Virtual time is a new paradigm for organizing and synchronizing distributed systems which can be applied to such problems as distributed discrete event simulation and distributed database concur-rency control. Virtual time provides a flexible abstraction of real time in much the same way that virtual memory provides an abstraction of real memory. It is implemented using the Time Warp mechanism, a synchronization protocol distinguished by its reliance on lookahead-rollback, and by its implementation of rollback via antimessages.

Traditional discrete-event simulations employ an inherently sequential algorithm. In practice, simulations of large systems are limited by this sequentiality, because only a modest number of events can be simulated. Distributed discrete-event simulation (carried out on a network of processors with asynchronous message-communicating capabilities) is

The achievements attained in accelerating the simulation of the dynamics of complex discrete event systems using parallel or distributed multiprocessing environments are comprehensively presented. While parallel discrete event simulation (DES) governs the evolution of the system over simulated time in an iterative SIMD way, distributed DES tries to spatially decompose the event structure underlying the system, and executes event occurrences in spatial subregions by logical processes (LPs) usually assigned to different (physical) processing elements. Synchronization protocols are necessary in this approach to avoid timing inconsistencies and to guarantee the preservation of event causalities across LPs. Included in the survey are discussions on the sources and levels of parallelism, synchronous vs. asynchronous simulation and principles of LP simulation. In the context of conservative LP simulation (Chandy/Misra/Bryant) deadlock avoidance and deadlock detection/recovery strategies, Con...

Discrete event simulation is widely used within the networking community for purposes such as demonstrating the validity of network protocols and architectures. Depending on the level of detail modeled within the simulation, the running time and memory requirements can be excessive. The goal of our research is to develop and demonstrate a practical, scalable approach to parallel and distributed simulation that will enable widespread reuse of sequential network simulation models and software. We focus on an approach to parallelization where an existing network simulator is used to build models of subnetworks that are composed to create simulations of larger networks. Changes to the original simulator are minimized, enabling the parallel simulator to easily track enhancements to the sequential version. In this paper we describe our lessons learned in applying this approach to the publicly available ns [9] software package, and converting it to run in a parallel fashion on a network of wo...

e-j 4r.,,D I-- " h",' _ k,) r,m '3'-. IC,-.-4 Z _ O

Conventional wisdom has it there are two basic approaches to parallel simulation: conservative (Chandy-Misra) and optimistic (time warp). All known protocols are thought to fall into one of these two classes. This dichotomy is false. There exists a spectrum of options that includes these approaches. We describe a design space that admits these as alternatives, we show how most of the well known parallel simulation approaches can be derived using our design alternatives, and we explore the implications of the existence of the design space we describe. In particular, we note there are many as yet unexplored approaches to parallel simulation. 1. INTRODUCTION Parallel simulation, generally called distributed simulation in the literature, is concerned with the parallel execution of discrete event simulations. Beginning with the research of Chandy and Misra [ChMi79] and Peacock et al. [PeWo79], a number of approaches have been described for coordinating cooperating processes so th...

There is reason to believe bounded aggressive processing (limiting the degree to which processes act on conditional knowledge) may be a good alternative to unbounded processing. Simulations characterized by substantial variance in logical process processing times can lead to conditions where, without bound, some processes may cause frequent repairs (e.g. rollbacks) to occur. We present an algorithm, SRADS/LR, in which we bound aggressiveness and discuss its expected impact on performance.

this paper are:

This paper proposes a new coherence method called “multicast snooping ” that dynamically adapts between broadcast snooping and a directory protocol. Multicast snooping is unique because processors predict which caches should snoop each coherence transaction by specifying a multicast “mask. ” Transactions are delivered with an ordered multicast network, such as an Isotach network, which eliminates the need for acknowledgment messages. Processors handle transactions as they would with a snooping protocol, while a simplified directory operates in parallel to check masks and gracefully handle incorrect ones (e.g., previous owner missing). Preliminary performance numbers with mostly SPLASH-2 benchmarks running on 32 processors show that we can limit multicasts to an average of 2-6 destinations (<< 32) and we can deliver 2-5 multicasts per network cycle (>> broadcast snooping’s 1 per cycle). While these results do not include timing, they do provide encouragement that multicast snooping can obtain data directly (like broadcast snooping) but apply to larger systems (like directories). 1

Recently there has been a great deal of interest in performance evaluation of parallel simulation. Most work is devoted to the time complexity and assumes that the amount of memory available for parallel simulation is unlimited. This paper studies the space complexity of parallel simulation. Our goal is to design an efficient memory management protocol which guarantees that the memory consumption of parallel simulation is of the same order as sequential simulation. (Such an algorithm is referred to as optimal.) We first derive the relationships among the space complexities of sequential simulation, Chandy-Misra simulation, and Time Warp simulation. We show that Chandy-Misra may consume more storage than sequential simulation, or vice versa. Then we show that Time Warp never consumes less memory than sequential simulation. Then we describe cancelback, an optimal Time Warp memory management protocol proposed by Jefferson. Although cancelback is considered as a complete solution for the storage management problem in Time Warp, some efficiency issues in implementing this algorithm must be considered. In this paper, we propose an optimal algorithm called artificial rollback. We show that this algorithm is easy to implement and analyze. An implementation of artificial rollback is given, which is integrated with processor scheduling to adjust the memory consumption rate based on the amount of free storage available in the system.