摘要:

AbstractMost methods for programming loosely coupled systems are based on message‐passing. Recently, however, methods have emerged based on ‘virtually’ sharing data. These methods simplify distributed programming, but are hard to implement efficiently, as loosely coupled systems do not contain physical shared memory. We introduce a new model,the shared data‐object model, that eases the implementation of parallel applications on loosely coupled systems, but can still be implemented efficiently.In our model, shared data are encapsulated in passive data‐objects, which are variables of user‐defined abstract data types. To speed up access to shared data, data‐objects are replicated. This ability to replicate objects is a significant difference with other object‐based models (e.g. Emerald and Amber). Also, by replicating logical objects rather than physical pages, our model has many advantages over shared virtual memory systems.This paper discusses the design choices involved in replicating objects and their effect on performance. Important issues are: how to maintain consistency among different copies of an object; how to implement changes to objects; which strategy for object replication to use. We have implemented several options to determine which ones are the

ISSN:1040-3108    

DOI:10.1002/cpe.4330040502

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

摘要:

AbstractThe present paper describes the implementation of multi‐block codes, used to model complex 2‐D geometries for applications in computational fluid dynamics on massively parallel architectures. The work starts with a brief description of ongoing and planned major aerospace projects and gives an estimate of the computing power needed. In order to provide this computational speed, one has to resort to massively parallel systems. In the first section the essential features of multi‐block grids, along with the grid generation equations are discussed and it is shown that overlapping multi‐block grids are inherently parallel by construction. Since the number of blocks is not fixed, but can be matched to a large extent to the number of available processors, there are no principal limitations of this parallelization approach, provided the ratio of computation time to communication time remains large enough, which leads to the discussion of problem scalability. The details of implementation on the Intel iPSC/2 of a general 2‐D multi‐block mesh‐generation code are outlined in sections 2 and 3, together with the listings of the major communication function (Section 4). In section 5 the results for this code are presented, clearly demonstrating that the multi‐block concept is a viable tool for massively parallel computers, which can be applied to virtually all problems in science and engineering where computational meshes are used. In section 5.2 an outlook on the parallelization of more complex problems is given, and estimates for speed‐up and efficiency, based on the present experiences, are provided. It turns out that, as long as computation dominates communication time, which is usually the case for complex aerospace applications, parallelization will be the tool to provide the additional orders of magnitude of computing power needed to routinely design and analyse future aircraft as well as spacecraft, in particular at high Mach numbers, when chemical reactions

ISSN:1040-3108    

DOI:10.1002/cpe.4330040503

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

摘要:

AbstractWe report our experience in developing a compiler and a distributed run‐time kernel for distributed computers, calledEchidna. Echidnais a software package which has been available for more than two years, and aims to support protocol designers in protocol modeling, validation and performance evaluation. It is basically oriented towards the rapid prototyping of distributed algorithms on distributed computers: the Intel iPSC hypercube, the Supernode machine, and networks of Transputers and Suns were considered. Algorithms are described using an ISO formal description technique calledEstelle.First, we present and justify the concept ofexperimentationon distributed algorithms for which our Estelle compiler has been designed. Then we discuss how Estelle constructs are mapped ontoCstructures and how they are interpreted by a distributed run‐time kernel. We conclude by presenting typical uses ofEchi

ISSN:1040-3108    

DOI:10.1002/cpe.4330040504

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

摘要:

AbstractOn‐line computation of forward and inverse Jacobian matrices is essential in robot manipulator controllers, where high‐speed robot motion is required. The complexity of Jacobian calculation is such that the computational burden is large, and parallel processing is necessary if on‐line computation is to be achieved. Various algorithms and parallel‐processing networks suitable for this are considered. All algorithms have been implemented on transputer networks and computation times measured. The paper emphasises the importance of including communication overheads in comparisons of the computational efficiency of alternative algorithms and processor networks. Theoretical processing times based on computer cycle times and arithmetic operation counts are shown to be a false basis for comparison.Whilst considering the specific case of computation of Jacobian matrices for a robot manipulator, the paper provides a useful example of the considerations and constraints involved in distributing any algorithm across a multi‐processo

ISSN:1040-3108    

DOI:10.1002/cpe.4330040505

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY

ISSN:1040-3108    

DOI:10.1002/cpe.4330040501

出版商:John Wiley&Sons, Ltd    

年代:1992

数据来源: WILEY