In the wave of digital transformation, edge computing, with its low latency and high efficiency, is gradually becoming an important support for various intelligent applications. Today, we would like to introduce a high-performance development board specially designed for edge computing scenarios—RISC-V + Self-developed DPU Edge Computing Development Board. It not only integrates an innovative RISC-V multi-core CPU and dedicated AI accelerator (DPU), but also demonstrates excellent capabilities in audio processing, visual recognition, live streaming, and other fields.
Heterogeneous Computing, Perfect Balance of Performance and Power Consumption
The biggest highlight of this development board is its heterogeneous computing capability. The collaborative work of the RISC-V multi-core CPU and dedicated DPU enables each to fully utilize its advantages when handling complex tasks, achieving efficient acceleration. Taking the YOLOv7-tiny model as an example, compared with the pure CPU solution, its inference speed is increased by more than 60%. The 6 TOPS INT8 computing power and typical 8-watt ultra-low power consumption achieve a perfect balance between performance and power consumption.
Scenario-based Deep Optimization to Meet Diverse Needs
For different scenarios, this development board offers three models: Audio-M/X, Vision-F/G, and Live-Edge. Audio-M/X is designed for audio scenarios, equipped with 2-channel 24-bit ADC + 1W DAC, combined with 1D FFT hardware unit, achieving ≤2ms audio stream processing delay, built-in AEC/ANS algorithm library, and real-time voice enhancement SNR increased by 20dB. Vision-F/G focuses on visual scenarios, dual MIPI-CSI-2 interfaces combined with feature extraction acceleration engine, processing a frame of 1080p image takes only 18ms, supports multi-person parallel detection, meeting the needs of access control gates, smart building attendance, etc. Live-Edge is born for live streaming, supports 1080p60 video hardware encoding and decoding, aggregates RTMP/SRT/WebRTC protocols, achieves 80ms end-to-end processing delay, and provides strong support for live e-commerce and real-time special effects live streaming.
Industrial-grade Quality, Stable Operation Without Worries
In terms of quality, this development board also performs excellently. The aluminum-magnesium alloy heat dissipation structure, combined with a wide temperature working range of -20℃ to 70℃, supports DIN rail installation. Whether in harsh industrial environments or complex outdoor scenarios, it can operate stably. This provides great convenience for developers to deploy in various application scenarios.
Rich Development Ecosystem, Helping Projects Land Quickly
To make it easier for developers to realize the full-process development from model training to edge deployment, this development board provides a complete toolchain support, including PyTorch, TensorFlow model conversion tools, as well as ONNX and TFLite quantization compilers. At the same time, it is equipped with a rich SDK, reference models, and sample projects. The professional team provides comprehensive technical support services, including hardware customization, algorithm adaptation, and other commercial services, to help projects land smoothly.
RISC-V + Self-developed DPU Edge Computing Development Board, with its powerful heterogeneous computing capability, scenario-based deep optimization, industrial-grade quality, and rich development ecosystem, injects new vitality into the field of edge computing. Whether you are a developer of audio processing, visual recognition, or live streaming, you can find a solution that meets your needs on this development board. Let us explore the infinite possibilities of edge computing together and jointly promote the process of digital transformation.
Source: Wuhan Hongan: https://www.honganinfo.com/computing-power/v-edge/
