System - Level Design Space Exploration of Reconfigurable Architectures

R econfigurable architectures are becoming increasingly popular as they bear a promise of combining the flexibility of software with the performance of hardware. Nevertheless, such architectures are subject to numerous constraints, such as performance, memory requirements, chip area, and power consumption. To create an efficient design, performing Design Space Exploration (DSE) at various stages is essential in order to effectively appraise several design alternatives. DSE at early design stages facilitates designers in rapid performance evaluation of different parameters, such as architectural characteristics, application-to-architecture mappings, scheduling policies, and hardware-software partitionings. DSE methodologies help traversing (typically) huge design spaces efficiently, thus performing DSE at a high level of abstraction facilitates design decisions to be made at very early design stages, which can significantly reduce the overall design time of a system. Towards this goal, in this dissertation, we develop a generic system-level framework, called rSesame, in order to perform modeling and simulation of dynamically reconfigurable architectures at early design stages. The framework can be deployed as a standard modeling and simulation framework for performing system-level DSE to explore several design parameters, while designing dynamically reconfigurable architectures. Performing runtime evaluations together with static explorations, enables reconfigurable architectures to be more efficient in terms of several design constraints. As a result, the rSesame framework combines both static and runtime explorations in order to facilitate system-level DSE of reconfigurable architectures with respect to architectural exploration, hardware-software partitioning and task mapping/scheduling. We deployed the rSesame framework to evaluate the Molen reconfigurable architecture by assessing and evaluating a wide range of applicationto-architecture-mappings. These mappings are evaluated based on different