WoTUG - The place for concurrent processes

BibTeX Proceedings details

BibTex html text dump of PaperDB database.

PaperDB is GPL Software, see: http://paperdb.sourceforge.net.

  title = "{RT}-{DOS} -- {A} real-time distributed operating system for transputers",
  author= "Tayh, M. and Bor, M. and Benmaiza, M. and Eskicioglu, M. R.",
  editor= "Zedan, Hussein S. M.",
  pages = "1--11",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "This paper describes the design philosophy of a real-time
     operating system, RT-DOS, and its architecture. RT-DOS is a
     generic name given to a prototype operating system, which is
     a fully-distributed, message-based system designed to run on
     a network of transputers. The main objective of RT-DOS is to
     provide a dynamically reconfigurable work platform that
     adapts to dynamic changes in workloads, allows system
     maintainability and dynamic upgrading, and increases system
     reliability and availability."
  title = "{O}n the {F}easibility of {R}un-{T}ime {P}rocess {M}igration in {M}ulti-transputer {M}achines",
  author= "Jones, Peter and Cha, Hojung",
  editor= "Zedan, Hussein S. M.",
  pages = "12--27",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "A number of techniques for the amelioration of message
     transport costs in networked multiprocessor machines have
     been explored. Among these are established approaches such
     as (i) software implementation of message through routing
     systems, (ii) run-time installation of point-to-point
     physical connections between otherwise remote processors and
     (iii) pre-load manipulation of the machine topology to bring
     communicating processors close together. A further technique
     is proposed. This technique requires the run-time transport
     of processes, between processors, in order to reduce the
     distance over which communication must take place. This
     paper investigates the feasibility of such an approach for
     multi-transputer machines and indicates ways in which it can
     be implemented in practice."
  title = "{B}roadcast communication in fault tolerant multicomputer systems",
  author= "Gresser, K.",
  editor= "Zedan, Hussein S. M.",
  pages = "28--34",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "Fault tolerance can be achieved by means of redundancy.
     Multicomputer architectures allow duplicated tasks to run on
     multiple processors. In this paper we focus on multicomputer
     systems consisting of a small number of processors (4 to
     16). Fault tolerance instrumented in the operating system
     level tolerates hardware faults. This type of fault
     tolerance is known as software implemented fault tolerance.
     The information exchange of the computers requires a
     powerful communication system with fault tolerance
     properties. Since no standard system meets all requirements
     a new design was necessary. This paper describes concept,
     hardware, firmware and performance of the transputer based
     broadcast communication system (BCS)."
  title = "{T}ransputer performance issues using the trollius operating system",
  author= "Beers, James R. Jr. and Leibensperger, Ros and Braner, Moshe and Fielding, David",
  editor= "Zedan, Hussein S. M.",
  pages = "35--50",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The Trollius Opeiating System for distributed multicomputer
     offers different types of message-passing services,
     incorporating different amounts of overhead. Each of the
     operating system activities required for interprocess
     communication is designed as an individual module or layer.
     The different types of Trollius communication can be
     accessed by turning operating system services on or off, or
     by directly utilizing different layers of
     communication.Basically, there are three layers of internode
     communication. The network layer can be used from any node
     to any other node; the data link layer is for
     nearest-neighbor communication; and Ihe physical layer uses
     the hardware in the most efficient way possible, but
     requires exclusive use of the link. A separate level, the
     kernel level, is used for intranode communication. Operating
     system services that can be turned on or off include
     buffering and virtual circuits.Two different types of
     benchmarks are used to evaluate the performance of the
     different types of Trollius message-passing. The first type
     calculates message-passing lime as the sum of the limes
     required for each individual service or component of Ihe
     process. The decrease in message-passing lime obtained when
     turning off services can ihus be easily measured. The second
     benchmark measures throughput, the amount of information
     that can be sent from one node to another in a given amount
     of time, for different layers of message-passing, utilizing
     different sendees. This is the more accurate measure when
     series of messages are sent from one node to another. The
     throughput measurements demonstrate the value of the
     physical layer and of virtual circuits."
  title = "{H}igh performance event and {I}/{O} handling on the transputer",
  author= "Harley, R. G. and Levy, D. C. and Hemme, A. W. M. and Webster, M. R.",
  editor= "Zedan, Hussein S. M.",
  pages = "51--60",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "If transputers are to be used in high performance control
     applications it is essential that the I/O and event handling
     capabilities of the transputer are well understood. This
     requires insight into the hardware architecture and the low
     level language (guy code) of the transputer. The interface
     between the transputer executing the corresponding control
     process, and the external event generating devices can be
     divided into two main processes, namely interrupt or event
     handling and the I/O or data handling. For maximum
     performance these processes must be handled as quickly and
     efficiently as possible. Some of the issues involved in
     using occam to establish an interrupt handler are discussed
     in [1]. This paper extends that work to show how better
     performance and multiple event handling can be obtained
     efficiently. Several methods of achieving this real world
     interface are examined and, based on an actual design, the
     paper concludes with some recommendations to make the event
     handler more efficient."
  title = "{E}valuation of two systems for distributed message passing in transputer networks",
  author= "Avramov, N. N. and Knowles, A. E.",
  editor= "Zedan, Hussein S. M.",
  pages = "61--73",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "While special chips for message transfer in multi-transputer
     machines do not yet exist, there are several software
     solutions which may be used. Dynamic link reconfiguration is
     fastest but requires additional switching hardware and a
     sophisticated common bus to control requests from all
     network nodes. Through-routing of messages via the network
     is slower but the method is applicable to any transputer
     network. This paper makes a comparison between two
     through-routing solutions for distributed message passing
     and shows how the parameters of a software implementation
     influence the communication throughput. Several conclusions
     about the improvement of particular through-routing software
     are made and the results of the experimental performance
     evaluation under different conditions are given."
  title = "{A}n environment for transputer {CPU} load measurements",
  author= "Pietro, Giuseppe de and Villano, Umberto",
  editor= "Zedan, Hussein S. M.",
  pages = "74--82",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "In a multiprocessor system an uneven load balancing can
     usually dramatically reduce the performance of the parallel
     program running on it Hence it is of paramount importance to
     be able to estimate the CPU and communication loads of every
     task before the program is actually executed so that the
     optimal application partitioning can be found. In this paper
     the problem of CPU load measurement is tackled, and a
     measurement environment is illustrated in which the
     processes to be allocated to the processors in the network
     are run in quasi-concurrence on a single Transputer. A
     technique based on active process list manipulation makes it
     possible to perform a fairly accurate measurement of the CPU
     activity of the parallel processes in the application using
     the Transputer internal tinier as a reference clock."
  title = "{A}n operating environment for control systems",
  author= "Wijbrans, K. C. J. and Tillema, H. G. and Bakkers, Andr\`{e} W. P. and Schoute, Albert L.",
  editor= "Zedan, Hussein S. M.",
  pages = "83--94",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "This article describes an operating environment for control
     systems. This operating environment contains the basic
     layers of a distributed operating system. Typical for this
     operating environment is that it is designed according to
     the needs of control systems. To do this, first the general
     requirements of a controller have been investigated. Based
     on the requirements posed by controllers as they can be
     found in complex systems, the requirements of the operating
     environment have been derived. This operating environment
     has been implemented on transputers. To test the performance
     of the operating environment, performance indicators were
     chosen and performance measurements were carried out for
     several different strategies. Due to the demanding nature of
     real-time control systems, special attention has been paid
     to an efficient implementation of a basic kernel."
  title = "{L}i{BRA} -- {A} load balancing tool for a reconfigurable parallel computer",
  author= "Tambwekar, Sanjay and Shukla, U. S. and Paulraj, A.",
  editor= "Zedan, Hussein S. M.",
  pages = "95--107",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "Load balancing in MIMD message-passing parallel computers is
     essential to make efficient use of the system resources and
     reduce the program runtime. For parallel computers that also
     provide topological \^{}configurability, it is necessary for
     the load balancing strategy to not only find an optimum
     distribution of tasks to processors, but also to determine
     the best-suited interconnection pattern for the processors.
     In this paper, we present an off-line tool, LiBRA, that will
     assist in automating the process of load balancing. The user
     specifies his problem in terms of a computation graph and
     the machine characteristics. LiBRA uses simulated annealing
     with an automatic annealing schedule to generate the optimal
  title = "{F}ault tolerant computing with transputers and occam",
  author= "Nieuwenhuis, L. J. M. and Blom, G. D.",
  editor= "Zedan, Hussein S. M.",
  pages = "108--118",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "In this paper the results of a case study with Transputers
     and Occam for a systematic approach of fault tolerant
     computing is presented. An arbitrary Transputer system can
     be transformed into a fault tolerant version without using
     additional special hardware. Fault tolerance is based on
     software implemented replication. The fault tolerant version
     consists of copies of the original system. Processes on the
     original Transputers can automaticly be transformed into
     versions which can be executed by the Transputers of the
     fault tolerant system. The reliability of the resulting
     system is optimal and performance optimizing properties of
     the original system are preserved."
  title = "{T}he {D}evelopment of occam: types, classes and sharing",
  author= "Barrett, Geoff",
  editor= "Zedan, Hussein S. M.",
  pages = "119--147",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The proposed extensions to the occam* language are aimed at
     providing • a more comprehensive type system• support
     for a modular programming style• a facility for sharing
     between processes.The type system is similar to that of many
     modern programming languages but with a careful treatment of
     union types and without recursive types.Although it is
     possible to describe shared objects in occam2, the required
     idiom has an implementation whose complexity is linear in
     the number of users. By introducing a special sort of shared
     bus of channels, this problem can be overcome.The class
     system is designed in such a way as to allow for separate
     compilation and alien code classes to be used in occam
     programs with little overhead and to provide some of the
     abstraction mechanisms which have been recognised as
     beneficial in object-oriented languages.There are also a
     number of new language features which do not significantly
     change the nature of the language but which do enhance its
     general expressiveness.The first part of this paper presents
     proposed changes to the occam2 reference manual ([1]). The
     second part is a commentary on the decisions which had to be
     made in order to produce the proposal. The section numbers
     of the manual changes correspond to the section numbers of
     the occam2 reference manual where a ' denotes a change to an
     existing section and a letter denotes the insertion of a new
  title = "{C}ombining configuration and allocation",
  author= "Du, Dong-Hui and Vidal-Naquet, Guy",
  editor= "Zedan, Hussein S. M.",
  pages = "148--157",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "This paper describes a method which makes hardware
     configuration transparent to Occam/Transputer users by doing
     automatic configuration of Transputer networks and
     allocation of programs. Some heuristic algorithms of
     combining network configuration with task allocation have
     been put forward. From an Occam program and an initial
     network topology, these algorithms construct a network
     topology which fits best the structure of the program, and
     allocate the program on the network so that its completion
     time is minimized."
  title = "{T}owards a distributed implementation of occam",
  author= "Debbage, Mark and Hill, Mark and Nicole, Denis A.",
  editor= "Zedan, Hussein S. M.",
  pages = "158--167",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "Progress has been made in providing a reasonable distributed
     implementation of the occam language. Primarily, this has
     involved the development of a routing kernel with a latent
     channel connection retaining occam syntax. Channel semantics
     are maintained by a message acknowledgement scheme and
     unrestricted message lengths. This provides the user with
     the potential for fully connected process communication
     without restrictions on node valencies or explicit PLACEment
     of any hard links.In addition the user program has been
     severed from any dependency on the topology by allowing
     multiple configuration level PROCESSORS to map onto a single
     transputer. Thus the user code can be run on any network
     which has been configured for the virtual channel
     router.Further development of the system has allowed us to
     implement the dynamic primitives that will be required by a
     compiler for distributed full occam. These include dynamic
     channel creation, remote procedure calls and facilities for
     moving channel ends."
  title = "{C}ayley graphs and transputer network configuration",
  author= "East, Ian R. and Jassim, Sabah",
  editor= "Zedan, Hussein S. M.",
  pages = "168--174",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The nature and use of Cayley graphs in understanding network
     topology design will be explained. The hypercube topology
     will be presented in group-theoretic form as an illustration
     and its isomorphism with tori, up to order four, will be
     shown. It will also be shown how to use the Cayley graph
     formulation to scalably configure a hypercube (with node
     process independent of identity within network). Lastly, we
     discuss the application of the Cayley formalism to infer and
     investigate new topologies which exhibit superior scaling of
     size and density to that of the hypercube, but which retain
     degree four and hence are suitable for transputer networks."
  title = "{C}ooperative priority scheduling in occam",
  author= "Sunter, Johan P. E. and Wijbrans, K. C. J. and Bakkers, Andr\`{e} W. P.",
  editor= "Zedan, Hussein S. M.",
  pages = "175--185",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "In this paper a scheduler for variable priority scheduling
     is presented. This scheduler assumes that the processes
     being scheduled cooperate with the scheduler. This
     cooperation introduces some latency in the scheduling of the
     processes. Analytic expressions describing the effect of
     this latency are derived. A variable priority scheduler was
     implemented and results from actual program executions are
     given. These results show that the scheduler can be used to
     schedule control algorithms with simple sequential processes
     with sample frequencies not higher than 2 kHz."
  title = "{I}mplementation of real-time scheduling algorithms in a transputer environment",
  author= "Caprani, Ole and Kristensen, Jens E. and M\ork, Claus and Pedersen, Henrik Bo and Rasmussen, Finn R.",
  editor= "Zedan, Hussein S. M.",
  pages = "186--197",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "A variety of real-time scheduling algorithms is considered.
     Each algorithm is implemented on the transputer either in
     pure Occam or in Occam supplied with a few low level
     assembly procedures to manipulate the workspace queues. The
     algorithms used are non-preemptive non-priority scheduling,
     time-sliced non-priority scheduling, preemptive fixed
     priority scheduling (rate-monotonic scheduling) and
     preemptive dynamic priority scheduling (deadline
     scheduling). Furthermore, it is shown how access to shared
     data can be scheduled to meet time constraints. All the
     implemented scheduling algorithms have been assessed through
     experiments in order to estimate the overhead introduced."
  title = "{M}ulti-priority scheduling for transputer-based real-time control",
  author= "Welch, Peter H.",
  editor= "Zedan, Hussein S. M.",
  pages = "198--214",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "A major requirement of real-time control applications is a
     set of cyclic processes — one for each
     \"control-law\". Each process must be managed so
     that it completes each cycle within a fixed time. The rate
     at which each process cycles will be constant, but will
     generally be different for different processes.Current
     transputer hardware provides very fast pre-emptive
     scheduling for two static priority levels, with
     \"round-robin\" management within each level. This
     is not sufficient to manage securely more than one such
     control-law per transputer — even at very low processor
     loadings. Efficient classical solutions (e.g.
     \"rate-monotonic\" or \"deadline\"
     scheduling) require multiple and time-varying
     priorities.This paper shows how to implement such solutions
     simply, and with an acceptable level of overhead, on the
     existing (and future) generation of transputer. Other
     promising scheduling methods are discussed. All solutions
     are expressed in occam with no assembler inserts and no
     security rules violated — the intended applications are
     safety-critical! Real performance figures are
     reported.Finally, a method of proving (or dis-proving) the
     security of any particular set of process loadings,
     operating under any particular scheduling algorithm, is
  title = "{T}ransputers and routers: {C}omponents for concurrent machines",
  author= "May, David and Thompson, Peter",
  editor= "Zedan, Hussein S. M.",
  pages = "215--231",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "A transputer is a complete microcomputer integrated in a
     single VLSI chip. Each transputer has a number of
     communication links, allowing transputers to be
     interconnected to form concurrent processing systems. The
     transputer instruction set contains Instructions to send and
     receive messages through these links, minimising delays in
     inter-transputer communication. Transputers can be directly
     interconnected to form specialised networks, or can be
     interconnected via routing chips. Routing chips are VLSI
     building blocks for interconnection networks: they can
     support system-wide message routing at high throughput and
     low delay."
  title = "{P}redictable response times and portable hard real-time systems with {TRANS}-{RTX}c on the {T}ransputer",
  author= "Verhulst, Eric and Thielemans, Hans",
  editor= "Zedan, Hussein S. M.",
  pages = "232--240",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The transputer is a flexible processor, very well suited for
     applications where modularity and distributed operation are
     prime requirements, like in process control. Nevertheless,
     its basic FIFO-scheduling algorithm, makes its application
     for real-time processing quite difficult. This problem has
     now been solved by the development of a preemptive
     scheduling algorithm. This algorithm was used to port an
     existing real-time kernel to the Transputer. In addition, by
     taking account of the specific nature of the Transputer,
     much better performance and flexibility have been obtained."
  title = "{P}rototyping transputer applications",
  author= "Hart, E. and Flavell, S.",
  editor= "Zedan, Hussein S. M.",
  pages = "241--247",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The design is discussed of a toolset capable of estimating
     the performance of a transputer application at the design
     stage, prior to detailed code being available. The
     Transim/Gecko package from the Polytechnic of Central London
     is discussed in detail as an example of such a toolset. The
     methodology of the package is discussed, its input language,
     output format and the transputer scheduling model to which
     it adheres. An example is described of a real transputer
     application where the tool has been successfully used to
     improve performance."
  title = "{D}iffusion limited aggregation: {A}n example of real-time parallelisation",
  author= "Morse, D. R. and Welch, A. M. and Welch, Peter H.",
  editor= "Zedan, Hussein S. M.",
  pages = "248--261",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The simulation of the growth of Diffusion-Limited Aggregates
     (DLA) is representative of a class of 'shared'
     data-structure computations that does not yield to
     traditional parallelisation methods (such as 'farming',
     'geometric decomposition' and 'data-flow'). The difficulty
     is that the shared data-structure is large and evolving, the
     required access to it from each processor is random and very
     high, and the computation per access is very low. These
     conditions also make these problems most unsuitable for
     shared-memory parallel computers.This paper presents a
     parallelisation technique that does give linear speed-up for
     this problem (at least, for up to 32 transputers). The
     cost-effectiveness of the solution compares favourably with
     those published that use vector-processing machines.The
     success of the parallelisation depends on real-time issues
     associated with keeping each worker transputer sufficiently
     up-to-date with all its colleagues. Some
     'quasi-relativistic' effects need to be taken into account
     as well!The speed-ups achieved through this parallelisation
     are used to investigate the effect of various parameters
     (such as stride length and background drift) on the kind of
     DLA growth that is obtained. These studies would not be
     practical without the savings in time that have been
     realised from a parallel implementation of the DLA
     simulation.Finally, we characterise the features of those
     applications for which this parallelisation method is
  title = "{A} formal top-down developement method for occam programs",
  author= "Roantree, Donal and Clint, Maurice",
  editor= "Zedan, Hussein S. M.",
  pages = "262--286",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "This paper describes a formal development method for
     mathematical applications in occam. Refinement rules form
     the basis of the method. A traces model is defined and used
     to give a formal semantics to occam. The rules are proved
     sound with respect to this semantics."
  title = "{A}n assessment of the use of occam for dependable real-time systems",
  author= "Burns, A. and Wellings, A. J. and Zedan, Hussein S. M.",
  editor= "Zedan, Hussein S. M.",
  pages = "287--294",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "As occam based technology matures, and more transputer
     implementations evolve, new application domains will open
     up. This in turn will place fresh requirements on the
     language. In some instances occam, in its current form, will
     be unsuitable and language changes will be inevitable. This
     is, of course, at odds with producing a
     \"standard\" and \"stable\" language
     definition.The purpose of this paper is to identify the
     outstanding issues that must be discussed in the short,
     medium or long-term, if occam (and the transputer) is to be
     used for a wide range of dependable real-time applications
     in current and future systems. Our approach is to start with
     application requirements, from these to indicate occam
     problem areas and then, if appropriate, offer potential
     solutions. The areas considered are those associated with
     hard real-time, software reliability, mode changes, dynamic
     change management and fault management."
  title = "{M}emory access synchronization in series expansion methods of parallel image reconstruction",
  author= "Nowinski, W. J.",
  editor= "Zedan, Hussein S. M.",
  pages = "295--304",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The paper is concerned with parallel image reconstruction.
     The background of parallel image reconstruction has been
     briefly reviewed. Parallel formulations of series expansion
     methods expressed in occam exploiting projection and ray
     parallelisms have been presented. Two types of mapping of
     the projection space onto the image space influencing a
     simultaneous memory access have been defined. Finally,
     several ways of synchronizing the activities of processes
     when accessing the memory have been discussed."
  title = "{D}istributing matrix eigenvalue calculations over transputer arrays",
  author= "Hopkins, Tim and Vowden, Barry",
  editor= "Zedan, Hussein S. M.",
  pages = "305--312",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "We discuss the parallel numerical solution of the matrix
     eigenvalue problem for real symmetric tridiagonal matrices.
     Instances occur frequently in practice. Two implementations
     of the Sturm sequence algorithm on transputer arrays are
     described. For the first the maximum size of matrices which
     may be accommodated is restricted by the amount of local
     memory available. The second implementation removes this
     constraint but requires an increased execution time."
  title = "{P}arallel panel methods",
  author= "Chalmers, Alan G. and Fiddes, Steven P. and Paddon, Derek J.",
  editor= "Zedan, Hussein S. M.",
  pages = "313--321",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The nature of panel methods makes parallelisation difficult,
     unless effective methods can be found that minimise the
     number of messages required to calculate the interaction of
     all panels in the problem domain. Here, we use a minimal
     path configuration of processors to give an effective
     solution and show its performance superiority over a
     solution obtained from a ring configuration of processors. A
     detailed description of the numerical model and the
     numerical methods that are used for a typical panel method
     problem is given.The importance of balancing the message
     generating parts of an algorithm are established by
     examining the influence matrix set-up time and the matrix
     solution time. The scalability and maximum performance
     characteristics of the algorithm and system configuration
     are analysed and reported."
  title = "{S}hared virtual memory on transputers via the data diffusion machine",
  author= "Raina, Sanjay and Warren, David H. D. and Cownie, James",
  editor= "Zedan, Hussein S. M.",
  pages = "322--330",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "The Data Diffusion Machine (DDM) is a novel multiprocessor
     architecture which is scalable to an arbitrary number of
     processors and at the same time provides a shared virtual
     address space. There is no fixed home location for data -
     instead data migrates from one processor to another on
     demand. A cache coherence protocol maintains memory
     consistency allowing replication, migration and replacement
     of data.In order to evaluate the DDM we are developing an
     emulator on the Meiko Computing Surface. This paper
     describes the DDM emulator together with additional support
     to turn the emulator into a platform for running real shared
     memory applications. We describe how the bus based snoopy
     protocol of the DDM can be modified to suit point-to-point
     interconnection networks."
  title = "{A}m interactive graphical debugger for occam programs",
  author= "Abdennadher, N. and Angue, J. C.",
  editor= "Zedan, Hussein S. M.",
  pages = "331--338",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "Parallel programs are usually described informally, and
     these descriptions are implemented on parallel computer
     systems. When the program does not work correctly, it is
     difficult to detect the semantic error: deadlock,
     starvation, etc ... We propose in this paper an interactive
     graphical debugger of parallel programs written in OCCAM and
     executed and developed on Transputer Network which is
     implemented on IBM PC motherboard. The debugger provides the
     programmer a graphical representation of the dynamic
     behaviour of the OCCAM programs."
  title = "{V}irtualising communication in the {C}-{NET} high level programming environment",
  author= "Adamo, Jean\_Marc and Bonneville, J. and Bonello, C.",
  editor= "Zedan, Hussein S. M.",
  pages = "339--349",
  booktitle= "{OUG}-13: {R}eal-{T}ime {S}ystems with {T}ransputers",
  isbn= "90 5199 041 3",
  year= "1990",
  month= "sep",
  abstract= "A tool for virtualising communication on the SuperNode
     machine is presented. The tool provides four facilities:
     virtual interprocessor communication channels, Transputer
     links multiplexing, link load-balancing, and consistent
     termination of aborted interprocessor communication."

If you have any comments on this database, including inaccuracies, requests to remove or add information, or suggestions for improvement, the WoTUG web team are happy to hear of them. We will do our best to resolve problems to everyone's satisfaction.

Copyright for the papers presented in this database normally resides with the authors; please contact them directly for more information. Addresses are normally presented in the full paper.

Pages © WoTUG, or the indicated author. All Rights Reserved.
Comments on these web pages should be addressed to: www at wotug.org

Valid HTML 4.01!