Specifications
Product Family
SOPHGO Micro Server SE9, SE7 and SE8 edge video analytics servers
SE9 16-BP1-11
BM1688, 16-channel HD video analysis, 8GB memory
SE9 16-BP1-18
BM1688, 16-channel HD video analysis, 16GB memory
SE9 8-BP1-11
CV186AH, 8-channel HD video analysis, 4GB memory
SE9 8-BP1-17
CV186AH, 8-channel HD video analysis, 8GB memory
SE7 32-EA4-23
BM1684X, 32-channel HD video analysis
SE7 32-BP1-11
BM1684X, 32-channel HD intelligent analysis
SE8 288-EA6-72
BM1684X, 288-channel HD video analysis — high-density analytics node
Processor — BM1688
8-core Arm Cortex-A53 @ 1.6 GHz with 16 TOPS INT8 TPU
Processor — BM1684X
8-core Arm Cortex-A53 @ up to 2.3 GHz fourth-generation SOPHON TPU
Processor — CV186AH
16M high-end intelligent deep learning vision processor
Video Decode
Up to 16 channels of full HD decoding on BM1688 platforms
Video Encode
Up to 10 channels of full HD encoding on BM1688 platforms
AI Precision
INT4 / INT8 / INT16 / FP16 / BF16 / FP32 mixed precision
Model Formats
ONNX, Caffe and TFLite import
Frameworks
TensorFlow, PyTorch, Paddle, TensorRT and BM1684 / BM1684X compatible toolchains
Ecosystem Products
Coeus-3550T and EC-1684JD4 deep learning compute boxes on BM1684
Scale-up Options
CSA1-N8S1684 1U cluster server (8x BM1684); IVP03X 32-channel analytics workstation
Deployment Targets
Visual computing, edge computing, intelligent transportation, smart campus and smart classroom
Memory Range
4GB up to 16GB across the SE9 line
Overview
The SOPHGO Micro Server family brings the company's SOPHON TPU architecture into compact, deployable edge appliances for multi-channel video analytics. The SE9 line runs on the BM1688 processor for 16-channel workloads or the CV186AH vision processor for 8-channel workloads, with memory from 4GB to 16GB; the SE7 line moves to the BM1684X for 32-channel analysis; and the SE8 288-EA6-72 scales to 288 channels for high-density analytics nodes.
These systems target the segment where an NVR cannot run the models and a GPU server is too costly and power-hungry: transportation networks, safe-city deployments, industrial campuses and utility inspection programmes that need dozens of concurrent video streams analysed locally with low latency.
Software compatibility spans the SOPHGO toolchain with ONNX, Caffe and TFLite model import, and development environments compatible with TensorFlow, PyTorch, Paddle, TensorRT and the previous BM1684 / BM1684X generations. That continuity lets integrators move existing SOPHGO pipelines onto newer silicon without re-quantising models.
QS Compute sources SOPHGO SE9, SE7 and SE8 configurations together with the related Coeus and EC deep learning compute boxes and 1U cluster servers. Contact us with your channel count, resolution and model mix for a configured quotation.
Key Benefits
Per-channel economics far below GPU servers for video analytics; a coherent upgrade path from CV186AH through BM1688 to BM1684X; 8 to 288 channels in the same product family; low power and passive cooling options for cabinet mounting; full SOPHGO toolchain and cross-generation model portability.
Applications
Safe-city and intelligent transportation analytics; video surveillance and NVR-class inference; industrial campus monitoring; smart classroom and campus deployments; power-line and infrastructure inspection; edge deep learning microservers.
Request a Quote — SOPHGO MICRO SERVER SE9 / SE7 / SE8 — BM1688 EDGE VIDEO ANALYTICS SERVERS
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →