Proceedings Euromicro Workshop on Parallel and Distributed Processing

25-27 Jan. 1995

Filter Results

Displaying Results 1 - 25 of 71
  • A parallel algorithm for "document segmentation"

    Publication Year: 1995, Page(s):516 - 521
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (467 KB)

    We present a parallel algorithm for physical segmentation of technical documents. The proposed method follows a "data parallel" approach, based on a divide and conquer implementation. A document page is statically partitioned into n equal-sized rectangular blocks, where n is the number of processors. Each processor independently finds a segmentation of its assigned block, according to the same rul... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Proceedings Euromicro Workshop on Parallel and Distributed Processing

    Publication Year: 1995
    Request permission for commercial reuse | PDF file iconPDF (22 KB)
    Freely Available from IEEE
  • Flexible data parallel training of neural networks using MIMD-Computers

    Publication Year: 1995, Page(s):27 - 32
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (432 KB)

    An approach to flexible and efficient data parallel simulation of neural networks on large scale MIMD machines is presented. We regard the exploitation of the inherent parallelism of neural network models as necessary if larger networks and training data sets respectively are to be considered. Nevertheless it is essential to provide the flexibility for investigating various training algorithms or ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A distributed diffusion method for dynamic load balancing on parallel computers

    Publication Year: 1995, Page(s):43 - 50
    Cited by:  Papers (8)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (616 KB)

    Parallel applications can be divided into tasks that can be executed simultaneously in different processors. Depending on prior knowledge about computational requirements of the problem, the assignment of tasks to processors can be guided in two ways: static and dynamic. We propose a new dynamic load balancing algorithm based on the diffusion approach which employs overlapping balancing domains to... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A load balancing approach for parallel database machines

    Publication Year: 1995, Page(s):51 - 58
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (584 KB)

    Parallel database systems have become a major tool for high performance information processing. These systems require efficient load balancing approaches to partition each relation and to allocate them to the parallel architecture. If the database is not partitioned in a balanced way, the execution of database operations might waste resources, reducing the performance of parallel system. In this s... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Distributing code in a parallel fine grain machine using the actor model

    Publication Year: 1995, Page(s):122 - 129
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (584 KB)

    One way to gain execution time on massively parallel systems is to minimise I/O operations by reducing code loading. A solution to this problem is to distribute code dynamically among all processors. The actor model is well suited to this execution model. Actors are modelled as processes which have to be duplicated at execution, at request and depending on the application growth. Our project aims ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A class hierarchy emulating virtual shared objects on message-passing systems

    Publication Year: 1995, Page(s):174 - 181
    Cited by:  Patents (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (724 KB)

    Multicomputer systems are increasingly becoming of primary interest for scalable performance. Due to their distributed memory they require use of the message-passing paradigm which, however, is considered as low level programming. Remote access facilities and reliable emulation of shared memory can hide message-passing behind a more abstract interaction protocol. In this paper we address parallel ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Dataflow model for handling cut in the LOGFLOW parallel prolog machine

    Publication Year: 1995, Page(s):60 - 67
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (504 KB)

    The Generalized Dataflow Model (GDM) and the Logicflow Model have been recently introduced for OR and pipeline AND-parallel execution of pure logic programs. An informal description of the COR, HCOR and CUT nodes of the Logicflow Model has been given and it was shown how the `cut' mechanism of Prolog can be realised based on the Logicflow Model. In the current paper a more formal and exact definit... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Word parallelism vs spatial parallelism: a performance optimization technique on the PAPRICA system

    Publication Year: 1995, Page(s):236 - 243
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (572 KB)

    Starting from the analysis of the hardware efficiency of SIMD systems which use an external memory for data storage, this paper discusses a critical point in hardware design. In particular it presents a technique aimed to the maximization of the data bus efficiency. This technique is based on the transformation of the initial data set into a packed one, and can be successfully implemented on syste... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Algorithm-based fault-tolerant programming in scientific computation on multiprocessors

    Publication Year: 1995, Page(s):374 - 382
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (668 KB)

    Efficient parallel algorithms proposed to solve many fundamental problems in scientific computation are sensitive to processor failures. Because of its low costs, algorithm-based fault tolerance is an interesting concept for introducing fault tolerance into existing multiprocessors. To facilitate fault-tolerant programming in scientific computation, we have modified and developed further an existi... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Experiences in dynamic placement of actors on multicomputer systems

    Publication Year: 1995, Page(s):130 - 137
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (700 KB)

    The availability of low-cost microprocessors and the advances in communication technologies have promoted a great interest in multicomputer architectures. These systems are characterized by high performance, availability and extensibility at low cost but nowadays lack of good programming tools which could help the programmer to really exploit the resources of the computing system. One of the most ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A PRAM oriented language

    Publication Year: 1995, Page(s):182 - 191
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (576 KB)

    A new programming language oriented to the PRAM model is presented and results of its implementation on transputer networks are reported. During the last four years of teaching at the Computer Science Faculty of La Laguna University, the new language has proved to be a practical tool for the study, design, analysis and verification of PRAM algorithms. The theoretical analysis predicts the good per... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A Shared Memory model based on the Dataflow paradigm

    Publication Year: 1995, Page(s):68 - 75
    Cited by:  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (500 KB)

    In this paper a Shared Memory model designed to be efficiently implemented on a collection of autonomous processing elements is presented. The model is inspired by the Dataflow paradigm. The main features of the model are: the single assignment principle, multi-dimensional addressing scheme, and associative data access primitives. The single assignment principle limits the number of updates of a u... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Correctness of a distributed deadlock resolution algorithm for the single request model

    Publication Year: 1995, Page(s):254 - 261
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (596 KB)

    We consider the problem of the distributed deadlock resolution. Starting from a high level specification of the problem and the resolution algorithm for a system with single request model, we provide successive levels of decreasing abstraction of the initial specification in order to achieve a solution in a complete distributed system. The successive refinements and the final distributed deadlock ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An object oriented approach to fault tolerant software

    Publication Year: 1995, Page(s):383 - 390
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (532 KB)

    In this paper we are interested in the use of software techniques for fault tolerance in the framework of distributed object oriented programming. We survey different degrees of integration between paradigms for object oriented software development and those for fault tolerance. We then provide the outline of a unifying approach to fault tolerance in object oriented systems. This approach, named M... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Exploiting locality on parallel irregular problem computations

    Publication Year: 1995, Page(s):2 - 9
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (524 KB)

    By now, irregular problems are difficult to parallelize in an automatic way because of their lack of regularity in data access patterns. Most times, programmers must hand-write a particular solution for each problem separately. In this paper we present two pseudo-regular distributions which can be applied to partition most problems achieving very good average case distributions. Also, we have desi... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Distributed shared memory for function-grained graph reduction machine

    Publication Year: 1995, Page(s):148 - 155
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (584 KB)

    The paper proposes an efficient DSM system, which enables tasks to access the virtual global memory, for the function grained graph reduction machine. The function grained graph reduction machine is a parallel machine with distributed memory, which stores the graph nodes representing function applications in the distributed memory and enables its processing elements to reduce them by executing tas... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Minimizing network contention for mapping tasks onto massively parallel computers

    Publication Year: 1995, Page(s):210 - 218
    Cited by:  Papers (1)  |  Patents (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (664 KB)

    This paper presents a new framework aimed at compile-time determination of satisfactory sub-optimal solutions to the mapping problem onto modern massively parallel computing systems. The approach incorporates realistic assumptions on the models both for parallel programs and target architectures. It is refined for the K-ary n-cube family of processor networks that use a deterministic routing algor... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Pattern-driven automatic program transformation and parallelization

    Publication Year: 1995, Page(s):76 - 83
    Cited by:  Papers (7)  |  Patents (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (732 KB)

    The PARAMAT system is able to automatically parallelize a wide class of sequential numeric codes operating on dense vectors, matrices etc. without any user interaction, for execution on distributed memory message-passing multiprocessors. A powerful pattern recognition tool locally identifies program semantics and concepts in scientific codes. It also works for dusty deck codes that have been `encr... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A control architecture for managing instructions among partitions of a data parallel structure

    Publication Year: 1995, Page(s):262 - 269
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (632 KB)

    The problem considered is the definition and evaluation of control strategies for handling image features spanning over several partitions of a parallel data structure, whenever such partitions are assigned to different nodes of a distributed memory MIMD parallel machine. The solutions proposed are based on a tree structure of MANAGER processes, which handle communication of information between WO... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A parallel modified block Lanczos' algorithm for distributed memory architectures

    Publication Year: 1995, Page(s):424 - 431
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (448 KB)

    In this paper we propose a parallel block Lanczos algorithm suitable for MIMD distributed memory message passing architectures. We first consider a direct parallelization of the classic block Lanczos algorithm and we evaluate its performance. Then, after a discussion of these results, we reorganize the block algorithm obtaining a modified version that has a better behaviour with respect to the per... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Hierarchical architectures for computer vision

    Publication Year: 1995, Page(s):392 - 398
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (564 KB)

    High computer performance depends only partially on using faster and more reliable hardware, but to a large extent it depends on the architecture and on the processing techniques. An effective platform that matches general planning strategies is given by the hierarchical paradigm. This is true particularly in the field of image processing and computer vision, which is characterized by very large q... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Iterated Runge-Kutta methods on distributed memory multiprocessors

    Publication Year: 1995, Page(s):12 - 19
    Cited by:  Papers (3)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (508 KB)

    In this article, we consider the iterated Runge-Kutta (IRK) method which is an iteration method based on a predictor-corrector scheme for the solution of ordinary differential equations. The method uses embedded formulae to control the stepsize. We present different algorithms of the IRK method on distributed memory multiprocessors using appropriate communication primitives. The theoretical perfor... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A hierarchical locality algorithm for NUMA compilation

    Publication Year: 1995, Page(s):106 - 113
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (560 KB)

    A compiler algorithm which exploits program locality and reduces the latency overhead in parallel hierarchical memory machines is described. By applying the appropriate transformation at different levels of the hierarchy, the amount of nonlocal accesses between processors is minimised. Similarly, the memory structure within a processor is exploited so reducing the amount of communication between l... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Enumeration of dense non-convex iteration sets

    Publication Year: 1995, Page(s):156 - 163
    Cited by:  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (600 KB)

    The enumeration of points contained in a polyhedron is one of the key algorithmic problems in the transformation of scientific programs. However, current algorithms can only operate on convex and “regularly non-convex” polyhedra. If the iteration sets to be enumerated do not fit in either category, the final code must scan a superset of the union of iteration domains and determine at r... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.