The parallel (and distributed) Discrete Event simulation (PADS) community - over it's more than 20 years of existence has achieved remarkable results in getting the execution of simulations of discrete event syste...
详细信息
ISBN:
(纸本)0769521118
The parallel (and distributed) Discrete Event simulation (PADS) community - over it's more than 20 years of existence has achieved remarkable results in getting the execution of simulations of discrete event systems faster, involving multiple, logically and/or physically dispersed, communicating computational resources. A solid, mature and empirically approved body of foundational concepts, models, systems, methods and algorithms are at hand and well documented in more than 50.000 pages of research papers. PADS simulation engines, model and software libraries, and operating systems have proven feasibility and applicability of these results in solving real world problems. Despite the (healthy) self-criticism of it's proponents ("Will the field survive?"), PADS was - and is - a success! With this talk we aim at a prospect for the challenges posed by the emerging new models of computational and communication, the evolving new paradigms of interaction and coordination, the new styles of system design, and even the availability of new technologies, for which - in my opinion - no other community qualifies better than the PADS community.
We propose a new simulation method, called interval branching, which aims to facilitate efficient simulation studies of systems that involve temporal uncertainty. The method uses interval timestamps for events and exp...
详细信息
ISBN:
(纸本)9780769531595
We propose a new simulation method, called interval branching, which aims to facilitate efficient simulation studies of systems that involve temporal uncertainty. The method uses interval timestamps for events and explores different execution orders of overlapping events by branching the simulation. We first present a sequential version of interval branching, and then extend it to parallelsimulation using the logical process paradigm. The parallel interval branching algorithm uses the logical-process cloning technique for efficient computation of branches. Our preliminary experiments show its potential as an efficient method for discrete-event simulation studies.
Several scheduling algorithms have been proposed to determine the ne,rr event to be executed on a processor in a Time Warp parallel discrete event simulation. However none of them is specifically designed for simulati...
详细信息
ISBN:
(纸本)0769506771
Several scheduling algorithms have been proposed to determine the ne,rr event to be executed on a processor in a Time Warp parallel discrete event simulation. However none of them is specifically designed for simulations where the execution time (or granularity)for different ty,pes of events has large variance. In this paper we present a grain sensitive scheduling algorithm which addresses this problem. In our solution, the scheduling decision depends on both timestamp and granularity values with the aim at giving higher priority, to small grain events even if their timestamp is not the lowest one (i.e. the closest one to the commitment horizon of the simulation). This implicitly limits the optimism of the execution of large grain events that, if rolled back, would produce large waste of CPU time. The algorithm is adaptive in that it relies on the dynamic recalculation of the length of a simulated time window within which the timestamp of ally good candidate event for the scheduling falls in. if the window length is set to Zeta, then the algorithm behaves like the standard Lowest-Timestamp-First (LTF) scheduling algorithm. simulation results of a classical benchmark in several different configurations are reported for a performance comparison with LTE These results demonstrate the effectiveness of our algorithm.
One of the difficulties inherent to conservative parallel discrete event simulation (PDES) is the need for extraction and explicit specification of the lookahead. In this paper we propose a simulator design which allo...
ISBN:
(纸本)0769516084
One of the difficulties inherent to conservative parallel discrete event simulation (PDES) is the need for extraction and explicit specification of the lookahead. In this paper we propose a simulator design which allows users to build simulation models with lookahead transparently, This is based on the cooperation between PDES experts and the users, and on the reusability, of model components. We introduce the principal design of the simulator and give the results related to its lookahead.
We describe MOSS, a small language of mobile distributed objects and system-wide references, uncommitted to any distributedsimulation protocol, but which can be executed as a distributed conservative simulation with ...
详细信息
We describe MOSS, a small language of mobile distributed objects and system-wide references, uncommitted to any distributedsimulation protocol, but which can be executed as a distributed conservative simulation with automatic deduction of lookahead. We show how the MOSS programmer can control the dynamic distribution and locality of simulation objects by simple means which provide natural modelling functions. Preliminary results show how programmed locality can reduce communication costs in simulation.
A new partitioning method for synchronous PDES simulations is proposed. The method exploits characteristics of both the simulation method and of the application domain to arrive at efficient partitionings. A performan...
详细信息
A new partitioning method for synchronous PDES simulations is proposed. The method exploits characteristics of both the simulation method and of the application domain to arrive at efficient partitionings. A performance study shows that the method outperforms existing partitioning methods in terms of four different performance metrics.
In this work we discuss the use and possible abuse of linear and inversive pseudorandom numbers (PRNs) in parallel and distributed environments. After an investigation of properties of PRNs which determine hour these ...
详细信息
ISBN:
(纸本)0818684577
In this work we discuss the use and possible abuse of linear and inversive pseudorandom numbers (PRNs) in parallel and distributed environments. After an investigation of properties of PRNs which determine hour these may be applied in such environments we introduce a software package which provides an unified and easy-to-use approach to the generating and handling of parallel streams of such PRNs. Experimental results are conducted which describe the features of the software package and compare the performance of two selected types of pseudorandom number *** this work we discuss the use and possible abuse of linear and inversive pseudorandom numbers (PRNs) in parallel and distributed environments. After an investigation of properties of PRNs which determine hour these may be applied in such environments we introduce a software package which provides an unified and easy-to-use approach to the generating and handling of parallel streams of such PRNs. Experimental results are conducted which describe the features of the software package and compare the performance of two selected types of pseudorandom number generators.
A Motif based graphical tool XTracker is described. XTracker can show Gannt-like charts of the activities on each node or it can show the event messages as traffic between simulation objects. XTracker can take its dat...
详细信息
A Motif based graphical tool XTracker is described. XTracker can show Gannt-like charts of the activities on each node or it can show the event messages as traffic between simulation objects. XTracker can take its data from sequential simulation runs and simulate a parallel execution under a number of simulation methods. XTracker can act as a performance modeling tool.
parallel and distributed discrete-event simulation (PDES) is a critical technology for an important class of very large complicated simulation models. However, with few exceptions, the bulk of empirical work in PDES h...
详细信息
parallel and distributed discrete-event simulation (PDES) is a critical technology for an important class of very large complicated simulation models. However, with few exceptions, the bulk of empirical work in PDES has been on small models. Furthermore, synchronization behavior is frequently complicated, which makes it very difficult to analytically prove anything about the performance executing large models on large machines. Scalability analysis asks how the performance of a certain application behaves as the application problem size increases and the parallel architecture executing it increases.
暂无评论