Aumann, Y., Rabin, M.O.: Clock construction in fully asynchronous parallel systems and PRAM simulation. Theoretical Computer Science, Vol. 128 (1994) 3--30 (Preliminary version: FOCS'92)  Home/Search   Document Not in Database   Summary   Related Articles   Check

....could o er programmers illusion that their programs run on a parallel system that is synchronous, while in fact the programs would be simulated on an asynchronous system. Simulations of a parallel system that is synchronous on a system that is asynchronous have been studied for over a decade now [3, 4, 5, 6, 10, 13, 14, 17, 19, 20, 21, 22, 28, 30, 35, 36]. Simplifying considerably, simulations assume that there is a system with p asynchronous processors, and the system is to simulate a program written for n synchronous processors. The simulations use three main ideas: idempotence, load balancing, and synchronization. Speci cally, the execution of ....

....stage, so as to ensure that the simulated program proceeds in lock step. One challenge in realizing the simulations is the problem of late writers i.e. when a slow processor clobbers the memory of a simulation with a value from an old phase. This problem has been addressed in various ways [4, 5, 20, 28, 29]. Another challenge is the development of ecient load balancing and synchronization algorithms. This challenge is abstracted as the Certi ed WriteAll (CWA) problem. In this problem, introduced in a slightly di erent form by Kanellakis and Shvartsman [17] there are p processors, an array w with n ....

[Article contains additional citation context not shown here]

Aumann, Y., Rabin, M.O.: Clock construction in fully asynchronous parallel systems and PRAM simulation. Theoretical Computer Science, Vol. 128 (1994) 3-30 (Preliminary version: FOCS'92)

....the goal of improving productivity of parallel computing involves using the synchronous PRAM model as the programming paradigm and then efficiently simulating PRAM programs on realistic machines. Simulations of PRAM on asynchronous parallel machines have been studied for over a decade now [2, 3, 4, 5, 9, 13, 14, 18, 20, 21, 22, 23, 27, 29, 34, 35]. Simplifying considerably, simulations assume that there is a machine with p asynchronous fault prone processors and the machine is to simulate a program written for n synchronous fault free processors. The simulations use three main ideas: idempotence, load balancing, and synchronization. ....

....every stage so as to ensure that the simulated program proceeds in lock step. One challenge in realizing the simulations is the problem of late writers i.e. when a slow processor clobbers the memory of a simulation with a value from an old phase. This problem has been addressed in various ways [3, 4, 21, 27, 28]. Another challenge is the development of efficient load balancing and synchronization algorithms. In this problem, introduced in a slightly different form by Kanellakis and Shvartsman [18] we have an array w with n cells and a flag f , all initially 0, and wish to set al..l elements of w to 1 and ....

[Article contains additional citation context not shown here]

Aumann, Y., Rabin, M.O.: Clock construction in fully asynchronous parallel systems and PRAM simulation. Theoretical Computer Science, Vol. 128 (1994) 3--30 (Preliminary version: FOCS'92)

....where the computing medium is subject to failures is a fundamental problem in distributed computing. Variations on this problem have been studied in in message passing models [3, 5, 7] using group communications [6, 9] and in shared memory computing using deterministic [12] and randomized [2, 13, 16] models. This work was supported by NSF Grant CCR 9988304 and a grant from AFOSR. The work of the third author was supported by a NSF Career Award. We consider the abstract problem of performing t tasks in a distributed environment consisting of n processors. We refer to this as the do all ....

Aumann, Y., Rabin, M.O.: Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation. Foundations of Comp. Sc. (1993) 147--156

....of the semantics of the data items, and furthermore take into account the state of each transaction while it is executing. Another setting in which the use of replicated variables to give probably correct results has proved to be useful is the ecient simulation of a PRAM on an asynchronous system [24, 3]. Speci cally, Kedem et al. 24] use replicated variables in a way that a correct copy can be reliably identi ed and probably exists. They then use these variables to create a global counter that processors use to determine whether they are roughly synchronized with other processors, and behave ....

....variables in a way that a correct copy can be reliably identi ed and probably exists. They then use these variables to create a global counter that processors use to determine whether they are roughly synchronized with other processors, and behave appropriately if they are not. Aumann and Rabin [3] exhibit a clock construction in an asynchronous system with multiple processors that use a shared memory to create an object that correctly behaves as a clock with high probability. They use the clock to ensure that processors stay synchronized throughout the computation. In both cases, the ....

Y. Aumann and M. Rabin. Clock construction in fully asynchronous parallel systems and PRAM simulation. In Proceedings of the 33rd IEEE Symposium on Foundations of Computer Science, pages 147-156, October 1992.

.... automated tool, e.g. a compiler, the latter requires sophisticated solutions because of the diculty of (re)using the auxiliary memory due to late writers (i.e. processors that are slow and that unknowingly write stale values to memory) Examples of solutions addressing these problems include [2, 3, 22]. The Write All problem has been substantially studied for synchronous failure prone processors [20] and a number of ecient randomized algorithms have been developed for the asynchronous setting, e.g. 2, 3, 7, 13, 22, 29, 28, 31] However there is a dearth of deterministic asynchronous ....

.... stale values to memory) Examples of solutions addressing these problems include [2, 3, 22] The Write All problem has been substantially studied for synchronous failure prone processors [20] and a number of ecient randomized algorithms have been developed for the asynchronous setting, e.g. [2, 3, 7, 13, 22, 29, 28, 31]. However there is a dearth of deterministic asynchronous algorithms. We measure the eciency of algorithms in terms of work that accounts for all machine instructions executed by the n processors during the computation (this generalizes the processortime product measure of synchronous parallel ....

Aumann, Y., Rabin, M.O.: Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation. Proc. of the 33rd IEEE Symp. on Foundations of Computer Science (1993) 147-156

.... tool, e.g. a compiler, the latter requires sophisticated solutions because of the diculty of (re)using the auxiliary memory due to late writers (i.e. processors that are slow and that unknowingly write stale values to memory) Examples of randomized solutions addressing these problems include [2, 1, 21]. Another important aspect of algorithm simulations is the use of on optimistic approach, where the computation may proceed for several steps assuming that all tasks assigned to active processors are successfully completed. For example, a series of potentially incorrect tentative steps can be ....

Aumann, Y., Rabin, M.O., \Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation", in Proc. of the 33rd IEEE Symp. on Foundations of Computer Science, pp. 147-156, 1993.

....where the computing medium is subject to failures is a fundamental problem in distributed computing. Variations on this problem have been studied in in message passing models [3, 5, 7] using group communications [6, 9] and in shared memory computing using deterministic [12] and randomized [2, 13, 16] models. This work appeared in the Proceedings of the 14th International Conference on Distributed Computing (DISC 2000) Springer Verlag LNCS Vol. 1914, pp. 119 133, Toledo, Spain, October 2000. This work was supported by NSF Grant CCR 9988304 and a grant from AFOSR. The work of the third ....

Aumann, Y., Rabin, M.O.: Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation. Foundations of Comp. Sc. (1993) 147-156

....in addition to other performance benefits that we shall see later. 2 Previous and Related Work Calypso has its roots in results by us and by our colleagues addressing fault tolerance, parallel program execution on fault prone and asynchronous abstract machines, and distributed systems [30, 31, 11, 26, 12, 24, 14, 21, 23, 22, 4, 25, 3, 15, 13]. The research leading to Calypso started as formal work which developed provable methods for executing parallel computations, initially on abstract machines with crash failing processors, and later on abstract machines with asynchronous processors. An outline of a network of workstations based ....

Y. Aumann and M. Rabin. Clock construction in fully asynchronous parallel systems and PRAM simulation. In Proc. 33rd IEEE Ann. Symp. on Foundations of Computer Science, 1992.

....spot contention and combining in multistage interconnection networks. IEEE Trans. on Computers, C 34(10) 943 948, 1985. Pre92] L. Prechelt. Measurements of MasPar MP 1216A communication operations. Technical report, Institut fur Programmstrukturen und Datenorganisation, Universitat Karlsruhe, Karlsruhe, Germany, November 1992. Ran89] A. G. Ranade. Fluent parallel computation. PhD thesis, Department of Computer Science, Yale University, New Haven, CT, May 1989. Rei93] J. H. Reif, editor. A Synthesis of Parallel Algorithms. Morgan Kaufmann, San Mateo, CA, 1993. Sny86] L. Snyder. Type architecture, shared memory ....

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and PRAM simulation. In Proc. 33rd IEEE Symp. on Foundations of Computer Science, pages 147--156, October 1992.

....been proved most sufficiently that soul is of all things the oldest since it is the first principle of motion. Plat] CHAPTER 3. MODELLING AND SIMULATION 64 [Guse84] Chandy et al. Chan85] Christian [Chri89] Dunigan [Duni91] Fidge [Fidg88] Fidg91] Basten et al. Bast94] Rabin [Rabi94] [Rabi94a] and Mizuno et al. Mizu95] With regard to distributed simulation, two main approaches have been developed to address the causality problem, usually referred to as time driven and event driven respectively; a further classification applies to these two main approaches which characterizes a ....

Rabin, M., "Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation", Lecture in the Sixth International School For Computer Science Researchers, July 1994, Lipari, Sicily.

....processors execute at the same rate. This extra overhead is due to designing for a cost metric that accounts for worst case asynchrony. The advantage of this algorithm arises only in cases of very large delays among the machine processors. Most asynchronous models, e.g. CZ89, CZ90, Nis90, And92, AR92, DHW93, LAB93] account for more general asynchrony among the processors in their cost metrics, and hence algorithms designed using these models suffer from similar overheads in order to more robustly handle cases with very large delays. We should point out though, that the qrqw asynchronous ....

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and PRAM simulation. In Proc. 33rd IEEE Symp. on Foundations of Computer Science, pages 147--156, October 1992.

....of the current generation of parallel machines. Thus, a number of alternative, intermediate models have been proposed and studied in recent years. These abstract models differ in what aspects of parallel machines are exposed. Some focus on dealing with asynchrony in a shared memory context (e.g. [8, 20, 21, 28, 32, 35, 49, 57, 61]) Others focus on accounting for the overheads in accessing the shared memory ( 2, 3, 25, 32, 41, 44, 52, 56] or in sending messages ( 5, 9, 10, 22, 23, 39, 53, 55, 69] Several models are primarily concerned with the memory hierarchy, especially disk I O ( 6, 62, 72] Others focus on ....

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and PRAM simulation. In Proc. 33rd IEEE Symp. on Foundations of Computer Science, pages 147-- 156, October 1992.

No context found.

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and pram simulation. Theoretical Computer Science, 128:3-30, 1994.

....for any c 0 there exists a proper choice of constants such that Pr [E] 1 Gamma n Gammac . The Phase Clock. Many of our algorithmic constructs operate in phases, each requiring of Theta(n) or Theta(n log n) work. Measuring work in the asynchronous setting is nontrivial. Aumann and Rabin [7] present a general Phase Clock structure which provides the necessary tool. We briefly review the main features of the [7] clock structure. At any point in time the Phase Clock has a value which is either an integer, or undefined . The integral values are non decreasing in time. The value of ....

....of our algorithmic constructs operate in phases, each requiring of Theta(n) or Theta(n log n) work. Measuring work in the asynchronous setting is nontrivial. Aumann and Rabin [7] present a general Phase Clock structure which provides the necessary tool. We briefly review the main features of the [7] clock structure. At any point in time the Phase Clock has a value which is either an integer, or undefined . The integral values are non decreasing in time. The value of the clock is initialized to 0. The Clock Update procedure, which takes O(1) operations, allows the processors to participate ....

[Article contains additional citation context not shown here]

Y. Aumann and M.O. Rabin. Clock construction in fully asynchronous parallel systems and pram simulation. Theoretical Computer Science, 128:3--30, 1994.

....powerful. We further describe why it is useful to bridge these elds and then proceed to the main results in this paper. 1. 1 Asynchronous Parallel Computation Executing parallel programs on heterogeneous processors is studied intensely in the area of asynchronous parallel computation [20, 19, 34, 32, 28, 5, 3, 2]. In this eld, the goal is to run a parallel program that is written assuming synchronization barriers, on a collection of asynchronous processors that do not have a synchronization primitive. In some special cases, such as numerical algorithms, the structure of the parallel program may be ....

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and pram simulation. Theoretical Computer Science, 128:3-30, 1994.

No context found.

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and PRAM simulation. In Proceedings of the 33rd Annual Symposium on the Foundations of Computer Science (FOCS), pages 147-156, 1992.

....of the instructions among processors. This measure directly gauges the efficiency with which the protocol uses whatever instruction cycles are provided by the processors. The total work measure is the standard measure used in the context of A PRAM (Asynchronous PRAM) computations (see e.g. [18, 9]) We say that an event E occurs with high probability (w.h.p. if for any c 0 there exists a proper choice of constants such that Pr [E] 1 Gamma n . Related Work. Fischer, Lynch, and Paterson [15] prove the impossibility of deterministic asynchronous consensus in the message passing ....

....0 flip) A central matter is how the processors know when to advance from one phase to the next. i.e. how do the processors know that a sufficient number of cells where filled in the current phase, so that the computation may advance to the next. For this we employ the Phase Clock construction of [9], which we describe next. 7 0 0 0 0 0 1 1 1 1 1 1 1 0 0 Figure 2: One of the butterflies from the weak coin procedure. There is a constant probability of having a 1 in the input level and a constant probability of having no 1 s in the input level. When subsequent levels are filled ....

[Article contains additional citation context not shown here]

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and pram simulation. Theoretical Computer Science, 128:3--30, 1994.

....and those of scheduling theory. We further describe these elds and then proceed to describe the main results in this paper. 1. 1 Asynchronous Parallel Computation Executing parallel programs on heterogeneous processors is studied intensely in the area of asynchronous parallel computation [16, 15, 29, 28, 24, 5, 3, 2]. In this eld, the goal is to run a parallel program written assuming synchronization barriers, on a collection of asynchronous processors that do not have a synchronization primitive. Processors are assumed to be arbitrarily erratic. That is, a processor may initially run so slowly that it is ....

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and pram simulation. Theoretical Computer Science, 128:3-30, 1994.

No context found.

Y. Aumann and M. O. Rabin. Clock construction in fully asynchronous parallel systems and PRAM simulation. In Proceedings of the 33rd Annual Symposium on the Foundations of Computer Science, pages 147-156, 1992.

No context found.

Aumann, Y., Rabin, M.O.: Clock construction in fully asynchronous parallel systems and PRAM simulation. Theoretical Computer Science, Vol. 128 (1994) 3--30 (Preliminary version: FOCS'92)

No context found.

Aumann, Y., Rabin, M.O.: Clock construction in fully asynchronous parallel systems and PRAM simulation. Theoretical Computer Science, Vol. 128 (1994) 3--30 (Preliminary version: FOCS'92)

No context found.

Aumann, Y., Rabin, M.O.: Clock construction in fully asynchronous parallel systems and PRAM simulation. Theoretical Computer Science, Vol. 128 (1994) 3--30 (Preliminary version: FOCS'92)

No context found.

Aumann, Y., Rabin, M.O.: Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation. Proc. of the 33rd IEEE Symp. on Foundations of Computer Science (1993) 147--156

No context found.

Aumann, Y., Rabin, M.O.: Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation. Foundations of Comp. Sc. (1993) 147--156

No context found.

Aumann, Y., Rabin, M.O.: Clock Construction in Fully Asynchronous Parallel Systems and PRAM Simulation. Proc. of the 33rd IEEE Symp. on Foundations of Computer Science (1993) 147-156

First 50 documents

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