We discuss the scalable parallel solution of the Poisson equation on irregularly shaped domains discretized by finite differences. the symmetric positive definite system is solved by the preconditioned conjugate gradi...
详细信息
ISBN:
(纸本)9783642281501;9783642281518
We discuss the scalable parallel solution of the Poisson equation on irregularly shaped domains discretized by finite differences. the symmetric positive definite system is solved by the preconditioned conjugate gradient algorithm with smoothed aggregation (SA) based algebraic multigrid (AMG) preconditioning. We investigate variants of the implementation of SA-AMG that lead to considerable improvements in the execution times. the improvements are due to a better data partitioning and the iterative solution of the coarsest level system in AMG. We demonstrate good scalability of the solver on a distributed memory parallel computer with up to 2048 processors.
A distributed computation is usually modeled as a finite partially ordered set (poset) of events. Many operations on this poset require computing meets and joins of subsets of events. the lattice of normal cuts of a p...
详细信息
Recently, collaborative works supported by computers through real time communication have been introduced to various human activities. New kinds of devices such as iPhone, iPad and Android terminals are in common use....
详细信息
Extending the researches on wavelength switched optical networks (WSON), efficient integration of the novel optical packet switching network and wavelength switching-based optical circuit switching network technologie...
详细信息
ISBN:
(纸本)9780819489883
Extending the researches on wavelength switched optical networks (WSON), efficient integration of the novel optical packet switching network and wavelength switching-based optical circuit switching network technologies which offers both best-effort packet delivery and QoS guaranteed lightpath services has been being studied. In addition, researches on the optical-layer transparent data processing, such as all-optical wavelength multicasting, all-optical 3R regeneration, etc, are conducted simultaneously. It is believed that future innovative optical network services (INSes) would be built on these novel future-proof technologies, and foster colorful applications in the new generation networks. Before the wide applications of INS in different fields, there would be a foreseeable strong requirement for INS firstly posed by pioneer grid applications, e. g., e-science, e-government, and e-banking, etc, which would require the high-performance underlying networks. Our research here is motivated to glue the optical networks and grid applications by integrating lightpath, geographically distributed INS systems and grid resources (e. g., computers, storages, instruments, etc.), and finally offering an easy-to-use high performance networked grid computing environment-optical grid network (OGN) to user applications. In this paper, we introduce our research activities of a distributed optical grid network infrastructure (OGNI), and the creation of the future easy-to-use INS based on OGNI. the proposals have been validated through field-trial experiments over a developed WSON testbed.
User Domains gathers and combines resources from multiple sources to create a per-user geographically distributed heterogeneous virtualization platform where user-provided virtual machines can be executed in user mode...
详细信息
the proceedings contain 29 papers. the topics discussed include: energy-efficient server rack cooling using feedforward control: benefits and problems;design of a reconfigurable pipelined switch for faulty on-chip net...
ISBN:
(纸本)9780889868649
the proceedings contain 29 papers. the topics discussed include: energy-efficient server rack cooling using feedforward control: benefits and problems;design of a reconfigurable pipelined switch for faulty on-chip networks;on BPC permutations admissibility to variable-stage hybrid optical shuffle-exchange networks;efficient data service design for a SOA approach to scientific computing;task allocation method for avoiding contentions by the information of concurrent communications;an online system for octuple-precision computation;performance evaluation of semi-fixed-priority scheduling on prioritized SMT processors;failure prediction in video-streaming servers through performance analysis of server and client-server interactions;adaptive thread scheduling techniques for improving scalability of software transactional memory;and reexamining the parallelization schemes for standard full tableau simplex method on distributed memory environments.
Efficient parallel programming has always been very tricky and only expert programmers are able to take the most of the computing power of modern computers. Such a situation is an obstacle to the development of the hi...
详细信息
We present an algorithm for multi-physics simulation of charged particles in electrokinetic flows. It includes a coupled simulation of charged rigid particles in fluid flows in an electric field. the parallel simulati...
详细信息
this paper describes chosen aspects of the European Integrated Tokamak Modelling Task Force (ITM-TF) effort. It covers topics related to latest developments towards providing fusion scientists with a software framewor...
详细信息
Adaptive mesh refinement and iterative traversals of unknowns on such adaptive grids are fundamental building blocks for PDE solvers. We discuss a respective integrated approach for grid refinement and processing of u...
详细信息
ISBN:
(纸本)9783642281440;9783642281457
Adaptive mesh refinement and iterative traversals of unknowns on such adaptive grids are fundamental building blocks for PDE solvers. We discuss a respective integrated approach for grid refinement and processing of unknowns that is based on recursively structured triangular grids and space-filling element orders. In earlier work, the approach was demonstrated to be highly memory-and cache-efficient. In this paper, we analyse the cache efficiency of the traversal algorithms using the I/O model. Further, we discuss how the nested recursive traversal algorithms can be efficiently implemented. For that purpose, we compare the memory throughput of respective implementations with simple stream benchmarks, and study the dependence of memory throughput and floating point performance from the computational load per element.
暂无评论