Falcon is a system for on-line monitoring and steering of large-scale parallel programs. The purpose of such interactive steering is to improve its performance or to affect its execution behavior. The Falcon system is composed of an application-specific on-line monitoring system, an interactive steering mechanism, and a graphical display system. In this paper, we present a framework of the Falcon system, its implementation, and evaluation of the system performance. A complex sample application -- a molecular dynamics simulation program (MD) -- is used to motivate the research as well as to evaluate the performance of the Falcon system. 1 Introduction The high performance of current parallel supercomputers is permitting users to interact with their applications during program execution. Such interactive executions of large-scale parallel codes typically make use of multiple networked machines working in concert on behalf of a single user, as computational engines, display engines, inpu...

Complex models may have model components distributed over a network and generally require significant execution times. The field of parallel and distributed simulation has grown over the past fifteen years to accommodate the need of simulating the complex models using a distributed versus sequential method. In particular, asynchronous parallel discrete event simulation (PDES) has been widely studied, and yet we envision greater acceptance of this methodology as more readers are exposed to PDES introductions that carefully integrate real-world applications. With this in mind, we present two key methodologies (con- servative and optimistic) which have been adopted as solutions to PDES systems. We discuss PDES terminology and methodology under the umbrella of the personal communications services application.

The increasing complexity of large-scale distributed real time systems demands powerful real time object computing technologies. Furthermore, systematic design approaches are needed to support analysis, design, test, and implementation of these systems. In this paper, we discuss RTDEVS/CORBA, an implementation of DEVS modeling and simulation theory based on real time CORBA communication middleware. With RTDEVS/CORBA, the software model of a complex distributed real time system can be designed and then tested in a virtual testing environment and finally executed in a real distributed environment. This model continuity and simulation-based design approach effectively manages software complexity and consistency problems for complex systems and increases the flexibility for test configurations as well as reduces the time and cost for testing. In the paper, the layered architecture and different implementation issues of RTDEVS/CORBA are studied and discussed. An example application is then given to show how RTDEVS/CORBA supports a framework for model continuity in distributed real time modeling and simulation.

This paper describes current research in real time operating systems. Due to its importance to real-time systems, we begin this survey with a brief summary of relevant results in realtime scheduling and synchronization. Real-time operating systems are described in terms of the primitives and constructs offered to application programs. In addition, the effects of underlying computer architectures on real-time operating systems are discussed, followed by a description of benchmarks and evaluation methods for real-time systems. College of Computing Georgia Institute of Technology Atlanta, Georgia 30332--0280 Readers' comments and suggestions for improvement are solicited. Please direct them to kaushik@cc.gatech.edu. 1 Functionality and Characteristics of Real-time Systems The embedded computer hardware of modern robots and industrial control systems is becoming increasingly complex. Typically, it consists of many interconnected computers operating at multiple levels of control or sup...

The implementation of operating system functions can significantly affect the performance of parallel programs. Our research concerns the customization of operating system functionality for different target hardware to improve the performance of application programs. In this paper, we describe our experience with a reconfigurable, multiprocessor Mach cthreads package on a 32-node KSR-1 supercomputer. Sample static and dynamic configurations address the exchange and on-line adaptation of threads schedulers, and the on-line adaptation of threads synchronization constructs. Experimental results are demonstrated with two different parallel application programs, (1) a parallel branch-and-bound application and (2) the runtime kernel of a Time Warp discrete event simulator. The lightweight threads package has been ported to several target architectures, including Sparcstations, a 32-node GP1000 BBN Butterfly, SGI multiprocessors, and the 32-node Kendall Square Supercomputer. College of Comp...

this memory management to multi-granular distributed computing environments. Issues that must be addressed include the lack of a global memory pool and high communication latencies. New, efficient memory management protocols for distributed memory systems will be developed in this project. Complex simulators may consist of separately developed continuous and discrete simulation programs that are combined into a single, integrated system. For example, one could envision adding continuous models for signal propagation to our discrete PCS network simulation. A key issue in these systems is to develop techniques to effectively integrate different synchronization mechanisms, particularly rollback-based mechanisms such as Time Warp with time-stepped simulation mechanisms. Open research questions regarding these federated simulators include: What is the best approach for integrating these protocols to maximize performance? Should one or more protocols be modified to accommodate interaction with other protocols? If so, how? What is the nature of the interaction of disparate protocols and how does one impact the performance of another?

) Kaushik Ghosh, Richard M. Fujimoto, and Karsten Schwan College of Computing Georgia Institute of Technology Atlanta, GA, 30332. June 24, 1994 Abstract This paper describes several of our experiences with a real-time scheduler. Using a robot control application program, we motivate the importance of supporting multiple schedulers within the same application program. We demonstrate the utility of speculative task execution in dynamic real-time systems, and describe the implementation of a scheduler for performing speculative execution and recovery. We show that existing real-time scheduler interfaces have scope for improvement, especially when scheduling latency must be low and when multiple schedulers used by a single application must co-exist on a single processor. A new scheduler interface is specified and its basic costs are evaluated experimentally. Preliminary measurements on a KSR-1 machine are quoted. The measurements demonstrate how the execution times of temporal queries ma...

Garbage collection can be carried out on-demand in a non-real-time system. However, a real-time system can afford this overhead only during intervals of `idle' time. We motivate the usefulness of reconfiguring the available memory for data structures, and the intervals of garbage collection of these data structures, in a parallel real-time system performing speculative execution. After briefly mentioning the data structures, we describe a scheme for reconfiguring garbage collection. The parameters of such reconfiguration are based on the available platform, and the amount of idle time available in the real-time system. Specific parameters are provided for one architecture -- the KSR2 parallel processor. Experimental performance evaluation of the scheme is currently under investigation. College of Computing Georgia Institute of Technology Atlanta, Georgia 30332--0280 1 Introduction The complexity and diversity of modern real-time applications is moving real-time systems res...

Distributed real-time simulation, distributed real-time databases, fault tolerance, eventual consistency, active databases, time warp ABSTRACT: In this paper we present an optimistic synchronization protocol for distributed real-time simulations that uses a database as communication and storage mechanism. Each node in the simulation is also a database node and communication in the simulation is done by storing and reading to the database. The underlying replication protocol in the database then makes sure that all updates are propagated. The progress in the simulation is optimistic, i.e., each node tries to simulate as far ahead as possible without waiting for input from any other node. Since the simulations are said to be real-time we must guarantee that no events can be delivered too early nor too late. Also, recovery of a node must be done within predictable time due to the real-time constraints. Since all updates in the simulation are done through transactions we have a well-defined foundation for recovery and we show how the recovery can be done deterministically. For the simulation to function (and keep deadlines) during network partitions we allow local commits in the database. This requires that all data required on a specific node must be reachable from that node, i.e., no remote accesses should be needed. However, allowing local commits may introduce conflicting updates. These conflicts are detected and solved predictably. 1.

ion and Its Implementation : : : : : : : : : : : : : : : : : : : : 96 5.2.1 Abstraction : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 96 5.2.2 Implementation : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 97 5.2.2.1 Reads : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 98 5.2.2.2 Writes : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 98 5.2.2.3 Postprocessing of Events : : : : : : : : : : : : : : : : : : : 99 5.2.2.4 Rollbacks : : : : : : : : : : : : : : : : : : : : : : : : : : : : 100 5.2.3 Other comments on the protocol : : : : : : : : : : : : : : : : : : : : 100 5.3 Correctness of protocol : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 101 5.3.1 The pseudo-code : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 101 vii 5.3.2 Correctness : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 104 5.3.2.1 The protocol is deadlock free : : : : : : : : : : : : : : : : 104 5.3.2.2 A committed event gets ...