SYSGO
REBECCA Hardware/Software Edge AI Platform Overview
Pages
3
Time to read
6 mins
Publication
Language
English
Pages
3
Time to read
6 mins
Publication
Language
English
This document is a technical report on the REBECCA project, which focuses on advancements in edge AI systems utilizing RISC-V technology. The project aims to create a multicore RISC-V-based architecture that integrates AI-specific accelerators and neuromorphic computing to enhance power efficiency and scalability. The REBECCA platform features the CVA6 processor, designed with a chiplet-based architecture and shared memory to optimize real-time AI processing. Initial prototypes have been validated using U55C development boards, demonstrating the feasibility of RISC-V for AI applications. The report outlines the architecture's key features, including hardware-accelerated AI processing, intrusion detection systems, and specialized machine learning accelerators. Additionally, it discusses the development of a HyperRAM module to improve memory management and data processing efficiency. Future research directions include enhancing neuromorphic computing capabilities and optimizing real-time AI performance. The collaboration between industry and academia is emphasized as crucial for the project's success.