-
Multi-Channel Video Processing: Supports 256-channel 1080p@25fps video decoding + AI analysis, ideal for security and real-time surveillance.
-
Flexible Configuration: Modular design allows customizing compute nodes (1–8 modules) and storage (SATA3.0 HDD/SSD expansion).
-
BMC Management: Integrated web-based monitoring for hardware control, troubleshooting, and secondary development.
-
Connectivity: Dual 10GbE SFP+ ports, GbE, optional 4G/5G, and UPS support for failover resilience.
-
Software Support: Includes SOPHON SDK for TensorFlow/PyTorch/Caffe frameworks and Docker deployment.
Parameter | Specification |
---|---|
Product Name | Computing Power Cluster Server |
Product Model | CSA1-N8S1684X |
AI Computing Power | 256TOPS@INT8 peak, 16TFLOPS@FP32 high-precision |
VPU | Video decoding: H.264 & H.265: 1080P @ 11520fps Video decoding resolution: 8192×8192 / 8K / 4K / 1080P / 720P / D1 / CIF Video encoding: H.264 & H.265: 1080P @ 600fps Video encoding resolution: 4K / 1080P / 720P / D1 / CIF JPEG decoding: 4800 images/sec @1080P |
Nodes | 8 compute nodes (up to 64 ARM cores) + 1 control node |
CPU | Compute node: BM1684 octa-core (A53×8) 64-bit, up to 2.3GHzControl node: RK3588S octa-core (4×Cortex-A76 + 4×Cortex-A55) 64-bit, up to 2.4GHz |
RAM | 16GB LPDDR4 × 8 (compute nodes) |
Storage | 32GB eMMC × 8 (compute nodes), supports 3.5-inch SATA3.0 hard disk |
Network | GE RJ45 Gigabit Ethernet port ×2 (control node + common port)SFP 10GE × 2Optional 4G/LTE/5G |
Display | HDMI 2.0, 4K@60Hz (main processor core board) |
USB | 2 × USB3.0 HOST, 1 × Type-C (for processor core debugging) |
Fan | 5 high-speed fans |
Power | 300W AC power supply (100–240V AC input) |
OS | Linux |
BMC | Integrated BMC management with web-based interface, supports secondary development |
Deep Learning Framework | TensorFlow / PyTorch / Paddle / Caffe / ONNX / MXNet / DarkNet |
Dimension | Standard 1U rack-mounted: 490mm × 390mm × 44.4mm |
Operating Temperature | 0℃ – 50℃ |
Operating Humidity | 10% – 90% RH |