In this paper, we present a parallel simulator (SWiMNet) for PCS networks using a combination of optimistic and conservative paradigms. The proposed methodology exploits event precomputation permitted by model indepen...
详细信息
In this paper, we present a parallel simulator (SWiMNet) for PCS networks using a combination of optimistic and conservative paradigms. The proposed methodology exploits event precomputation permitted by model independence within the PCS components. The low percentage of blocked calls is exploited in the channel allocation simulation of precomputed events by means of an optimistic approach. Experiments were conducted with various call arrival rates and mobile host densities on a cluster of Pentium workstations. Performance results indicate that the SWiMNet achieves a speedup of 6 employing 8 workstations, and a speedup of 12 with 16 workstations.
Environment aware, self adaptive Time Warp logical processes (LPs) implementing a pro-active performance control scheme, thus addressing the timeliness of control decisions, are introduced. The ability of these pro-ac...
详细信息
Environment aware, self adaptive Time Warp logical processes (LPs) implementing a pro-active performance control scheme, thus addressing the timeliness of control decisions, are introduced. The ability of these pro-active TW LPs to adapt to sudden load changes is examined. Experimental results show that the protocol is able to capture abrupt changes in both computational and communication resource availability.
The performance of parallel discrete event simulation protocols is heavily dependent on the lookahead of the simulation model. Identifying and expressing correct lookahead for a model isn't easy, nor is it a well-...
详细信息
The performance of parallel discrete event simulation protocols is heavily dependent on the lookahead of the simulation model. Identifying and expressing correct lookahead for a model isn't easy, nor is it a well-defined process. In this paper, a global view of a PDES model as a set of data flows is presented. Using this view, we show how the lookahead of the model can be optimized, and we present various simplified implementations of this global view and the significant performance improvements generated when applied to real world models.
Clusters of workstations represent today a satisfactory alternative to MPPs and supercomputers in many areas of application. The rapidly reduction of the cost of high performance workstations/PCs makes this technology...
详细信息
Clusters of workstations represent today a satisfactory alternative to MPPs and supercomputers in many areas of application. The rapidly reduction of the cost of high performance workstations/PCs makes this technology ever more available. Moreover, new concepts for the integration of individual workstations through Local Area Networks are emerging. High speed interconnection networks and optimized protocol system architectures are the most important objectives of current research in this field of study. In this contribution, we attempt to propose a simple but effective performance model of systems with distributed computational and I/O resources when executing parallel scientific applications characterized by communication bursts and by intensive I/O phases. By means of queueing network techniques, the analysis of the model lends to the definition of a speedup surface which captures the relative influence of processors and disks parallelism in the performance of applications that alternates computations and I/O operations in a cyclic fashion.
Branch and Bound is a standard method for searching an optimal solution in the scope of continuous and discrete Global Optimization. It iteratively creates a search tree where each node represents a problem which is d...
详细信息
Branch and Bound is a standard method for searching an optimal solution in the scope of continuous and discrete Global Optimization. It iteratively creates a search tree where each node represents a problem which is decomposed in several subproblems provided that a feasible solution can be found by solving this set of subproblems. The computational power needed to solve most of the Branch and Bound Global Optimization problems and their high degree of potential parallelism make them suitable candidates to be solved in a multiprocessing environment. With parallel processing in mind Branch and Bound techniques can be considered as irregular and dynamic problems. So, their parallel implementations are not straightforward and require the use of dynamic load balance methods where the workload of a subproblem is a crucial parameter. In this paper an efficient parallel approach to the Branch and Bound continuous Global Optimization problem is described. It is based on a centralized asynchronous parallel model and on the prediction of the work loan of the set of subproblems containing a feasible solution. The proposed dynamic load balancing model obtains an almost perfect work load balance with low communication overhead.
This paper describes how the VHDL hardware description language can be used to aid the analysis, design and implementation of adaptive array beamformers for use in teleconferencing environments using parallel architec...
详细信息
This paper describes how the VHDL hardware description language can be used to aid the analysis, design and implementation of adaptive array beamformers for use in teleconferencing environments using parallel architectures. Beamformer models are often implemented in traditional programming languages such as C, Pascal and FORTRAN. This paper argues that VHDL provides a number of benefits to designers of such systems including improved vision of algorithm architecture, a faster design cycle and a more parallel and implementation-orientated design. The level of abstraction that VHDL gives is an important aid to overcoming the implementation problems associated with designing microphone array beamformers. The traditional view of VHDL as tool for making implementations in silicon can thus be extended to view VHDL as a tool for thinking with, in some circumstances better than with conventional algorithmic computer languages.
A set of computers connected by the Internet can be seen as a MIMD virtual parallel computer with distributed memory. We call such a computer a Massively parallel Virtual Computer (MPVC). The implementation of an MPVC...
详细信息
A set of computers connected by the Internet can be seen as a MIMD virtual parallel computer with distributed memory. We call such a computer a Massively parallel Virtual Computer (MPVC). The implementation of an MPVC will be feasible in the next few years, as new technologies brine to Internet communication speeds undreamed of nowadays. Problems such as Monte Carlo simulations and image processing and generation are good candidates to take advantage of MPVCs, since they can use practically all the computing power they can get. Massively parallel Processing on the Internet introduces the problem of scalability, that is, how to administrate efficiently a large number of computers and an even larger number of tasks running in them, Previous approaches either do not scale well or let the program mer solve the problem by himself. This paper explains the Dynamic Virtual Hypercubes, a structure created to logically organize an MPVC. Dynamic Virtual Hypercubes provide a scalable and efficient way to handle a large number of computers and tasks, since they can adapt to changes in workload and number of available machines.
The aim of this paper is to present an easy and efficient method to implement alternating-line processes on current parallel computers. First we show how data locality has an important impact on global efficiency, whi...
详细信息
The aim of this paper is to present an easy and efficient method to implement alternating-line processes on current parallel computers. First we show how data locality has an important impact on global efficiency, which leads us to the conclusion that one-dimensional compositions are the most convenient ones for 2D problems. Once this is asserted, a parallel algorithm is presented for the solution of the distributed tridiagonal systems along the partitioned direction. The key idea is to pipeline the simultaneous resolution of many systems of equations, not parallelising each resolution separately. This approach presents good numerical and architectural properties, in terms of memory usage and data locality, and high parallel efficiencies are obtained. For the case of alternating-line processes, the election of the optimal decomposition is studied. The experimental results have been obtained on a Cray T3E.
LIPS is a system for distributed computing using idle-cycles in heterogeneous networks of workstations. Especially data- and compute-intensive applications in the field of cryptography and computer algebra have used t...
详细信息
LIPS is a system for distributed computing using idle-cycles in heterogeneous networks of workstations. Especially data- and compute-intensive applications in the field of cryptography and computer algebra have used the system. The system provides its user with the tuple space based generative communication paradigm of parallel computing as known from the coordination language LINDA. In LIPS, failures (fail stop failures) like crashed machines are handled transparently for the application. Dynamic Load Adaption, meaning removing application processes from machines not being idle any longer and migrating those processes to idle machines is based on the detection of crashed application processes and the (re)start of application processes on an idle machine. The implementation of Dynamic Load Adaption for LIPS applications is easy, because checkpoint generation and the restart from a checkpoint is independent from the other application processes. As the crash of an application process (assuming the machine and the operating system the application process resides survive) can be detected very fast, the used mechanism allows for fast adaption of the applications distribution to changes in the NOW availability.
暂无评论