Results 1 - 10
of
11
Multicast grouping for dynamic data distribution management
- In Proceedings of the 31st Society for Computer Simulation Conference
, 1999
"... multicast, RTI The High Level Architectures Data Distribution Management services are the most recent in a succession of systems designed to reduce the amount of data received by individual simulations in large-scale distributed simulations. A common optimization in these interest management systems ..."
Abstract
-
Cited by 26 (2 self)
- Add to MetaCart
(Show Context)
multicast, RTI The High Level Architectures Data Distribution Management services are the most recent in a succession of systems designed to reduce the amount of data received by individual simulations in large-scale distributed simulations. A common optimization in these interest management systems is the use of multicast groups for sending data to a selected subset of all potential receivers. The use of multicast has met with considerable success in this application. However, its use to date has relied on a priori knowledge of communication patterns between simulations and static assignment of multicast groups to these patterns. As larger, more complex, and less predictable simulations are built, the need has arisen for more efficient use of multicast groups as they are a restricted resource. This paper analyzes the potential benefits of using multicast as a way of motivating algorithms for performing dynamic grouping. An initial grouping algorithm is described and results of simulating its effectiveness are given.
Parallel Execution Of A Sequential Network Simulator
, 2000
"... Parallel discrete event simulation (PDES) techniques have not yet made a substantial impact on the network simulation community because of the need to recast the simulation models using a new set of tools. To address this problem, we present a case study in transparently parallelizing a widely used ..."
Abstract
-
Cited by 11 (1 self)
- Add to MetaCart
Parallel discrete event simulation (PDES) techniques have not yet made a substantial impact on the network simulation community because of the need to recast the simulation models using a new set of tools. To address this problem, we present a case study in transparently parallelizing a widely used network simulator, called ns. The use of this parallel ns does not require the modeler to learn any new tools or complex PDES techniques. The paper describes our approach and design choices to build the parallel ns and presents preliminary performance results, which are very encouraging.
RTI Performance on Shared Memory and Message Passing Architectures
- in Proceedings of the 1999 Spring Simulation Interoperability Workshop
, 1999
"... : This paper compares the performance of HLA time management algorithms across three computing architectures: 1) a shared memory multiprocessor (SGI Origin), 2) a cluster of workstations interconnected via a low latency, high-speed switched network (Myricomm's Myrinet), and 3) a traditional LAN ..."
Abstract
-
Cited by 3 (1 self)
- Add to MetaCart
: This paper compares the performance of HLA time management algorithms across three computing architectures: 1) a shared memory multiprocessor (SGI Origin), 2) a cluster of workstations interconnected via a low latency, high-speed switched network (Myricomm's Myrinet), and 3) a traditional LAN using TCP/IP. This work is based on the RTI-Kit software package described in a paper presented at the Fall 1998 SIW. This software implements group communication and time management functions on the platforms described above. A time management algorithm designed to efficiently compute LBTS values on shared memory multiprocessors is described. This algorithm exploits sequentially consistent shared memory to achieve very efficient time management. We present results comparing the performance of this time management algorithm with a messagepassing algorithm implemented over shared memory. 1. Introduction The U.S. Department of Defense (DoD) mandated that the High Level Architecture (HLA) be the s...
Multicast Grouping for Data Distribution Management. Simulation Practice and Theory
, 2002
"... The High Level Architecture’s Data Distribution Management services are the most recent in a succession of systems designed to reduce the amount of data received by individual simulations in large-scale distributed simulations. A common optimization in these interest management systems is the use of ..."
Abstract
-
Cited by 2 (0 self)
- Add to MetaCart
(Show Context)
The High Level Architecture’s Data Distribution Management services are the most recent in a succession of systems designed to reduce the amount of data received by individual simulations in large-scale distributed simulations. A common optimization in these interest management systems is the use of multicast groups for sending data to a selected subset of all potential receivers. The use of multicast has met with considerable success in this application. However, its use to date has relied on a priori knowledge of communication patterns between simulations and static assignment of multicast groups to these patterns. As larger, more complex, and less predictable simulations are built, the need has arisen for more efficient use of multicast groups as they are a restricted resource1. This paper presents two algorithms for performing grouping, and the message delivery time improvements resulting from applying the algorithms to selected data sets. 1
Topology-Aware Communication in Wide-Area Message-Passing
- in Recent Advances in Parallel Virtual Machine and Message Passing Interface, Lecture Notes in Computer Science
"... Abstract. This position paper examines the use of topology-aware communication services to support message-passing in wide-area, distributed environments, i.e., grids. Grid computing promises great benefits in the flexible sharing of resources but poses equally great challenges for highperformance c ..."
Abstract
-
Cited by 2 (1 self)
- Add to MetaCart
(Show Context)
Abstract. This position paper examines the use of topology-aware communication services to support message-passing in wide-area, distributed environments, i.e., grids. Grid computing promises great benefits in the flexible sharing of resources but poses equally great challenges for highperformance computing, that is to say, how to execute large-scale computations on a grid with reasonable utilization of the machines involved. For wide-area computations using a message-passing paradigm, these issues can be addressed by using topology-aware communication, i.e., communication services that are aware of and can exploit the topology of the network connecting all relevant machines. Such services can include augmented communication semantics (e.g., filtering), collective operations, content-based and policy-based routing, and managing communication scope to manage feasibility. While such services can be implemented and deployed in a variety of ways, we propose the use of a peer-to-peer, middleware forwarding and routing layer. In a related application domain (time management in distributed simulations) we provide emulation results showing that such topology-awareness can play a major role in performance and scalability. Besides these benefits, such communication services raise a host of implementation and integration issues for their operational deployment and use in grid environments. Hence, we discuss the need for proper APIs and high-level models. 1
Migrating to a realtime distributed parallel simulator architecture
- In Proceedings of the 2007 Summer Computer Simulation Conference
, 2007
"... A legacy non-distributed logical time simulator is migrated to a distributed architecture to parallelise execution. The ex-isting Discrete Time System Specification (DTSS) modelling formalism is retained to simplify the reuse of existing mod-els. This decision, however means that the high simulation ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
A legacy non-distributed logical time simulator is migrated to a distributed architecture to parallelise execution. The ex-isting Discrete Time System Specification (DTSS) modelling formalism is retained to simplify the reuse of existing mod-els. This decision, however means that the high simulation frame rate of 100Hz used in the legacy system has to be re-tained in the distributed one—a known difficulty for existing distribution technologies due to inter-process communication latency. A specialised publish-subscribe simulation model is used for the new simulator architecture. The simulation model, in-cluding the process synchronisation, is implemented using a low latency peer-to-peer TCP messaging protocol. The TCP send and receive buffers and TCP’s Nagle algorithm are also tweaked to ensure low latency communication. Gigabit Eth-ernet is used at the hardware layer. A parallelised execution speed-up of four to five times is reached with six to eight ma-chines at a simulation frame rate of 100Hz.
Multicast Grouping For Data Distribution Management
, 2001
"... The High Level Architecture's Data Distribution Management services are the most recent in a succession of systems designed to reduce the amount of data received by individual simulations in large-scale distributed simulations. A common optimization in these interest management systems is the u ..."
Abstract
- Add to MetaCart
The High Level Architecture's Data Distribution Management services are the most recent in a succession of systems designed to reduce the amount of data received by individual simulations in large-scale distributed simulations. A common optimization in these interest management systems is the use of multicast groups for sending data to a selected subset of all potential receivers. The use of multicast has met with considerable success in this application. However, its use to date has relied on a priori knowledge of communication patterns between simulations and static assignment of multicast groups to these patterns. As larger, more complex, and less predictable simulations are built, the need has arisen for more efficient use of multicast groups as they are a restricted resource . This paper presents two algorithms for performing grouping, and the message delivery time improvements resulting from applying the algorithms to selected data sets.
Network Aware Time Management and Event Distribution
- In Proc. of the 14th Workshop on Parallel and Distributed Simulation
, 2000
"... In this paper we discuss new synchronization algorithms for Parallel and Distributed Discrete Event Simulations (PDES) which exploit the capabilities and behavior of the underlying communications network. Previous work in this area has assumed the network to be a Black Box which provides a one-to-on ..."
Abstract
- Add to MetaCart
In this paper we discuss new synchronization algorithms for Parallel and Distributed Discrete Event Simulations (PDES) which exploit the capabilities and behavior of the underlying communications network. Previous work in this area has assumed the network to be a Black Box which provides a one-to-one, reliable and in-order message passing paradigm. In our work, we utilize the Broadcast capability of the ubiquitous Ethernet for synchronization computations, and both unreliable and reliable protocols for message passing, to achieve more efficient communications between the participating systems. We describe two new algorithms for computation of a distributed snapshot of global reduction operations on monotonically increasing values. The algorithms require O(N ) messages (where N is the number of systems participating in the snapshot) in the normal case. We specifically target the use of this algorithm for distributed discrete event simulations to determine a global lower bound on time-s...
Pre-Sampling as an Approach for Exploiting Temporal Uncertainty
, 2000
"... In this paper we describe an approach to exploit temporal uncertainty in parallel and distributed simulation by utilizing time intervals rather than precise time stamps. Unlike previously published work that propose new message ordering semantics, our approach is based on conservative, time sta ..."
Abstract
- Add to MetaCart
In this paper we describe an approach to exploit temporal uncertainty in parallel and distributed simulation by utilizing time intervals rather than precise time stamps. Unlike previously published work that propose new message ordering semantics, our approach is based on conservative, time stamp order execution and enhancing the lookahead of the simulation by pre-drawing random numbers from a distribution that models temporal uncertainty. The advantages of this approach are that it allows time intervals to be exploited using a conventional Time Stamp Order (TSO) delivery mechanism, and it offers the modeler greater statistical control over the assigned time stamps. An implementation of this approach is described and initial performance measurements are presented. 1.
