Kendryte K210 is a dual-core RISC-V processor specifically designed for edge AI and machine vision applications. Operating at a clock speed of 400 MHz, it is built on the RV64IMAFDC architecture, enabling efficient processing for various AI tasks.
Key Features
-
Dual-Core Architecture: The K210 features two 64-bit RISC-V cores, providing a solid foundation for parallel processing tasks, which is essential for real-time AI applications.
-
Neural Network Processor (KPU): The K210 includes a dedicated KPU that can perform up to 0.8 TOPS (Tera Operations Per Second). This hardware accelerator is optimized for convolutional neural networks (CNN), making it suitable for tasks such as image recognition and classification.
-
Memory Configuration: It is equipped with 8 MB of SRAM, which allows for efficient data handling and storage during processing tasks. Additionally, the K210 supports external storage through a microSD card slot, enhancing its versatility in various applications.
-
Low Power Consumption: The K210 is designed to operate efficiently, consuming less than 1 watt during active use and dropping to 300 mW when idle. This low power requirement is crucial for battery-operated devices and edge computing scenarios.
-
Development Support: The K210 supports programming environments like MicroPython, allowing developers to create applications more easily. It is compatible with various development boards, enhancing its accessibility for prototyping and deployment.
Applications
The Kendryte K210 is particularly well-suited for applications in machine vision, such as:
-
Facial Recognition: Capable of processing facial recognition algorithms at 60 frames per second (fps) at QVGA resolution, making it ideal for security and surveillance applications.
-
Object Detection: The K210 can classify and detect objects in real-time, which is beneficial for robotics and automated systems.
-
Audio Processing: It includes an audio processing unit (APU) for tasks such as sound source localization and audio beamforming, expanding its use cases in smart home devices and interactive systems.
The K210’s combination of processing power, efficiency, and flexibility makes it a compelling choice for developers looking to implement AI capabilities in edge devices and machine vision systems[1][2][4].
Further Reading
1. Kendryte K510 tri-core RISC-V AI processor deliver up to 3 TOPS – CNX Software
2. Kendryte’s KD233 Is a Dual-Core RISC-V SoC Designed for AI Applications – Hackster.io
3. New Part Day: The RISC-V Chip With Built-In Neural Networks | Hackaday
4. UnitV k210 Edge Computing AI Camera(OV7740) | m5stack-store
5. M5StickV AI Camera – Kendryte K210 Chipset (no Wi-Fi) : ID 4321 : Adafruit Industries, Unique & fun DIY electronics and kits