Energy-efficient noC-based systems for real-time multimedia applications using approximate computing

• Ano de defesa: 2020
• Autor(a) principal: Penny, Wagner Ishizaka
• Orientador(a): Zatt, Bruno
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Pelotas
• Programa de Pós-Graduação: Programa de Pós-Graduação em Computação
• Departamento: Centro de Desenvolvimento Tecnológico
• País: Brasil
• Palavras-chave em Português: Computação; NoC; Approximate computing; Machine learning; Video coding; Hardware acceleration; Computação aproximada; Aprendizagem de máquina; Codificação de vídeo; Aceleração em hardware
• Área do conhecimento CNPq: CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO
• Link de acesso: http://guaiaca.ufpel.edu.br/handle/prefix/7248

This thesis presents an energy-efficient NoC-based system for real-time multimedia applications employing approximate computing. The proposed video processing system, called SApp-NoC, is efficient in both energy and quality (QoS), employing a scalable NoC architecture composed of processing elements designed to accelerate the HEVC Fractional Motion Estimation (FME). SApp-NoC architecture is organized using neighbor Tiles, sized to enable scalability across distinct throughput demands - depending on video resolution and frame rate - whereas reaching real-time processing for 4K UHD videos at 120 fps. Approximate computing is deployed using four types of processing elements implemented as dedicated hardware accelerators with distinct levels of approximation, designed based on the application error resiliency analysis. Therefore, two solutions are proposed: HSApp-NoC (Heuristc-based SApp-NoC), and MLSApp-NoC (Machine Learning-based SApp-NoC). At design time, video encoder statistical behavior is used to propose algorithms aiming the tiling definition, to properly size the NoC and to instantiate and place the approximate processing elements within SApp-NoC. At run-time, our application-aware dynamic task-mapping algorithm guarantees real-time processing while reducing energy consumption with low QoS degradation. When compared to a precise solution processing 4K videos at 120 fps, HSApp-NoC and MLSApp-NoC reduce about 48.19% and 31.81% the energy consumption, at small quality reduction of 2.74% and 1.09%, respectively. A set of schedulability analysis is proposed in order to guarantee the meeting of timing constraints at typical workload scenarios. Moreover, our system design methodology is suitable to be applied to other error-resilient processing kernels targeting energy saving with high throughput requirements.