Adopts Rockchip, integrates quad-core Mali-T860 GPU and carries modular neural deep-learning accelerator NPU with no external buffer memory needed. It has strong peak power and super high performance, supports Pytorch, Caffe deep-learning framework, and provides complete and user-friendly model training tools and web training model cases, enabling it to be rapidly used for mobile edge computing, smart home devices, facial recognition, AI server, etc.
Equipped with ARM Cortex-A72 architecture, six-core 64-bit high-performance processor, frequency up to 1.8 GHz. It integrates a quad-core Mali-T860 GPU, supports H.265 HEVC and VP9, H.264 encoding and 4K HDR, and with a powerful hard decoding capability as high as 4K.
Carried an AI embedded neural network processor NPU, which has the peak performance of up to 5.6 Tops and computing performance 2.8 Tops, with an energy efficiency of up to 9.3 Tops/W. This ensures a powerful peak power while maintaining extremely low power consumption, giving it huge advantages in the field of edge computing used in terminal devices.
AIO-3399C(AI) adopts AI-specific MPE matrix engine and APiM (AI processing in Memory) architecture, local parallel AI computing that combines storage and computing, one-time upgrade network preload with no need of instructions, bus, and external DDR cache. This configuration improves the processing speed hugely and lowers processing energy consumption a lot compared with processors built in traditional architecture approaches.
It provides complete and easy-to-use PyTorch-based model training tool PLAI (People Learn AI), can be developed on Windows 10 and Ubuntu 16.04 systems, and supports fast and easy adding of custom network models, greatly reducing the technical barriers of using AI and allowing more people to open the gate of AI easily.
Supports the following three network training model examples such as GNet1, GNet18 and GNetfc with more network instances continuing to emerge subsequently, making it possible to easily test a large number of deep learning applications on the device.
Equipped with industrial grade metal casing, featuring small size, fanless efficient heat dissipation, dustproof and anti-disruption. It has various installation methods and can be flexibly embedded in various smart devices.
AIO-3399C(AI) can be powered by POE+ (802.3 AT, output power 30W) enhanced Ethernet. It with external expansion interfaces such as RS232, RS485 and 2 TTL, which is convenient for connecting various industrial devices.
Specification | |
CPU |
RK3399,Dual-core(Cortex-A72)+Quad-core(Cortex-A53), frequency up to 1.8 GHz |
GPU |
Quad-core ARM Mali-T860 Support OpenGL ES 1.1/2.0 /3.0, OpenVG1.1, OpenCL, Directx11 |
NPU |
SPR2801S, Adopt MPE and APiM unique AI architecture Computing performance up to 2.8 Tops and 9.3Tops/W energy efficiency |
DDR |
2GB / 4GB dual-channel LP DDR4 |
Storage |
8GB -128GB High-SpeedeMMC, TF Card Slot |
Hardware Features | |
Network |
RJ45interface Gigabit Ethernet On-boardWIFI/BTmodule, support 2.4GHz/5GHzdual-bandWiFi, 802.11a/b/g/n/ac protocol Support Bluetooth 4.1(Support BLE) Mini PCIe(Used to expand 3G/4G modules, use with Micro SIM card slot) |
Multimedia |
Support 4K VP9 and 4K 10bits H265/H264 video decoding, up to 60fps 1080P Multi-format video decoding(VC-1,MPEG-1/2/4,VP8) 1080P video decoding, support H.264,VP8 formats Video post processor: deinterlacing, denoising, edge/ detail/ color optimization |
Display |
Dual VOP: support 4096X2160 and 2560X1600 resolution HDMI2.0 support 4K 60Hz display, support HDCP 1.4/2.2 Support eDP 1.3(4 line, 10.8Gbps),can directly drive multiple resolutions LCD screen which with eDP interface Support dual 6/8 bit LVDS interface, up to 24-bit 1920×1200 resolution Support Rec.2020 and Rec.709color gamut conversion DP 1.2 (DisplayPort)x 1, support 4K@60 fps output |
Interface |
Dual ISPpixel processing capability up to13MPix/s, supportsimultaneous input of two-way camera data Support USB3.0 HOST and Type-C ADC x 1, SPI / GPIO, LED×2, I²C×1, Gravity sensor×1(Scalable) |
SD Card |
Support SD Card |
RTC |
Support RTCreal-time clock |
On/Off |
Support timer switch |
Audio |
PHONEx 1, LINE-INx 1, LINE-OUTx 1, Microphone (left and right channel) |
USB |
Type-C(OTG), USB3.0x1, USB2.0x4(interface x 2, socket x 2) |
Key |
Power Key(key×1, socket×1), Recover Key(key×1, socket×1) |
Serial port |
RS232×1, RS485×1, Debug serial port×1, on-board 2-way TTL |
IR |
With a one-way infrared receiver, support infrared remote control |
Power |
With a one-way infrared receiver, support infrared remote control DC 12V-2A(DC5.5 × 2.1mm), Support for external connection( Power socket×1) Can power by POE+(802.3 AT, Output Power30W) Ethernet |
OS/Software | |
System |
Support Android\Linux\Ubuntu system |
Framework |
Support PyTorch , Caffe framework, follow-up support TensorFlow |
Tools |
PLAI model training tool(Support for GNet1, GNet18 and GNetfc network models which based on VGG) |
Appearance | |
Size |
126 mm× 91.3mm |