Scale4Edge

Scalable low-power RISC-V-based processor platform with application-specific AI accelerators

01/05/2020 to 31/12/2025

Scale4Edge provides an ecosystem for a scalable and flexibly expandable edge computing platform based on the free RISC-V instruction set architecture. The ecosystem offers a scalable, customized complete solution including support. This also includes standards-compliant design processes, partially open-source software and also the safety (security) of the platform as well as an intensive review and analysis of the developments. The RISC-V platform of the ecosystem provides the user with three levels for implementation: (1) CPU instruction level, defined by the RISC-V ISA standard; (2) software level, defined by the C standard and complementary standards such as MISRA-C; and (3) operating system level with system services, resource allocations and configuration interfaces. The activities at the University of Tübingen aim at exploring and developing energy- and resource-efficient hardware accelerator architectures for a RISC-V platform to be used in edge computing systems that can be embedded in their environment in a well-tailored manner. This enables efficient processing of sensor data streams with an average electrical power consumption well below 1 mW.