Core-3399-JD4 Six-core High-performance AI Core Board
Adapts Rockchip high-performance six-core 64-bit processor RK3399, on-board AI neural network processor NPU SPR2801S,
with high computing capacity, ultra-high performance, powerful hardware decoding ability and rich extended interfaces.
It can be flexibly applied to cluster server, computer vision, commercial display equipment, etc.
Six-core 64-bit High-performance Processor
Adapts RK3399 six-core 64-bit (A72x2+A53x4) high-performance processor, frequency up to 1.8GHz, integrated quad-core ARM
high-end GPU Mali-T860. Provide a variety of storage configuration options, users only need to expand the function
backplane to quickly achieve project development.
AI Neural Network Processor NPU
Onboard AI neural network acceleration chip SPR2801S, adopts AI special architecture APiM, its computing capability up
to 2.8 Tops, 9.3 Tops/W ultra-high efficiency. It can be used in the application domain of low energy consumption and
high computing performance products.
Powerful Hardware Decoding Capability
Supports dual LVDS, EDP, HDMI, DP1.2 multiple display output interfaces, dual-screen identical display/dual-screen
differential display, supports 4K VP9, 4K 10bits H265/H264, 1080P (VC-1, MPEG-1/2/4, VP8) multi-format video decoding,
and 1080P (H.264, VP8) video coding.
Stable And Reliable
With SODIMM 260P interface, the data transmission and expansion performance can be best achieved, immersion gold process
pin, corrosion resistant, 2 studs fixed, stable and reliable. Designed measurement is only 69.6mm x 55mm for saving more
precious space.
Rich Extension Interfaces
With rich interfaces such as I2C, SPI, UART, ADC, PWM, GPIO, PCIe, USB3.0, I2S (supports 8-way digital microphone array
input), and so on.
High-performance Industry Backplane
Core-3399-JD4 core board can be combined with the backplane to form a complete high-performance industry application
mainboard, the expansion interfaces are richer, the performance is more powerful, the mainboard can be directly applied
to a variety of intelligent products to accelerate products development.
Support For Multiple OS
Supports Android, Linux+QT, Ubuntu multiple operating system, the performance is stable and reliable
Application
Core-3399-JD4 is suitable for cluster servers, high-performance computing/storage, computer vision, gaming equipment,
commercial display equipment, medical equipment, vending machines, industrial computers, etc.
-
Industrial Computer
-
Smart Robot
-
Game Equipment
-
Commercial Display
-
Medical Equipment
-
Vending Machine
-
Intelligent POS
-
Interactive Printer
Specification
| Specification |
| SoC |
Rockchip RK3399 (28 nm HKMG manufacture procedure) |
| CPU |
ARM® Six-core 64-bit processor, frequency up to 1.8GHz
Based on the big.LITTLEcore architecture, dual-core Cortex-A72+quad core Cortex-A53
|
| GPU |
ARM® Mali-T860 MP4 Quad-core GPU
Support OpenGL ES1.1/2.0/3.0/3.1, OpenVG1.1, OpenCL, DX11
Support AFBC(frame buffer compression)
|
| VPU |
Support 4K VP9 and 4K 10bits H265/H264 video decoding, up to 60fps
1080P multi-format video decoding (WMV, MPEG-1/2/4, VP8)
1080P video coding, support H.264，VP8 format
Video post processor, de-interlacing, de-noising, edge/detail/color optimization
|
| RAM |
2GB/4GB LPDDR4 |
| Storage |
8GB high-speed eMMC 5.1 （8GB/16GB/32GB/128GB）
|
| NPU |
Core-3399-JD4（AI）onboard AI neural network processor SPR2801S
Computing power up to 2.8 TOPS，peak up to 5.6Tops，9.3Tops/W ultra-high efficiency
Support PyTorch, Caffe framework, follow-up support TensorFlow
Support for PLAI model training tools (currently support VGG-based GNet1, GNet18 and GNetfc network models)
|
| Hardware Features |
| Ethernet |
10 / 100 / 1000 MbpsEthernet interface（RJ45）
|
| Display |
1 x HDMI 2.0, support 4K@60HZ output and HDCP 1.4/2.2
1 x MIPI-DSI , support single channel 2560x1600@60fps output
1 x eDP 1.3 ( 4 lanes with 10.8Gbps )
1 x DP 1.2 (DisplayPort), support 4K@60Hz output
Support dual-screen identical display/dual-screen differential display
|
| Audio |
1 x HDMI 2.0 audio output
1 x SPDIF, for audio output
2 x I2S for audio output and input
|
| Camera |
2x MIPI-CSI camera interface( built-in dual-ISP, Maximum support single 13Mpixel or dual 8Mpixel )
|
| USB |
2 x USB2.0 Host, 1 x USB3.0 Type-C |
| Interface |
I2C, SPI, UART, ADC, PWM, GPIO, PCIe, USB3.0,
I2S（Support 8-way digital microphone array input）
|
| Power |
DC input voltage 5V |
| OS/Software |
| OS |
Android 、Linux |
| Appearance |
| Size |
69.6mm × 55 mm |
| Type |
Gold finger（SODIMM 260P, 0.5mmpitch） |
| PCB |
10-layer board design |
Core-3399-JD4 V2
Six-core High-performance AI Core Board
Equipped with high-performance core processor RK3399 and AI neural network processor NPU SPR5801S, supporting stable operation of multiple operating systems, combined with the backplane can form complete industry main board, with more powerful performance and can be directly applied to cluster servers, industrial computers, commercial display devices, etc.
Six-core Processor RK3399
Adopts RK3399 six-core 64-bit (A72x2+A53x4) high-performance processor, frequency up to 1.8GHz, integrated quad-core GPU Mali-T860. Provide a variety of storage configuration options, users only need to expand the function backplane to quickly achieve project development.
Powerful AI computing performance
Onboard AI neural network acceleration chip SPR5801S, adopts AI special architecture APiM, its computing capability up to 2.8 Tops, 9.3 Tops/W ultra-high efficiency. It can be used in the application domain of low energy consumption and high computing performance products.
Powerful Hardware Decoding Capability
Supports dual LVDS, EDP, HDMI, DP1.2 multiple display output interfaces, dual-screen identical display/dual-screen differential display, supports 4K VP9, 4K 10bits H265/H264, 1080P (VC-1, MPEG-1/2/4, VP8) multi-format video decoding, and 1080P (H.264, VP8) video coding.
Stable And Reliable
With SODIMM 260P interface, the data transmission and expansion performance can be best achieved, immersion gold process pin, corrosion resistant, 2 studs fixed, stable and reliable. Designed measurement is only 69.6mm x 55mm for saving more precious space.
Multiple OS
Supports Android, Linux+QT, Ubuntu multiple operating system, the performance is stable and reliable
Rich Extension Interfaces
With rich interfaces such as I2C, SPI, UART, ADC, PWM, GPIO, PCIe, USB3.0, I2S (supports 8-way digital microphone array input), and so on.
Industry Application Main Board
Combined with the backplane to form complete high-performance industry application main board, the expansion interfaces are richer, the performance is more powerful, the main board can be directly applied to a variety of intelligent products.
Application
Suitable for cluster servers, high-performance computing/storage, computer vision, gaming equipment, commercial display equipment, medical equipment, vending machines, industrial computers, etc
-
Cluster Server
-
Computing/storage
-
Smart Robot
-
Gaming equipment
-
Advertising machine
-
Medical equipment
-
Vending machine
-
Industrial computer
Specification
| Specification |
| SoC |
Rockchip RK3399 (28nm HKMG process) |
| CPU |
ARM®Six-core 64-bit processor, frequency up to1.8GHz
Based on big.LITTLE core architecture, dual Cortex-A72(big core)+quad Cortex-A53(little core)
|
| GPU |
ARM® Mali-T860 MP4 Quad-core GPU
Support OpenGL ES1.1/2.0/3.0/3.1, OpenVG1.1, OpenCL, DX11
Support AFBC(frame buffer compression)
|
| VPU |
Support 4K VP9 and 4K 10bits H265/H264 video decoding, up to 60fps
1080P multi-format video decoding (WMV, MPEG-1/2/4, VP8)
1080P video coding, support H.264，VP8 format
Video post processor: de-interlacing, de-noising, edge/detail/color optimization
|
| RAM |
2GB/4GB LPDDR4
|
| Storage |
32GB high-speed eMMC 5.1 (16GB/32GB/128GB)
Support TF card expansion, M.2 PCIe M-KEY expansion NVMe protocol SSD
|
| NPU |
SPR5801S Neural network AI processor (Optional)
• Computing power up to 2.8 Tops, 9.3Tops/W ultra-high efficiency
• Support PLAI(PyTorch), MDK(Caffe) Model training and conversion tools, TensorFlow
• Support picture classification model: VGG-16(GNet1), GNet18 and GNetfc,
Target detection model: SSD(Based on VGG)
|
| Hardware Features |
| Ethernet |
Integrated GMAC Ethernet controller
Support extended RTL8211F to realize 10/100/1000Mbps Ethernet
|
| WiFi |
Extend WiFi & Bluetooth through SDIO3.0
|
| Display |
1 x HDMI 2.0, Support 4K@60HZ output and HDCP 1.4/2.2
1 x MIPI-DSI, Support single channel 2560x1600@60fps output
1 x eDP 1.3 (4 lanes with 10.8Gbps)
1 x DP 1.2 (DisplayPort), support 4K@60 fps output
Support dual-screen identical display/dual-screen differential display
|
| Audio |
1 x HDMI 2.0 audio output
1 x SPDIF, for audio output
2 x I2S for audio input and output
|
| Camera |
2x MIPI-CSI camera interface(built-in dual ISP, Maximum support single 13Mpixel or dual 8Mpixel)
|
| USB |
3 x USB2.0 (If NPU is selected, 1 USB2.0 will be occupied)
1 x USB3.0
|
| Other |
I2C×8, SPI×5, UART×5, ADC×4, PWM×3, GPIO×107, PCIe×1,
I2S×2（Support 8-way digital microphone array input）
|
| Power |
DC 5V
|
| OS/Software |
| OS |
Android, Linux+QT, Ubuntu
|
| Appearance |
| Size |
69.60 mm × 50.27 mm |
| Type |
Connecting finger(SODIMM 260P, 0.5mm pitch) |
| PCB |
8-layer board design |
