this paper identifies some of the issues that need to be explored in systems that support content-based delivery. distributed systems today are based on address-based delivery of messages. Each message carries the uni...
ISBN:
(纸本)0818691948
this paper identifies some of the issues that need to be explored in systems that support content-based delivery. distributed systems today are based on address-based delivery of messages. Each message carries the unique addresses of the intended recipients. the idea in content-based delivery is that information is delivered to those objects that have subscribed for that information. the ideas on content-based delivery reported here are based partly on Roman Ginis' doctoral dissertation at Caltech. the ideas on directories reported here are based partly on Eve Schooler's doctoral dissertation at Caltech.
this paper addresses the design of distributed methods which incorporate numerical extrapolation into adaptive multivariate integration, in order to increase the functionality of the integration algorithms. When attem...
详细信息
ISBN:
(纸本)0818691948
this paper addresses the design of distributed methods which incorporate numerical extrapolation into adaptive multivariate integration, in order to increase the functionality of the integration algorithms. When attempting to deal with singularities, adaptive integration algorithms need a very fine subdivision in the proximity of these "hot spots". this is not practical in higher dimensions where a vast number of subregions result. these problems may be alleviated through the use of a suitable extrapolation strategy. We present a strategy which, incorporated as a global extrapolation method in distributed adaptive integration, allows for load balancing.
in this paper, we extend the Reactive Approach to distributedcomputing. Reactive Systems allow one to use instantaneous broadcast events, which is a very powerful communication means between parallel components, in a...
详细信息
ISBN:
(纸本)081869209X
in this paper, we extend the Reactive Approach to distributedcomputing. Reactive Systems allow one to use instantaneous broadcast events, which is a very powerful communication means between parallel components, in addition to implicit synchronization. Extension to distribution implies that instantaneous broadcast becomes available over the network between local reactive areas. We describe a model of distributed reactive systems, called Synchronized Systems, and several algorithms of detection of instant termination to implement distributed instantaneous broadcast, in the infrastructure we choose to implement systems consist of distributed reactive machines connected to synchronizers. We present various implementations in Java, using SugarCubes, the RMI mechanism and another distributed Processing Environment.
Fine grained data distributions are widely used to balance computational loads across complete processes in parallel scientific applications. When a fine grained data distribution is used in memory, performance of I/O...
详细信息
ISBN:
(纸本)0818685794
Fine grained data distributions are widely used to balance computational loads across complete processes in parallel scientific applications. When a fine grained data distribution is used in memory, performance of I/O intensive applications can be limited not only by disk speed bur also by message passing, because a large number of small messages may be generated by the implementation strategy used in the underlying parallel file system or parallel I/O library. Combining (or packetizing) a set of small messages into a large message is generally known to speed up parallel I/O. However, overall I/O performance is affected not only by small messages bur also by other factors like cyclic block size and interconnect characteristics. We describe small message combination and communication scheduling for fine grained delta distributions in the Panda parallel I/O library mid analyze I/O performance on parallel platforms having different interconnects: IBM SP2, IBM workstation cluster connected by FDDI and Pentium It cluster connected by Myrinet.
Networks of workstations (NOWs) are cost-effective alternatives to multiprocessor systems. Recently, NOWs have been proposed for parallel query processing. Idle CPU cycles of workstations in a NOW-based system can be ...
详细信息
ISBN:
(纸本)0818691948
Networks of workstations (NOWs) are cost-effective alternatives to multiprocessor systems. Recently, NOWs have been proposed for parallel query processing. Idle CPU cycles of workstations in a NOW-based system can be used to process database query operations. In this paper Mle report performance of the hierarchical architecture for parallel query processing on a NOW: We have implemented the hierarchical architecture using PVM on a Pentium-based NOW. the experimental results reported here suggest that the hierarchical architecture is successful in achieving good scaleups and speedups indicating that the idle processor cycles are effectively used for query processing. Hierarchical system can also handle both light and heavy workloads in a load sharing fashion. Our results also suggest that the performance is sensitive to the minimum fragmentation size (chunk size) for partial operations and the structure of queries.
Dynamic Local Search [1] has been applied to the evolution of interactions between protein-like structures. these are composed of it randomly selected sequence of amino acids that are linked together to form linear po...
详细信息
ISBN:
(纸本)3540643591
Dynamic Local Search [1] has been applied to the evolution of interactions between protein-like structures. these are composed of it randomly selected sequence of amino acids that are linked together to form linear polymers in three dimensions. the objective function chosen for optimisation is the potential energy given by a Toy protein model. Proteins fold, move and interact with other chains to minimise their objective function at a given rate, F-rate, depending on the sum of the rates for re-organisation of their structures. the interaction between different proteins gives a whole range of local attraction/repulsion regimes that result in new structures with new bonds, broken bonds and recursive loops.
Concurrency and communication are two of the key features of distributed systems. these features can make systematic resting of distributed systems a complex task. A major problem is the explosion of the test space be...
详细信息
ISBN:
(纸本)0818691948
Concurrency and communication are two of the key features of distributed systems. these features can make systematic resting of distributed systems a complex task. A major problem is the explosion of the test space because of the potential for arbitrary interference of concurrent threads. this paper describes an approach for systematic resting of such systems in an object-oriented context. We use state-charts for system specification, and model the system behavior as event-sequences. A test case, therefore, is primarily an event-sequence with concurrent threads represented as interleaving events. Communication-states with associated events represent communication between objects. the test-space explosion is controlled by an extension to Chow's algorithm for generating test sequences for finite state machines. the number of test sequences we require is O(n(2)), where n is the sum of all events in ail concurrent statecharts.
this paper proposes a new compile time scheduling algorithm for distributed-memory systems, called Global Load Balancing (GLB). GLB is intended as the second step in the multi-seep class of scheduling algorithms. Expe...
详细信息
ISBN:
(纸本)0818691948
this paper proposes a new compile time scheduling algorithm for distributed-memory systems, called Global Load Balancing (GLB). GLB is intended as the second step in the multi-seep class of scheduling algorithms. Experimental results show that compared with known scheduling algorithms of the same low-cost complexity, the proposed algorithm improves schedule lengths lip to 30%. Compared to algorithms with higher order complexities, the typical schedule lengths obtained withthe proposed algorithm are at most twice longer.
We present a dynamic distributed load balancing algorithm for parallel, adaptive finite element simulations using preconditioned conjugate gradient solvers based on domain-decomposition. the load balancer is designed ...
详细信息
ISBN:
(纸本)3540648097
We present a dynamic distributed load balancing algorithm for parallel, adaptive finite element simulations using preconditioned conjugate gradient solvers based on domain-decomposition. the load balancer is designed to maintain good partition aspect ratios. It can calculate a balancing flow using different versions of diffusion and a variant of breadth first search. Elements to be migrated are chosen according to a cost function aiming at the optimization of subdomain shapes. We show how to use information from the second step to guide the first. Experimental results using Bramble's preconditioner and comparisons to existing state-ot the-art load balancers show the benefits of the construction.
the proceedings contain 35 papers. the topics discussed include: a threaded sliding window executor for irregular computation on a NOW;parallel profile matching for large scale webcasting;large-scale SVD and subspace-...
ISBN:
(纸本)3540648097
the proceedings contain 35 papers. the topics discussed include: a threaded sliding window executor for irregular computation on a NOW;parallel profile matching for large scale webcasting;large-scale SVD and subspace-based methods for information retrieval;thick-restart Lanczos method for symmetric eigenvalue problems;portable parallel adaptation of unstructured 3D meshes;locality preserving load balancing with provably small overhead;tree-based parallel load balancing methods for solution-adaptive unstructured finite element models on distributed memory multicomputers;coarse-grid selection for parallel algebraic multigrid;overlapping and short-cutting techniques in loosely synchronous irregular problems;using multithreading for the automatic load balancing of adaptive finite element meshes;a robust and scalable library for parallel adaptive mesh refinement on unstructured meshes;and parallelization of an unstructured grid, hydrodynamic-diffusion code.
暂无评论