Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.643306
Title: Simulation models of shared-memory multiprocessor systems
Author: Coe, Paul
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2000
Availability of Full Text:
Access through EThOS:
Full text unavailable from EThOS. Please try the link below.
Access through Institution:
Abstract:
Multiprocessors have often been thought of as the solution to today's every increasing computing needs; but they are expensive, complex and difficult to design. This thesis focusses on the development of multiprocessor simulations that would aid the design and evaluation of such systems. The thesis starts by outlining the various possibilities for multiprocessor design and discusses some of the more common problems that must be addressed. A selection of simulation environments and models that have been developed to study complex computer systems are then described. The major problem with these simulation systems is that they generally focus on a small area of multiprocessor systems design in order to produce fast simulations that generate results quickly; consequently they provide very little flexibility and room for exploration. The aim of this project was to design and implement a flexible multiprocessor model within the HASE simulation environment, enabling the designer to explore a large design space with a minimum of effort, focussing more on flexibility and less on simulation speed. A parameterised simulation model has been developed that presents the designer with many design options with which to experiment. The parameters allow simple alternatives to be explored, for example, different component speeds or bus widths, as well as more complicated features, for example, coherence protocols, synchronisation primitives and architecture configurations. The model was designed in a modular manner that allows new parameter values to be incorporated, as well as new implementations of the various entities. To support this new model, the HASE system was extended to provide better support for multiprocessor modelling.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.643306  DOI: Not available
Share: