T-Chip Group
Intelligent hardware website
Innovate intelligent hardware and solutions.
Server Business Website
Server products and efficient solutions.
Consumer Business Website
StationPC NAS products and services.
CSA1-N8S1684X Overview Specifications Wiki Download
Buy Now DataSheet
Buy Now DataSheet

CSA1-N8S1684X AI Computing Power Server

Powered by high-computing BM1684X, this server features up to 8 BM1684X computing modules. It delivers exceptional computing power,
enabling up to 256TOPS (INT8) peak computing power, or 128TFLOPS (FP16/BF16) computing power, or 16TFLOPS (FP32) high-precision
computing power, with the power of multi-channel video processing capability. It provides multiple algorithm migration and a one-stop deep
learning development toolkit. With the BMC management system, secondary development is supported

AI processor BM1684X

The BM1684X SoC with AI computing power features an octa-core Cortex-A53 + NPU. Cortex-A53 with a frequency of 2.3G. NPU
delivers 32Tops(INT8) / 16TFLOPS(FP16/BF6) / 2TFLOPS(FP32) computing power. The server provides 16GB LPDDR4 and 128GB
eMMC large storage capacity

Eight computing modules, flexible configuration

The server can be equipped with up to 8 BM1684X computing modules. Users can customize the number and storage configuration of
modules. This flexibility makes it suitable for various scenarios

256TOPS powerful computing power

It supports up to 256TOPS (INT8) peak computing power, or 128TFLOPS (FP16/BF16) computing power, or 16TFLOPS (FP32) high-precision
computing power, which can meet the application requirements for deep-learning model development

Incredible multi-channel video processing capability

The server supports up to 256-channel H.264 1080p@25fps video decoding, 256-channel H.264 1080P@25fps HD video processing
(decoding + AI analysis), and 96-channel H.264 1080p@25fps video encoding

Various algorithms, strong practicality

This server supports algorithm migration for pedestrian/vehicle/object recognition, video structuring, trajectory behavior analysis, etc.,
featuring high security and reliability. So it can be flexibly applied to various product development

Open SDK, one-stop AI development kit

SOPHON SDK (BMNNSDK2), a 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

BMC management system

BMC intelligent management system provides a visualization console interface through HDMI output. It easily achieves real-time monitoring,
software configuration, hardware management, troubleshooting, and system upgrade. Secondary development is also supported

Dual 10GbE meets high bandwidth requirements

Dual 10GbE ports and GbE ports meet business scenarios with high bandwidth requirements. The independent BMC management network
interface separates the management network and the service network, ensuring the security and reliability of network communication

Support SATA3.0 hard disk expansion

A 3.5-inch hard disk bay supports SATA3.0 HDD/SSD hard drive expansion, allowing the device to be easily expanded to a TB storage capacity

Standard 1U rack server

Featuring a standard 1U rack server chassis design with a depth of 576mm, this server perfectly matches a wide range of data center cabinets

A wide range of application scenarios

This server can be widely used in various scenarios, including edge computing, cloud storage, blockchain, multi-channel video
encoding and decoding, and intelligent security

Edge computing

Security

Cloud storage

Blockchain

Multi-channel video
encoding and decoding

Financial industry

Specifications

Specifications

CSA1-N8S1684X
Technical Specifications
 Server form

1U rack-mounted computing power server
Architecture

ARM architecture
Number of nodes

8 distributed computing nodes + 1 control node
Compute nodes

Octa-core 64-bit processor BM1684X, up to 2.3GHz
Video encoding

H.264: 3×4K@25fps, 12×1080P@25fps
Video decoding

H.264: 8×4K@25fps, 32×1080P@25fps
Control nodes

Octa-core 64-bit processor RK3588S, main frequency up to 2.4GHz, the highest computing power is 6TOPS
AI computing power

256TOPS (32T × 8, INT8)
RAM

16GB LPDDR4/LPDDR4X × 8
Storage

64GB eMMC × 8 (64GB/128GB Optional)
Storage expansion

3.5-inch/2.5-inch SATA3.0 SSD hard drive slot × 1 (Supports hot swapping; BMC can directly operate the hard drive, and computing child nodes can indirectly access the hard drive through the network sharing method provided by BMC)
Power consumption

Normal: 300W, Max: 430W
Fan module

6 high-speed cooling fans
Physical Specifications
 Size

490.0mm(L) × 417.3mm(W) × 44.4mm(H)
Installation requirements

IEC 297 Universal Cabinet Installation: 19 inches wide and 800 mm deep and above Retractable slideway installation: The distance between the front and rear holes of the cabinet is 543.5mm~848.5mm
Full weight

Server net weight: 6.7kg, total weight with packaging: 8.9kg
Environment

Operating Temperature: 0ºC ~ 42ºC, Storage Temperature: -40ºC ~ 60ºC, Operating Humidity: 20% ~ 80%RH (non-condensing)
Software Specifications
 BMC

The BMC management system is integrated with the web-based management interface, supporting Redfish, VNC, NTP, monitoring advanced and virtual media, and the BMC management system can be redeveloped
Large language models

BM1684X: Support the privatization of ultra-large-scale parametric models under the Transformer architecture, such as Deepseek-R1 series, Gemma series, Llama series, ChatGLM series, Qwen series, Phi series and other large language models
Visual large model

BM1684X: Support the privatization deployment of large visual models such as ViT, Grounding DINO, SAM, etc.
AI Painting

BM1684X: Support the private deployment of Flux, Stable Diffusion, and Stable Diffusion XL image generation models
Deep learning

All models: Support traditional network architectures such as CNN, RNN, LSTM, and support various deep learning frameworks such as TensorFlow, PyTorch, PaddlePaddle, ONNX, and Caffe. Support custom operator development and Docker containerization management technology
Interface Specifications
 Internet

2 × 10G Ethernet (SFP+), 2 × Gigabit Ethernet (RJ45, 1 management network port, 1 ordinary network port)
Display

1 × HDMI2.0 (Maximum resolution 1080P, main processor core board display)
USB

2 × USB3.0 HOST, 1 × Type-C (USB3.0 OTG, processor core board debugging)
Others

1 × SIM Card, 2 × 4G antenna