Accurate simulation of large parallel applications can be facilitated withthe use of direct execution and parallel discrete event simulation. this paper describes the use of COMPASS, a direct execution-driven, parall...
详细信息
Accurate simulation of large parallel applications can be facilitated withthe use of direct execution and parallel discrete event simulation. this paper describes the use of COMPASS, a direct execution-driven, parallel simulator for performance prediction of programs that include both communication and I/O intensive applications. the simulator has been used to predict the performance of such applications on bothdistributed memory machines like the IBM SP and shared-memory machines like the SGI Origin 2000. the paper illustrates the usefulness of COMPASS as a versatile performance prediction tool. We use both real-world applications and synthetic benchmarks to study application scalability, sensitivity to communication latency, and the interplay between factors like communication pattern and parallel file system caching on application performance. We also show that the simulator is accurate in its predictions and that it is also efficient in its ability to use parallelsimulation to reduce its own execution time which, in some cases, has yielded a near-linear speedup.
the proceedings contain 18 papers. the topics discussed include: scheduling resources in multi-user, heterogeneous, computing environments with SmartNet;the Globus project: a status report;Netsolve: a network-enabled ...
ISBN:
(纸本)0818683651
the proceedings contain 18 papers. the topics discussed include: scheduling resources in multi-user, heterogeneous, computing environments with SmartNet;the Globus project: a status report;Netsolve: a network-enabled solver;examples and users;implementing distributed synthetic forces simulations in metacomputing environments;CCS resource management in networked HPC systems;a dynamic matching and scheduling algorithm for heterogeneous computing systems;dynamic, competitive scheduling of multiple DAGS in a distributed heterogeneous environment;the relative performance of various mapping algorithms is independent of sizable variances in run-time predictions;modeling the slowdown of data-parallel applications in homogeneous and heterogeneous clusters of workstations;specification and control of cooperative work in a heterogeneous computing environment;a mathematical model, heuristic, and simulation study for a basic data staging problem in a heterogeneous networking environment;modular heterogeneous system development: a critical analysis of java;fault-tolerance: Java's missing buzzword;heterogeneous parallel computing with Java: jabber or justified?;on the interaction between mobile processes and objects;steps toward understanding performance in Java;and heterogeneous programming with Java: gourmet blend or just a hill of beans?.
the proceedings contain 36 papers. the topics discussed include: feature interactions of livelock type in IN: a detailed example;detecting feature interactions at specification stage;implementing an in service 'a ...
ISBN:
(纸本)0780349059
the proceedings contain 36 papers. the topics discussed include: feature interactions of livelock type in IN: a detailed example;detecting feature interactions at specification stage;implementing an in service 'a la TINA';secure codless terminal mobility - a challenge for manufacturers and providers;architectural impact of new services in a mobile network;database usage in telecommunications through CORBA;service animation - integrating call processing and management scripts;IN controls for broadband networks;open intelligent network service management point;TINA based service control in an intelligent broadband network;distributed network information systems for intelligent B-ISDN infrastructure;thin clients and distributed servers for service management;another dimension of flow control for the intelligent node;simulation of load limitation and interface performance in an IN configuration;intelligent networks as a solution for internet related telephone network congestion;don't underestimate the use of formal methods - how to make more out of your existing IN;call party handling aspects evolution from IN CS-2 to IN CS-3;and the evolution of IN service logic.
A distributed, parallel implementation of the widely used Modular Semi-Automated Forces (ModSAF) distributed Interactive simulation (DIS) is presented, with Scalable parallel Processors (SPPs) used to simulate more th...
详细信息
ISBN:
(纸本)0818683651
A distributed, parallel implementation of the widely used Modular Semi-Automated Forces (ModSAF) distributed Interactive simulation (DIS) is presented, with Scalable parallel Processors (SPPs) used to simulate more than 50,000 individual vehicles. the single-SPP code is portable and has been used on a variety of different SPP architectures for simulations with up to 15,000 vehicles. A general metacomputing framework for DIS on multiple SPPs is discussed and results are presented for an initial system using explicit Gateway processes to manage communications among the SPPs. these 50K-vehicle simulations utilized 1,904 processors at six sites across seven time zones, including platforms from three manufacturers. Ongoing activities to both simplify and enhance the metacomputing system using Globus are described.
Data staging is an important data management problem for a distributed heterogeneous networking environment, where each data storage location and intermediate node may have specific data available, storage limitations...
详细信息
ISBN:
(纸本)0818683651
Data staging is an important data management problem for a distributed heterogeneous networking environment, where each data storage location and intermediate node may have specific data available, storage limitations, and communication links. Sites in the network request data items and each item is associated with a specific deadline and priority. It is assumed that not all requests can be satisfied by their deadline. this work concentrates on solving a basic version of the darn staging problem in which all parameter values for the communication system and the data request information represent the best known information collected so far and stay fixed throughout the scheduling process. A mathematical model for the basic data staging problem is introduced. then, a multiple-source shortest-path algorithm based heuristic for finding a suboptimal schedule of the communication steps for data staging is presented. A simulation study is provided, which evaluates the performance of the proposed heuristic. the results show the advantages of the proposed heuristic over two random based scheduling techniques. this research, based on the simplified static model, sen es as a necessary step toward solving the more realistic and complicated version of the data staging problem involving dynamic scheduling, fault tolerance, and determining where to stage data.
the unthrottled optimism underlying the Time Warp (TW) parallelsimulation protocol can lead to excessive aggressiveness in memory consumption due to saving state histories, and waste of CPU cycles due to overoptimist...
详细信息
ISBN:
(纸本)081867931X
the unthrottled optimism underlying the Time Warp (TW) parallelsimulation protocol can lead to excessive aggressiveness in memory consumption due to saving state histories, and waste of CPU cycles due to overoptimistically progressing simulations that eventually have to be ''rolled back''. Furthermore, in TW simulations executing in distributed memory environments, the communication overhead induced by the rollback mechanism can cause pathological overall simulation performance. In this work direct optimism control mechanisms are used to overcome these shortcomings by probabilistically controlling simulation progression based on the forecasted time stamp of forthcoming messages. Several forecast methods are presented and their performance is compared for very large Petri net simulation models executed withthe TW protocol on the Meiko CS-2.
distributed computing systems making available large memories and powerful computational resources, allow the analysis of problems of unprecedented complexity. the exploitation of the potential offered by these system...
详细信息
ISBN:
(纸本)081867931X
distributed computing systems making available large memories and powerful computational resources, allow the analysis of problems of unprecedented complexity. the exploitation of the potential offered by these systems highly depends upon the availability of efficient tools devoted to the solution of specific problems. the paper describes a distributed approach to the analysis of large GSPN models whose applicability to real systems was often constrained by the explosion in the size of the underlying state space representation. A performing algorithm for distributed construction of large state space graphs generated by GSPN models is described in detail the subsequent distributed numerical solution of the Markov chain is then presented and discussed, along with its current limitations. the implementation of the complete distributed solution tool is integrated withthe well known GreatSPN package. the use of the widespread MPI Message Passing library makes the tool portable on a large class of distributed and parallel architectures. Performance results in the case of nets with large state space cardinalities are discussed, considering an experimental low-cost, general purpose PC cluster.
We present an approach to stochastic performance guarantees for multimedia sewers with mixed workloads. Advanced multimedia applications such as digital libraries or teleteaching exhibit a mixed workload with accesses...
详细信息
ISBN:
(纸本)0818678496
We present an approach to stochastic performance guarantees for multimedia sewers with mixed workloads. Advanced multimedia applications such as digital libraries or teleteaching exhibit a mixed workload with accesses to both 'continuous' and conventional, 'discrete' data, where the fractions of continuous-data and discrete-data requests vary over time. We assume that a server shares all disks among continuous and discrete data, and we develop a stochastic performance model for the resulting mixed workload, using a combination of analytic and simulation-based modeling. Based on this model we devise a round-based scheduling scheme with stochastic performance guarantees: for continous-data requests, we bound the probability that 'glitches' occur and for discrete-data requests, we bound the probability that the response time exceeds a certain tolerance threshold. We present early results of simulation studies.
It is well known that Time Warp may suffer from poor performance due to excessive rollbacks caused by overly optimistic execution. Here we present a simple flow control mechanism using only local information and GVT t...
详细信息
It is well known that Time Warp may suffer from poor performance due to excessive rollbacks caused by overly optimistic execution. Here we present a simple flow control mechanism using only local information and GVT that limits the number of uncommitted messages generated by a processor, thus throttling overly optimistic TW execution. the flow control scheme is analogous to traditional networking flow control mechanisms. A `window' of messages defines the maximum number of uncommitted messages allowed to be scheduled by a process. Committing messages is analogous to acknowledgments in networking flow control. the initial size of the window is calculated using a simple analytical model that estimates the instantaneous number of messages that a process will eventually commit. this window is expanded so that the process may progress up to the next commit point (generally the next fossil collection), and to accommodate optimistic execution. the expansions to the window are based on monitoring TW performance statistics so the window size automatically adapts to changing program behaviors. the flow control technique presented here is simple and fully automatic. No global knowledge or synchronization (other than GVT) is required. We also develop an implementation of the flow control scheme for shared memory multiprocessors that uses dynamically sized pools of free message buffers. Experimental data indicates that the adaptive flow control scheme maintains high performance for `balanced workloads', and achieves as much as a factor of 7 speedup over unthrottled TW for certain irregular workloads.
this conference proceedings contains 22 papers on advances in the design and analysis of algorithms for parallel and distributedsimulation. Topics discussed include selecting the checkpoint interval in time warp para...
详细信息
ISBN:
(纸本)1565550552
this conference proceedings contains 22 papers on advances in the design and analysis of algorithms for parallel and distributedsimulation. Topics discussed include selecting the checkpoint interval in time warp parallelsimulation, parallel algorithms for simulating continuous time Markov chains, determining initial states for time-parallelsimulations, global synchronization for optimistic parallel discrete event simulation, an algorithm for minimally latent global virtual time, a parallel partitioning technique for use with conservative parallelsimulation, disseminating critical synchronization information in parallel discrete event simulations, shared variables in distributedsimulation, high performance parallel logic simulation on a network of workstations, corolla partitioning for distributed logic simulation of VLSI circuits, efficient implementation of event sets in time warp, an analytical comparison of periodic checkpointing and incremental state saving, parallelsimulation of communicating finite state machines, the effect of synchronization requirements on the performance of distributedsimulations, and time warp simulation in time-constrained systems.
暂无评论