Quality-driven Design of Deep Neural Network Accelerators for CPS and IoT Applications

Ahstract-This paper presents the results of our analysis of the main problems that have to be solved in the design of highly parallel high-performance accelerators for deep neural networks (DNN) used in Cyber-Physical System (CPS) and Internet of Things (IoT) devices, in application areas such as smart automotive, health and smart services in social networks (Facebook, Instagram, Twitter), etc. Our analysis demonstrates that to arrive at a high-quality DNN accelerator architecture, complex mutual trade-offs have to be resolved among the accelerator micro- and macro-architecture, and the corresponding memory and communication architectures, as well as, among the performance, power consumption and area. Therefore, we developed a multi-processor accelerator design methodology involving an automatic design-space exploration (DSE) framework that enables a very efficient construction and analysis of DNN accelerator architectures, as well as, an adequate trade-off exploitation. Keeping in view the large on-chip memory demands for DNN, we extend our quality-driven model-based multi-processor accelerator design methodology with some novel data reuse techniques. Currently, we are beginning to apply this methodology with the proposed data reuse techniques to the design of DNN accelerators for IoT applications