32Tops High Computing Power AI Mainboard

Powered by SOPHON AI processor BM1684X, this mainboard can be configured with 16GB of RAM. With up to 32TOPS of INT8 computing power,
it provides mainstream programming frameworks, open SDK, a one-stop AI development toolkit and a complete, easy-to-use toolchain,
featuring easy deployment of algorithm applications and low cost of algorithm migration. This mainboard can be applied to intelligent security,
intelligent transportation, smart park, smart retail, intelligent finance, smart city, smart industry, smart energy and more

New-gen Octa-core AI processor BM1684X

This mainboard is powered by SOPHON AI processor BM1684X, which features octa-core ARM Cortex-A53, frequency of up to 2.3GHz and 12nm
lithography process. With up to 32Tops (INT8) computing power, or 16TFLOPS (FP16/BF16), or 2Tops (FP32), it supports mainstream programming
frameworks, and can be widely used in artificial intelligence inference for cloud and edge applications

Powerful multi-channel video AI performance

The mainboard supports up to 32-channel 1080P H.264/H.265 video decoding. It is able to simultaneously process and analyze over
16-channel HD video, making it ideal for various AI applications such as face detection and license plate recognition on video streaming

Complete software and hardware

With complete software and hardware, artificial intelligence inference for cloud and edge applications can be easily achieved. All of them accelerate
the development of edge applications, such as face recognition, video structuring, abnormal alarm, equipment inspection, and situation prediction, etc

One-stop toolkit, convenient and efficient

The SOPHON SDK one-stop deep learning development toolkit provides a series of software tools, including the underlying driver environment,
compiler, and inference deployment tool. It supports mainstream frameworks: Caffe/TF/PyTorch/Mxnet/Paddle, mainstream network model and
custom operator development, Docker containerization, and rapid deployment of algorithm applications

Abundant algorithms

The mainboard supports multiple algorithms porting such as "person/vehicle/object" recognition, video structuring, and trajectory behavior,
featuring high security and reliability. It can be flexibly applied to various product development

A variety of interfaces

A variety of interfaces includes M.2 (5G), Mini PCIe, SATA3.0, Gigabit Ethernet, USB3.0, RS485, RELAY and GPIO

Firm and tight BTB interface

A core board is equipped with a high-speed industrial-grade connector with high reliability and stability. The BTB interface provides a tight and firm
connection, allowing the core board to be combined with a backplane to form a complete, high-performance industrial mainboard.

Extensive development resources

SDK, tutorials, technical documents, and development tools are provided, making development simpler and more convenient

A wide range of applications

The mainboard supports flexible deployment in multiple fields and scenarios, such as intelligent security, intelligent transportation,
smart park, smart retail, intelligent finance, smart city, smart industry, and smart energy

Smart finance
Smart city
Smart industry
Smart retail


Basic specifications



High-performance octa-core ARM A53, 12nm lithography process, frequency up to 2.3GHz


Built-in tensor computing module TPU,

computing power up to: 32TOPS (INT8), 16TFLOPS (FP16/BF16), 2TFLOPS (FP32)

Support mainstream programming frameworks: TensorFlow / Caffe / PyTorch /

Paddle / ONNX / MXNet / Tengine / DarkNet


32-channel H.265/H.264 1080p@25fps video decoding

32-channel 1080P@25fps HD video processing(decoding + AI analysis)

Support 12-channel H.265/H.264 1080p@25fps video encoding




64GB eMMC 、128MB SPI Flash


2*1000Mbps Ethernet


4 * USB3.0(Max:1A)

Other Interfaces

M.2(5G)、Mini PCIe、SATA3.0、Gigabit Ethernet、USB3.0、RS485、RELAY、GPIO


DC 12V/5A(5.5mm*2.1mm)




183mm * 105mm * 41.8mm


≈380g(Including heat sink)


Operating Temperature: -20℃~60℃,

Storage Temperature:-20℃~70℃,

Operating humidity: 10%~90%(non-condensing)