Search results

With the availability of powerful personal computers (PCs), workstations and networking devices, the recent trend in parallel computing is to connect a number of individual workstations (PC and PC symmetric multiprocessor systems (SMP)) to solve computation‐intensive tasks in parallel way on such clusters (networks of workstations (NOW), SMP and Grid). In this sense, it is not more true to consider traditionally evolved parallel computing and distributed computing as two separate research disciplines. Current trends in high performance computing are to use NOW (and SMP) as a cheaper alternative to traditionally used massively parallel multiprocessors or supercomputers and to profit from unifying of both mentioned disciplines. The purpose of this paper is to consider the individual workstations could be so single PC as parallel computers based on modern SMP implemented within workstation.

Such parallel systems (NOW and SMP), are connected through widely used communication standard networks and co‐operate to solve one large problem. Each workstation is threatened similarly to a processing element as in a conventional multiprocessor system. But, personal processors or multiprocessors as workstations are far more powerful and flexible than the processing elements in conventional multiprocessors. To make the whole system appear to the applications as a single parallel computing engine (a virtual parallel system), run‐time environments such as OpenMP, Java (SMP), message passing interface, Java (NOW) are used to provide an extra layer of abstraction.

To exploit the parallel processing capability of such cluster, the application program must be paralleled. The effective way how to do it for (parallelisation strategy) belongs to a most important step in developing effective parallel algorithm (optimisation). To behaviour analysis, all overheads that have the influence to performance of parallel algorithms (architecture, computation, communication, etc.) have to be taken into account. In this paper, such complex performance evaluation of iterative parallel algorithms (IPA) and their practical implementations are discussed (Jacobi and Gauss‐Seidel iteration). On real application example, the various influences in process of modelling and performance evaluation and the consequences of their distributed parallel implementations are demonstrated.

The paper usefully shows that better load balancing can be achieved among used network nodes (performance optimisation of parallel algorithm). Generally, it claims that the parallel algorithms or their parts (processes) with more communication (similar to analyzed Gauss‐Seidel parallel algorithm) will have better speed‐up values using modern SMP parallel system as its parallel implementation in NOW. For the algorithms or processes with small communication overheads (similar to analysed Jacobi parallel algorithm) the other network nodes can be used based on single processors.

To address the problems of high performance computing by using the networks of workstations (NOW) and to discuss the complex performance evaluation of centralised and distributed parallel algorithms.

Defines the role of performance and performance evaluation methods using a theoretical approach. Presents concrete parallel algorithms and tabulates the results of their performance.

Sees that a network of workstations based on powerful personal computers belongs in the future and as very cheap, flexible and perspective asynchronous parallel systems. Argues that this trend will produce dynamic growth in the parallel architectures based on the networks of workstations.

We would like to continue these experiments in order to derive more precise and general formulae for typical used parallel algorithms from linear algebra and other application oriented parallel algorithms.

Describes how the use of NOW can provide a cheaper alternative to traditionally used massively parallel multiprocessors or supercomputers and shows the advantages of unifying the two disciplines that are involved.

Produces a new approach and exploits the parallel processing capability of NOW. Gives the concrete practical examples of the method that has been developed using experimental measuring.

This paper considers correction aspects of computer communication networks modelling with emphasis on their performance evaluation. In general, the problem is to achieve the highest possible performance given constraints on the system.

The paper reviews the application of the analytical methods, based on the queueing theory, to the computer communication systems and makes an extension of theory to the improvement of the developed analytical models. In this sense the paper describes the derivation of a correction factor for analytical models to study more precise their basic parameters (end‐to‐end delay, performance, etc.).

The contribution is in incorporating the derived correction factor to account for the real non‐exponential nature of the input to the transmission channels of computer communication systems. The produced results by corrected analytical model are compared with results previously reported in the literature to estimate the magnitude of improvement.

The improved analytical models were tested under various ranges of parameters, which influence the architecture of the computer communication networks and which are important for practical use.

The rapid rate of growth of computer‐based communication systems (e.g. distributed computer networks, mobile data networks) has resulted in a renewed and intensive interest in this area. Efficient design of their service facilities leads to the sharing of resources among users. Such public shared networks are largely oversubscribed by independent users, which make random demands on the network resources. The optimal resource allocation to satisfy such demands and the proper settlement of contention when demands exceed the capacity of the resources, constitute the problem of being able to understand and to predict system behaviour.

To behaviour analysis we can use both analytical and simulation methods. Modelling and simulation are methods, which are commonly used by performance analysts to represent constraints and optimise performance. Principally the application of analytical queuing theory results belongs to the preferred method in comparison to the simulation method, because of their ability to analyse also very large networks.

As new computer communication systems, such as distributed computer networks or mobile data networks, grow in scale and complexity, the problem of being able to understand and to predict system behaviour becomes increasingly important. To their behaviour analysis we can use both analytical and simulation methods. Principally the application of analytical queuing theory results belongs to the preferred method in comparison to the simulation method. In this sense the article describes the development, realisation and verification of the new analytical model for the study of the basic parameters (end‐to‐end delay, performance etc.) of distributed data networks (computer networks, mobile data networks). The suggested model considers for every node of the data network one part for its own node's activities (communication functions) and another one for the modelling of each node channel for data transmission. When using a multiprocessor system, as the modern node communication processor, the model for its own node activities is the more realistic M/D/r system (Poisson arrival process/Deterministic service time distribution/r server system) and for the every node transmission channel the M/M/1 system (Exponential service time, Single server system). The new developed analytical model includes the influence of the communication functions to the whole delay in each node of a computer communication network. The achieved results of the developed model are compared with the results of the commonly used analytical and simulation models to estimate the magnitude of improvement. Likewise the developed analytical model was tested under various ranges of parameters, which influence the architecture of the distributed data networks and which are important for practical use.

Describes the important role of decomposition strategy in parallel computing, and its influence is illustrated with the example of a very frequently used parallel algorithm for matrix multiplication. On the basis of the performed analysis of the parallel computers used worldwide divides them into the two basic groups which are considered from the programmer‐developer point of view to be very different. They also introduce the typical principal structures for both these groups of parallel computers and also their models. The paper then in an illustrative way describes the development of concrete parallel algorithm for matrix multiplication on various parallel systems. For each individual practical implementation of matrix multiplication there is introduced the derivation of its calculation complexity. The individual ways of developing parallel matrix multiplication and their implementations are compared, analysed and discussed from the point of view of the programmer‐developer and user. The very important role of the decomposition strategies that are mainly of the class of asynchronous parallel computers is highlighted.