Showing papers on "Distributed memory published in 1970"

Sharing of microprograms between processors

Abstract: A multiprocessor system includes two microprogrammed processors, each having a different instruction repertoire and capable of executing separate programs or portions thereof independently. Both processors share a common memory unit and communicate through established groups of memory storage locations. One processor is word-oriented and processes data using a fixed word format while the other processor is character-oriented and processes data using a variable length format. The microprogrammable control elements of both processors are interconnected to permit the fixed word processor to share microprograms of the variable length processor for executing instructions not included in its repertoire.

Computer Simulation of Short-Term Memory: A Component-Decay Model

Abstract: Publisher Summary This chapter discusses the computer simulation of short-term memory. The first of these systems is the sensory storage or very-short-term memory, viewed as a peripheral or perceptual storage. There is general agreement that information stored in this type of memory decays with time and is lost in a matter of a few seconds or less. Another type of memory, short-term memory (STM), is the focus of these various models. Despite general agreement that information is lost from STM, there are two views as to the nature of this loss. One view is that information is lost as a result of decay; an alternative is that STM has a limited capacity so that items are lost by being replaced by new items entering the system. The model represents the human at an information-processing level of description. At this level, the human is viewed as having available a variety of processing mechanisms which can be employed to operate upon information.

The Organization of High-Speed Memory for Parallel Block Transfer of Data

Abstract: This paper describes the organization of a multi-module memory, designed to facilitate parallel block transfers. All modules are assumed to be identical, and the individual modules can fetch or store no more than one word or word group during any single memory cycle. Parallel block transfers are made possible in multimodule memories by utilizing a device called the memory circulator and by organizing the memory in a particular way. The memory circulator consists of a bank of interconnected registers, one for each memory, and control circuitry. The memory system is organized so that ascending logical addresses are distributed cyclically among the modules. If there are 2b modules, then any individual word is accessed by using the least significant b bits of a memory address to select a module and by using the remaining bits to select an address within a module. The memory circulator can load and store a contiguous block of 2b words by selecting all modules and broadcasting a single address to all modules. A contiguous block can be displaced in memory by a multiple of 2b words by broadcasting different load and store addresses for a block of data. The circulator control circuitry includes a masking capability so that blocks smaller than 2b can be moved in this fashion. When the displacement of a block transfer is not a multiple of 2b a physical circulation of the data in the memory circulator registers is required.

On Static Memory Allocation in Computer Systems

Abstract: An algorithm for allocation of memory resources is presented making use of a priori information about a computer program. This static allocation algorithm is based on a directed graph model of computations with particular attention to the memory allocation of programs assigned for execution on multiprocessor systems.

Memory Protection in Multiprocessor Computer Systems.

Abstract: : The paper presents a study of the memory protection problem in a multiprocessing computer system. The general area of memory protection is investigated in terms of the following three categories: memory access conflict resolution, private data protection, and common data protection. Memory access conflict resolution may be subdivided into two areas--fixed and dynamic priority assignment. Special emphasis is given to a procedure for the design of a dynamic memory access conflict resolution circuit. An example of this procedure is also given. Techniques for the protection of private and common data within memory are reviewed. (Author)

Symmetry And BEM On Transputers

Abstract: Geometrical and material symmetries are found in many linear field problems. They can be taken into account thanks to a rationale called the Group Representation Theory. This method consists in reducing an original problem into a family of smaller ones, the overall solution is the sum of the subproblems. In this paper, we propose the implementation of the Boundary Element Method on distributed memory computers for two-dimensional problems in this context. Different parallel strategies are analysed and compared. Experimental results are obtained on a Meiko Computing Surface with 16 T800 transputers.

A Parallel Multi-block Navier-Stokes Solver OnDistributed Memory Machines

Abstract: This work describes the development of a parallel communication strategy on distributed memory platforms and its application to a Navier-Stokes solver for compressible fluid flows. In the present method, distributed message passing libraries are used in order to allow inter -processor communications. In particular, the implementation on the IBM-9076 SP1 available at CRS4 has been performed using the native IBM message-passing library [3] in order to enhance the usage of the inter-processor switch.

A Parallel Finite Element Software Package:Design And Implementation

Abstract: This paper describes the design and implementation of a parallel finite element software package capable of solving linear boundary value problems on a network of transputers. Many of the design and development issues are discussed that had to be considered in the evolution of efficient software for running on a distributed memory MIMD computer. Typical experimental results are presented and discussed.

Parallelization Of Software For Coastal Hydraulic Simulations For Distributed Memory Parallel Computers Using FORGE 90

Abstract: Due to the increasing availability of powerful distributed memory parallel computers, the parallelization of existing sequential software is a very important issue. Since this is often difficult and time-consuming, the usage of software tools for analysis and (semi-)automatic parallelization may be of great help. In this study, the usage of the FORGE 90 software tool is illustrated by its application to two software packages in the field of coastal hydraulics. FORGE 90 consists of two parts : 'BaWmg FCUfGE W provides many facilities to analyze sequential Fortran programs, and the 'FORGE 90 parallelized is an interactive tool which can be used to create a parallel version of a Fortran program targeted to execute on a distributed memory machine. The former function of the package has been used to analyze the water quality model DIVAST. The latter function of FORGE 90 has been used to parallelize the hydrodynamic model mu-CSM.