K. Mani Chandy and Jayadev Misra. A Foundation of Parallel Program Design. Addison- Wesley, Reading, MA, 1988.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the possible interleavings that may occur in a distributed system and may specify which of those interleavings are to be considered fair to the individual system components. Examples include CSP [4] in which system components communicate by sending messages over synchronous channels, and UNITY [1], in which components communicate by reading and modifying shared variables. The I O automaton model [7, 8] is particularly well suited for modelling distributed algorithms described using message passing. The I O automaton model is a (not necessarily finite) state machine model that provides ....

....shared memory automata, we need the following t. echnical definition. Let II be a set of actions and X a set of variables. We say that II is complete for X iff Vr 6 II, if In some sense, this is the reverse of what is often done to incorporate message passing into a shared memory model. In UNITY [1], for example, shaxed queue variables are declared to model channels and atomic accesses to these shared queues model sending and receiving da. ta aross the channels. 4These triples axe action names, not to be confused with the steps of an a.utomaton. r (fJc,a, fx) is a shared memory ....

K. Mani Chandy and Jayadev Misra. A Foundation of Parallel Program Design. Addison- Wesley, Reading, MA, 1988.

.... to enter the meeting room (i.e. to access a shared resource) Problems concern synchronous group mutual exclusion where a set of processes must synchronize in order to access a resource, or a process must possess all needed resources in order to continue, have been addressed by Chandy and Misra [CM84, CM88]. 1 Throughout the paper, in a forum is used synonymously with in the meeting room. So, to attend leave a forum is synonymously with to enter exit the meeting room. 2 Note further that the Congenial Talking Philosophers problem is more general than the classical n process mutual ....

K. M. Chandy and J. Misra. A Foundation of Parallel Program Design, Chapter 14: Committee Coordination. Addison-Wesley, 1988.

....Our randomized algorithm is presented in Section 3 and analyzed in Section 4. Section 5 concludes. 2 The Committee Coordination Problem The problem of scheduling multiparty interactions in asynchronous systems has been elegantly characterized by Chandy and Misra as one of Committee Coordination [5]: Professors (cf. processes) in a certain university have organized themselves into committees (cf. interactions) and each committee has a fixed membership roster of one or more professors. From time to time, a professor may decide to attend a committee meeting; it starts waiting and ....

K. Chandy and J. Misra. A Foundation of Parallel Program Design. Addison-Wesley, 1988.

....system in which interactions are fixed and multipartied. Disjunctive parallelism is permitted by allowing interactions to serve as guards in an alternative command. The scheduling problem in this setting has been elegantly characterized by Chandy and Misra as one of committee coordination [CM88] each professor in a university serves on one or more committees, the members of which are fixed; furthermore, a committee cannot convene until all its members are present. The problem then is to schedule committee meetings such that no two committees with a common member convene simultaneously. ....

....to a multiparty 4 interaction. The horizontal nondeterminism is exactly what is needed to avoid duplicate elements (see [EFK89] for further details) A number of distributed algorithms have been proposed for various instances of the multiparty interaction scheduling problem (e.g. Ram87, CM88, Bag89, GB89, PK90, Kum90, JS90a, JS90b, JS94] but, to our knowledge, we are the first to present a comprehensive analysis of the complexity of multiparty interaction. The algorithms of [Ram87, CM88, Bag89, GB89, PK90, Kum90] concern fixed interactions in the presence of disjunctive ....

[Article contains additional citation context not shown here]

K.M. Chandy and J. Misra. A Foundation of Parallel Program Design. Addison-Wesley, 1988.

....in which interactions are primitive and multipartied. Disjunctive parallelism is permitted by allowing interactions to serve as guards in an alternative command. The implementation problem in this setting has been elegantly characterized by Chandy and Misra as one of committee coordination [5]: each professor in a university serves on one or more committees, the members of which are fixed; furthermore, a committee cannot convene until all its members are present. The problem then is to schedule committee meetings such that no two committees with a common member convene ....

....design a synchronous language that uses SCCS like interaction, then we may allow conjunctive and disjunctive parallelism to be used in isolation. Related Work A number of distributed algorithms have been proposed for various instances of the multiparty interaction implementation problem (e.g. [26, 5, 2, 11, 24, 19, 15, 16, 17]) but, to our knowledge, we are the first to present a comprehensive analysis of the complexity of multiparty interaction. The algorithms of [26, 5, 2, 11, 24, 19] concern primitive interactions in the presence of disjunctive parallelism, such as those found in Action Systems, IP, LOTOS, PPSAs, ....

[Article contains additional citation context not shown here]

Chandy, K. M and Misra, J. A Foundation of Parallel Program Design, Addison-Wesley (1988).

....are developed starting from the initial design phase and extending through low level coding decisions and performance tuning. The examples show that good performance can be achieved in modular programs. The early stages of the approach use the same refinement methods that are proposed for Unity [CM88] but our goals in refinement are different. Unity is used as a programming language, and the presumption is that Unity compilers can be developed to generate efficient parallel code from an appropriately refined Unity program, but the practicality of the approach has not been demonstrated. Our ....

....implementations. One reason for using transitionaxiom specifications as a design language is that standard techniques exist for reasoning about such specifications. Correctness arguments can be made at various stages in program development using abstraction functions and invariants [Lam83, LT87, CM88] We include examples of correctness arguments to illustrate their integration into the development process. This thesis also contributes a number of parallel algorithms, which can be divided into two different classes: algorithms that use parallelism internally to reduce latency, and algorithms ....

[Article contains additional citation context not shown here]

K. Mani Chandy and Jayadev Misra. A Foundation of Parallel Program Design. Addison-Wesley Publishing Company, 1988.

....the possible interleavings that may occur in a distributed system and may specify which of those interleavings are to be considered fair to the individual system components. Examples include CSP [9] in which system components communicate by sending messages over synchronous channels, and UNITY [4], in which components communicate by reading and modifying shared variables. The I O automaton model [15, 16] is particularly well suited for modelling distributed algorithms described using message passing. The I O automaton model is a (not necessarily finite) state machine model that provides ....

....existing shared action mechanism. We begin by defining a special type of action called a shared memory action that will be used to model accesses to the shared variables. In some sense, this is the reverse of what is often done to incorporate message passing into a shared memory model. In UNITY [4], for example, shared queue variables are declared to model channels and atomic accesses to these shared queues model sending and receiving data across the channels. However, we emphasize that we do not implement shared memory on top of message passing. Rather, both are modelled directly in ....

K. Mani Chandy and Jayadev Misra. A Foundation of Parallel Program Design. Addison-- Wesley, Reading, MA, 1988.

....of the same group to synchronize to access the resource. Problems concern synchronous group mutual exclusion where a set of processes must synchronize in order to access a resource, or a process must possess all needed resources in order to continue, have been addressed by Chandy and Misra [CM84, CM88] In this paper we present a problem, which we refer to as the Congenial Talking Philosophers , to model the design issue for concurrency while mutual exclusion. The problem concerns a set of n philosophers which spend their time thinking alone and talking in a forum. Given that there is only one ....

K. M. Chandy and J. Misra. A Foundation of Parallel Program Design, Chapter 14: Committee Coordination. Addison-Wesley, 1988.

....to developing a unified model that supports composition, superposition, and encapsulation is to start with two existing models, each supporting only some of these features, and then to develop extensions to one of the models to support the features of the other. The two models we study are UNITY [3], which provides a superposition operator, and I O Automata [13] which provides composition and encapsulation. The result of this work is an extended I O automaton model that supports all three of these features and includes a formal specification mechanism for layered systems that allows the set ....

....components in order to prove properties about executions of the entire system. However, the I O automaton model does not provide a mechanism for constructing layered systems in which higher level modules can observe the states of lower level ones. In the UNITY model, defined by Chandy and Misra [3], a program consists of a set of statements that access a global shared memory. At each step in the execution, a statement is selected and executed, possibly updating the memory. Superposition in UNITY is defined to be a program transformation that adds a layer on top of a program, while ....

[Article contains additional citation context not shown here]

K. Mani Chandy and Jayadev Misra. A Foundation of Parallel Program Design. Addison-- Wesley, Reading, MA, 1988.

....tools based on that model, it is easier to integrate the tasks of algorithm specification, design, debugging, analysis, and proof of correctness. A number of formal models have been developed for the study of concurrent systems. Examples include CSP [22] CCS [36] Statecharts [18, 19] UNITY [10], and Swarm [42, 43] Some of these models have accompanying programming languages or simulation tools. For example, the Occam system [39] provides an implementation of a subset of CSP, the Statemate system [17] provides a simulation tool for systems described with Statecharts, and a programming ....

....Boug e and Francez [8] argue in favor of the use of superposition as a language construct by describing a number of important applications for its use, and they offer a compositional approach to superposition with a syntactic representation in CSP. In the UNITY model, defined by Chandy and Misra [10], a program consists of a set of statements that access a global shared memory. At each step in the execution, a statement is selected and executed, possibly updating the memory. Superposition in UNITY is defined to be a program transformation that adds a layer on top of a program, while ....

[Article contains additional citation context not shown here]

K. Mani Chandy and Jayadev Misra. A Foundation of Parallel Program Design. Addison-- Wesley, Reading, MA, 1988.

....Our randomized algorithm is presented in Section 3 and analyzed in Section 4. Section 5 concludes. 2 The Committee Coordination Problem The problem of scheduling multiparty interactions in asynchronous systems has been elegantly characterized by Chandy and Misra as one of Committee Coordination [5]: Professors (cf. processes) in a certain university have organized themselves into committees (cf. interactions) and each committee has a fixed membership roster of one or more professors. From time to time, a professor may decide to attend a committee meeting; it starts waiting and continues to ....

....For example, 4 A scheme of synchrony loosening is therefore proposed in [7] for reducing the size of an interaction in practical applications. 5 These algorithms and Multi all allow professors to distributedly establish a committee meeting on their own. Other deterministic algorithms such as [5, 4, 14] employ managers to coordinate committee meetings; the time complexity of these algorithms then depends on the number of managers they use. 6 Three, if you count the Boolean test. accesses to a counter can be executed concurrently. 7 To see this, consider three possible interleaved ....

K.M. Chandy and J. Misra. A Foundation of Parallel Program Design. Addison-Wesley, 1988.

....: 2 c[ Gamma : 2 c[ Gamma : Here interaction a has a fixed set of participants P , Q, and R, and so P can execute a[ if and when Q and R agree to execute their own a[ actions. The guardscheduling problem for zeroth order interactions has been elegantly characterized by Chandy and Misra [1988] as one of Committee Coordination: each professor in a university serves on one or more committees, the members of which are fixed; furthermore, a committee cannot convene until all its members are present. As such, no two committees with a common member can convene simultaneously. For the ....

....interaction can be concurrently activated. As more than one process can enrole into the same role, the failure of any one of them should not preclude the others from filling the role. Thus, first order interactions are inherently fault tolerant. The distributed algorithms of [Ramesh 1987; Chandy and Misra 1988; Bagrodia 1989a; Kumar 1990; Park and Kim 1990] for multiparty interaction guard scheduling are zeroth order in nature. The algorithms of [Chandy and Misra 1988; Bagrodia 1989a; Park and Kim 1990] employ a fixed coordinator for each interaction and thus cannot realize concurrent activations of a ....

[Article contains additional citation context not shown here]

Chandy, K. and Misra, J. 1988. A Foundation of Parallel Program Design.

....] Gamma : R : a[ Gamma : 2 c[ Gamma : P can execute a[ if and when Q and R agree to execute their own a[ actions. The guard scheduling problem for zeroth order interactions has been elegantly characterized by Chandy and Misra as one of committee coordination [3]: each professor in a university serves on one or more committees, the members of which are fixed; furthermore, a committee cannot convene until all its members are present (as such, no two committees with a common member can convene simultaneously) For the first order case, an enrolement ....

....of a first order interaction can be concurrently activated. ffl As more than one process can enrole into the same role, the failure of any one of them should not preclude the others from filling the role. Thus, first order interactions are inherently fault tolerant. The distributed algorithms of [17, 3, 2, 13, 15] for multiparty interaction guard scheduling are zerothorder in nature. The algorithms of [3, 2, 15] employ a fixed coordinator for each interaction and 2 In IP the situation is somewhat different. IP s team is an abstraction mechanism for multiparty interactions, the zeroth order construct of ....

[Article contains additional citation context not shown here]

Chandy, K. M. and Misra, J. A Foundation of Parallel Program Design, Addison-Wesley, 1988.

....a weakest precondition semantics. cf. for example, Bac80, Mor87, GM91] Parallelism, in the framework of program refinement, is usually treated UNITY style: the (fair) interleaved execution of an action system consisting of a fixed number of processes with communication through shared variables ([CM88, BKS88, Ser90, Luk91]) Some results are established dealing with the superposition of action systems. e.g. BS92] and [BS94] where additionally procedures are incorporated. In general, however, the underlying weakest precondition semantics is not tuned to a compositional treatment of concurrency. For some ....

K.M. Chandy and J. Misra. A Foundation of Parallel Program Design. Addison-Wesley, 1988.

.... using a higher level language as input is presented by Barros et al. 3] The partitioning algorithm is a two stage clustering algorithm which selects groups of code based on similarity measures obtained from classification of assignments in the input specification, which is described in UNITY [9]. Codesign systems in which hardware software partitioning is obtained by user interaction have also been investigated. Among these are POLIS [10] PARTIF [28] and CASTLE [8] In POLIS analysis and transformations are done on a uniform and formal internal hardware software representation called ....

K.M. Chandy and J. Misra. A Foundation of Parallel Programs Design. Printice Hall, 1988.

No context found.

K. Mani Chandy and Jayadev Misra. A Foundation of Parallel Program Design. Addison-- Wesley, Reading, MA, 1988.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC