Towards Computational Efficiency of Next Generation Multimedia Systems

To address throughput demands of complex applications (like Multimedia), a next-generation system designer needs to co-design and co-optimize the hardware and software layers. Hardware/software knobs must be tuned in synergy to increase the throughput efficiency. This thesis provides such algorithmic and architectural solutions, while considering the new technology challenges (power-cap and memory aging). The goal is to maximize the throughput efficiency, under timing- and hardware-constraints.