distributedcomputing provides a cost-effective solution for computation intensive problems. With the emerging of networking operating system for personal computer (PC), such as WindowsNT, it is now feasible to develo...
详细信息
ISBN:
(纸本)0819425885
distributedcomputing provides a cost-effective solution for computation intensive problems. With the emerging of networking operating system for personal computer (PC), such as WindowsNT, it is now feasible to develop distributedcomputing on a network of PCs. In addition, the computing power delivered by a PC is kept increasing whilst the cost is decreasing. Implying that the performance/cost factor for a PC is high and tile computing power delivered by the network is enormous. In this paper, we describe a software system which enables users to develop distributedcomputing program using the SPMD (Single Program Multiple Data) paradigm very quickly under the WindowsNT operating system. The programming model for the system is simple and a user can control the system through a graphical interface. The results show that our system provides a reasonable speedup in solving Image processing problems.
Advanced general-purpose parallel systems should be able to support diverse applications with different resource requirements without compromising effectiveness and efficiency. We present a new resource management mod...
详细信息
Advanced general-purpose parallel systems should be able to support diverse applications with different resource requirements without compromising effectiveness and efficiency. We present a new resource management model for cluster computing that allows multiple scheduling policies to co-exist dynamically. In particular, we have built Octopus, an extensible and distributed hierarchical scheduler that implements new space-sharing, gang-scheduling and load-sharing strategies. A series of experiments performed on an IBM SP2 suggest that Octopus can effectively match application requirements to available resources, and improve the performance of a variety of parallel applications within a cluster.
An optimal parallel algorithm for computing all-pair shortest paths on doubly convex bipartite graphs is presented here. Our parallel algorithm runs in O(logn) time with O(n(2)/logn) processors on an EREW PRAM and is ...
详细信息
ISBN:
(纸本)0818682272
An optimal parallel algorithm for computing all-pair shortest paths on doubly convex bipartite graphs is presented here. Our parallel algorithm runs in O(logn) time with O(n(2)/logn) processors on an EREW PRAM and is time-and-work-optimal. As a by-product, we show that the problem can be solved by a sequential algorithm in O(n(2)) time optimally on any adjacency list or matrix representing a doubly convex bipartite graph. The result in this paper improves a recent work on the problem for bipartite permutation graphs, which are properly contained in doubly convex bipartite graphs.
distributed Shared Memory (DSM) offers programmers a shared memory abstraction on top of an underlying network of distributed memory machines. advances in network technology and price/performance of workstations sugge...
详细信息
distributed Shared Memory (DSM) offers programmers a shared memory abstraction on top of an underlying network of distributed memory machines. advances in network technology and price/performance of workstations suggest that DSM will be the dominant paradigm for future high-performance computing. However, as long running DSM applications scale to hundreds or even thousands of machines, the probability of a node or network link failing increases. Fault tolerance is typically achieved via `checkpointing' techniques that allow applications to `roll back' to a recent checkpoint rather than restarting. High-performance DSM systems using relaxed memory consistency are significantly more difficult to checkpoint than uniprocessor or message passing architectures. This paper describes previous approaches to checkpointing message passing parallel programs along with extensions to DSM systems.
Frameworks for parallelcomputing have recently become popular as a means for preserving parallel algorithms as reusable components. Framework for parallelcomputing in general, and POET in particular, focus on findin...
详细信息
ISBN:
(纸本)0818677430
Frameworks for parallelcomputing have recently become popular as a means for preserving parallel algorithms as reusable components. Framework for parallelcomputing in general, and POET in particular, focus on finding ways to orchestrate cooperation between objects implementing parallel algorithms. POET seeks to be a general platform for scientific parallel algorithm components which can be modified, linked "mixed and matched" to a user's specification. The paper centers on the characteristics that allow these components to cooperate, and at the same time, remain "black-boxes" that users can drop into the frame without having to know the exquisite details of message-passing, data layout, etc.
In this paper, we suggest a new object-oriented functional language, OOId, which is based on a functional language fd and extended with object-oriented language features. We also discuss how to map OOId program onto D...
详细信息
ISBN:
(纸本)0818682272
In this paper, we suggest a new object-oriented functional language, OOId, which is based on a functional language fd and extended with object-oriented language features. We also discuss how to map OOId program onto DAVRID, which is a kind of multithreaded architecture based on the dataflow model. The comparison of execution time of two programs written in fd and OOId for the matrix multiplication problem, shows we can effectively run OOId program with just a little slow-down of execution time. In addition, the rate of stow-down is decreased, as the size of the matrix is larger. That means we can use good features od object-oriented language by paying just a little slow-down of execution time.
Gigabit network technologies have made it possible to combine workstations into a distributed, massively-parallel computer system. Middleware, such as distributed shared objects (DSO), attempts to improve programmabil...
详细信息
ISBN:
(纸本)0818678135
Gigabit network technologies have made it possible to combine workstations into a distributed, massively-parallel computer system. Middleware, such as distributed shared objects (DSO), attempts to improve programmability of such systems, by providing globally accessible 'object' abstractions, Researchers have developed consistency protocols for replicated 'memory' objects. These protocols are well suited to scientific applications but less suited to multimedia or groupware applications. This paper addresses the state sharing needs of complex distributed applications with (1) high-frequency symmetric accesses to shared objects, (2) unpredictable and limited locality of accesses, (3) dynamically changing sharing behavior, and (4) potential data races. We show that a DSO system exploiting application-level temporal and spatial constraints an shared objects can outperform shared object protocols which do not exploit application-level constraints. We compare our S(emantic) DSO against entry consistency using a sample application having the four properties mentioned above.
The C* language is a data-parallel extension of the C language which incorporates parallel data types. Since the C++ language provides operator overloading, a C++ library can implement the C* parallel extensions with ...
详细信息
ISBN:
(纸本)0818678763
The C* language is a data-parallel extension of the C language which incorporates parallel data types. Since the C++ language provides operator overloading, a C++ library can implement the C* parallel extensions with a similar syntax. Although library implementations are highly portable, some overheads make them impractical. The two major overheads incurred are temporaries in each operator application, and the inability to detect regular communication patterns The C++ overloading mechanism forces a temporary for each operator application. Also, regular communications in. C* are syntactically indistinguishable from general point-to-point communications. We tackled these problems extensively in a library. The template mechanism, a type parameterization in C++, is used to eliminate temporaries by delaying operator application and evaluating the entire expression at once. The polymorphic type dispatch mechanism is used to detect regular communications by assigning particular types to potentially regular communications. We have implemented the library on the CM-5, and compared its performance with the C* compiler using three simple examples. The techniques presented offers improved performance comparable to the C* compiler, which is close or 1.5 times slower in two examples, and even faster in one example.
The proceedings contains 14 papers on Concurrency: Practice and Experience. Topics discussed include: array remapping;parallel supercomputing;parallel programming;high performance computing;Nexus communication systems...
详细信息
The proceedings contains 14 papers on Concurrency: Practice and Experience. Topics discussed include: array remapping;parallel supercomputing;parallel programming;high performance computing;Nexus communication systems;Web-based parallel programming environment;Java-based global computing;and coarse grain distributedcomputing.
PVM-based parallel/distributed computation tools have been designed, implemented, and applied to two important mathematical algorithms. The tools make PVM easier to use and applicable to a wider class of computations....
详细信息
PVM-based parallel/distributed computation tools have been designed, implemented, and applied to two important mathematical algorithms. The tools make PVM easier to use and applicable to a wider class of computations. The application shows how advanced algebraic algorithms can take advantage of modern parallel/distributedcomputing with the aid of such tools. The tools and interfaces described include PVM-ET (a set of enhancement tools for PVM), PvmJobs (a general bag of jobs library that works with any user created job structure in a master/slave paradigm), and SaclibPvm (a simple software package interfacing SACLIB to PVM). The ability to interface symbolic computing to PVM allows us to tackle the parallelization of the Grobner Bases algorithm and the Characteristic Sets method, two very compute intensive algorithms important in algebraic computations. These algorithms, their parallelization, and experimental results are presented.
暂无评论