The NV8600-Nano AI Developer Kit is built around the NVIDIA Jetson Orin Nano 8GB module, featuring an active cooling solution (heatsink + fan) and mainboard PBA-TON9. It offers powerful performance and extensive connectivity for edge AI and industrial applications.
Key Hardware Features
- Processor: 6-core Arm Cortex-A78AE, up to 1.728 GHz
- Memory: 8GB LPDDR5
- Storage: 256GB NVMe SSD (Phison E13 controller)
- Expansion:
- 3× M.2 slots (Key-E for wireless, Key-B for LTE/5G, Key-M for SSD)
- 40-pin Raspberry Pi-compatible GPIO
- 2× MIPI CSI camera interfaces
- 6× USB 3.0 ports
- 1× SATA
- CAN, COM, NC-SI, debug UART
- Networking: 4× GbE (3× Intel I210 + 1× Realtek RTL8111), PoE supported
- Video Output: HDMI
- Power: 12V/5A (60W adapter), industrial-grade temperature range (-25°C to 70°C)
Included Accessories
- Power adapter (without cable)
- Raspberry Pi Camera Module 2
- USB 2.0 camera
System Setup
The kit comes pre-installed with Ubuntu 22.04.5 LTS + JetPack 6.2 (Linux 5.15).
1. Peripheral Connections
- Connect the USB camera to any USB 3.0 port.
- Connect the Raspberry Pi Camera Module 2 to MIPI CSI0 via the 22-pin ribbon cable.
- Connect power adapter, HDMI display, and keyboard/mouse.
2. Power-On
- Plug in the power; the green LED lights up, and the fan starts spinning.
- Press the power button to boot the system.
3. Login Credentials
- Username: aaeon
- Password: aaeon
Hardware Detection & Testing
- System Info: Identified as AAEON BOXER-8654AI_RefKit, shares some device tree components with Orin NX.
- USB Camera: Works correctly.
- Raspberry Pi Camera Module 2: Initially unrecognized; configured with Jetson-IO as Camera IMX219-C and works properly.
- SSD Performance: Read 2.14 GB/s, write 1.08 GB/s.
USB 3.0 Performance: Six ports show consistent read speeds of 940–953 MB/s. - GbE Performance (iperf3):
- LAN1: Rx 938 Mbps / Tx 744 Mbps
- LAN2–4: ~938–939 Mbps, optimal performance
Benchmark Results
- CPU Performance (sbc-bench): No throttling observed; peak temperature ~51°C.
- Multi-core performance slightly below RK3588, higher than Raspberry Pi 5.
- Browser Performance (Speedometer 2.0): 85.7 RPM (better than RPi5 at 63.5 RPM).
- GPU Performance:
- GLmark2-ES2: 2488 points
- Supports OpenGL 4.6, significantly stronger GPU than most Arm platforms
- Chromium does not enable GPU acceleration by default (manual compilation required)
AI Functionality (Nx Meta Platform)
The kit integrates Network Optix Nx Meta, enabling rapid AI video analysis:
Camera Support:
- Automatic detection of USB cameras
- ONVIF cameras supported (e.g., Reolink TrackMix PoE)
AI Pipeline:
- 80-class object detector (320×320 / 640×640)
- Face detection
Demonstrated Features:
- Object and face detection
- Object counting, line-crossing detection, loitering detection
- PTZ camera control (ONVIF-compatible)
Summary
The NV8600-Nano AI Developer Kit offers enhanced hardware design and expansion compared to the official NVIDIA Orin Nano DevKit:
Advantages:
- Large-capacity SSD, more USB 3.0 ports, 4× GbE with PoE, 3× M.2 expansion
- Industrial temperature range suitable for edge AI deployments
- Nx Meta platform accelerates AI development and prototyping
Limitations:
- Chromium GPU acceleration disabled by default
- LAN1 bidirectional throughput slightly lower than Intel I210 ports
Ideal Use Cases:
- Multi-camera integration
- Edge AI video analytics
- Industrial AI applications
Pricing and Availability:
- The NV8600-Nano AI developer kit is available on the UP Shop for $649 (plus taxes and shipping)
- The UP HD camera is priced at $35
- Compared to NVIDIA’s Jetson Orin Nano Super Developer Kit, it is more expensive but includes additional accessories (256GB SSD with JetPack 6.2 SDK and Raspberry Pi Camera Module 2)
- Provides more features and ports, designed for operation in a wide temperature range (-25°C to 70°C), with potentially better availability than NVIDIA’s low-cost option
Image source: CNX Software
