Upbeat Technology and SiFive have introduced the UP201/UP301 microcontroller family, a dual-core RISC-V MCU platform designed for low-power edge-processing applications. The devices integrate two SiFive Essential processor cores with two hardware AI accelerators developed by Upbeat, enabling real-time inference and digital signal processing within a constrained power budget.
The architecture supports near-threshold operation and incorporates an error-detection-and-correction (EDAC) scheme and SRAM enhancements to maintain reliability at low operating voltages. The MCUs deliver measured energy efficiency of 16.8 µW/MHz/DMIPS and scale up to 400 MHz or 717 DMIPS, providing a balance of throughput and standby efficiency suitable for always-on systems such as IoT nodes, wearables, drones, and edge sensors.
Integrated AI acceleration offloads inference workloads and DSP tasks from the CPU cores, enabling real-time processing while maintaining extended battery life. The dual-core structure supports heterogeneous workloads, allowing one core to handle control functions while the other manages compute-intensive operations.
Engineering samples, development tools, and reference designs are available, including an evaluation platform designed for low-power edge AI applications. Demonstrations will be shown at the RISC-V Summit alongside Upbeat’s ultra-low-power MEMS bone-conduction microphone.
The MCUs target applications that require continuous sensing and local decision-making while maintaining long operating life on limited energy sources. These include predictive-maintenance nodes, audio-processing systems, robotics, portable devices, and edge-AI systems.
Leave a Reply