Parallel discrete event simulation on shared memory multiprocessors
Abstract: "This thesis presents the implementation and the performance study of various parallel discrete event simulation (PDES) methods on a centralized shared memory machine (an Alliant FX-80). One of the most important features of this implementation is that it provides a single user interf...
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Buch |
| Sprache: | English |
| Veröffentlicht: |
Urbana, Ill.
1991
|
| Schriftenreihe: | Center for Supercomputing Research and Development <Urbana, Ill.>: CSRD report
1074 |
| Schlagworte: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Zusammenfassung: | Abstract: "This thesis presents the implementation and the performance study of various parallel discrete event simulation (PDES) methods on a centralized shared memory machine (an Alliant FX-80). One of the most important features of this implementation is that it provides a single user interface for all the implemented methods. These methods include the Chandy-Misra paradigm with deadlock avoidance (using a shared memory model), the Time Warp approach with direct, aggressive and lazy cancellation, and a hybrid approach, which exploits the parallelism available at each point in simulated time Two issues proved to be significant in the efficient implementation of the above methods: partitioning the tasks among the available processors, and processor self-scheduling. In this thesis, we examine various partitioning and scheduling strategies and assess their impact on the performance of the system. To study the performance of the implemented simulation methods, we simulated two kinds of systems: a synchronous multiprocessor machine with an Omega interconnection network, and an asynchronous toroid network where each node is a simple first-come- first-serve server. The performance of the implemented methods is discussed and conclusions are drawn from the obtained results. |
|---|---|
| Beschreibung: | 79 S. |