the problems pointed out before and provides support for fault tolerance. The original version of the library offers high-level support for parallelism in a portable way and can be used to write grid-based parallel applications which have a regular decomposition. In this paper, we will describe the fault-tolerance issues that were incorporated into the PUL-RD, giving special attention to the functionality of the checkpointing scheme. © 1995 IEEE.

关键词： Fault tolerance

来源：评论

学校读者我要写书评

暂无评论

The collaborative P-GRADE Grid Portal

引用

5th International Conference on Computational Science - ICCS 2005

作者： Lewis, Gareth J. Sipos, Gergely Urmetzer, Florian Alexandrov, Vassil N. Kacsuk, Peter Advanced Computing and Emerging Technologies Centre School of Systems Engineering University of Reading WhiteKnights Reading RG6 6AY United Kingdom MTA SZTAKI Laboratory of Parallel and Distributed Systems H-1518 Budapest Hungary

ISBN: (纸本)9783540260448

Grid portals are increasingly used to provide uniform access to the grid infrastructure. This paper describes how the P-GRADE Grid Portal could be used in a collaborative manner to facilitate group work and support the notion of Virtual Organisations. We describe the development issues involved in the construction of a collaborative portal, including ensuring a consistent view between participants of a collaborative workflow and management of proxy credentials to allow separate nodes of the workflow to be submitted to different grids. © Springer-Verlag Berlin Heidelberg 2005.

关键词： Computational methods

来源：评论

学校读者我要写书评

暂无评论

A true random numbers service and its applications

A true random numbers service and its applications

引用

31st International Convention Proceedings: Microelectronics, Electronics and Electronic Technology, MEET and Grid and Visualizations Systems, GVS 2008

作者： Stevanović, Radomir Skala, Karolj Karaivanova, Aneta Atanassov, Emanouil Gurov, Todor Centre for Informatics and Computing Rudjer Boskovic Institute Bijenicka c. 54 Zagreb Croatia Institute for Parallel Processing Bulgarian Academy of Sciences Acad. G. Bonchev Str. Bl. 25-A Sofia Bulgaria

ISBN: (纸本)9789532330366

We give an overview of random number generators in scientific use today, concentrating on the "truly random" ones. Quantum Random Bits Generator Service serves as an example of these, being readily available - online and free, with known qualities of served content. Sensitivity analysis of a typical stochastic scientific applications (dominantly Monte Carlo integration methods) to a choice of random numbers generator is laid out, with results showing optimal choice of generator, concerning application-specific needs for data randomness, and speed. As a concrete real-world example of an randomnesscritical simulation, the SALUTE application is used, with results supporting the use of a truly random generators for parallel Monte Carlo simulations of physical systems. © 2008 by Mipro.

关键词： Sensitivity analysis

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：