A Cost Efficiency Analysis and Mechanism for Dynamic Partially Reconfigurable Computing Systems

Abstract

Dynamic Partially Reconfigurable Computing Systems have proven to be useful in environments where a multiplicity of tasks is required. These systems used Dynamic Virtual Components (DVCs) for reconfiguration. However, the computing architecture (dedicated, software, hybrid) of the DVC must be selected during the design stage. In this thesis, a mechanism in which we evaluate and analyze the cost efficiency of a DVC based on a cost efficiency factor (CEF) is proposed. Data centric and stream centric experimental tests were performed and the CEF of a 3D stereo-panoramic augmented reality hybrid DVC was de termined. The results show that development costs and number of units to be produced influence the cost efficiency of a DVC. From the results it is concluded that the CEF can be a useful tool for selecting the computing architecture of a DVC, particularly when only a few units are to be deployed.