Atlas 800 Training Server
Ultra-high computing density
This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>
Enterprise products, solutions & services
Atlas 800 Training Server (Model: 9010)
The Atlas 800 training server (model: 9010) is an AI training server based on the Intel processors and Huawei Ascend 910 processors. It features ultra-high computing density and high network bandwidth. The server is widely used in deep learning model development and training scenarios, and is an ideal option for computing-intensive industries, such as smart city, intelligent healthcare, astronomical exploration, and oil exploration.
The ultimate computing density
• Up to 2.24 PFLOPS FP16 in a 4U space
High-speed network
• 8 100G RoCE v2 ports, slashing cross-server chip interconnect latency by 10–70%
High-speed network bandwidth
• 8 x 100G RoCE v2 high-speed interfaces
• Cross-server chip interconnection latency slashed by 10% to 70%
Specifications
Parameters | Atlas 800 Training Server Model: 9010 |
Form Factor | 4U AI server |
Processor | 2 Intel V5 Cascaded Lake processors |
Processor Memory | Up to 24 DDR4 DIMM slots, supporting RDIMMs |
AI Processor | 8 Ascend 910 processors |
HBM | 32GB, 1228GB/s |
AI Computing Power | 2.24 PFLOPS FP16 1.76 PFLOPS FP16 |
Local storage | • 2 x 2.5'' SATA + 8 x 2.5'' SAS/SATA • 2 x 2.5'' SAS/SATA + +6 x 2.5'' NVMe |
RAID | RAID 0, 1, 10, 5, 50, 6, or 60 |
Network | 8 100GE 1 OCP NIC 3.0 standard card, supporting 2 25GE |
PCIe Expansion | Up to 6 PCIe 3.0 x8 slots |
PSUs | 4 hot-swappable 3 kW AC PSUs, supporting 2+2 redundancy |
Power Supply | • 200–240 V AC • 240 V DC |
Power Consumption | Maximum: 5.6 kW |
Cooling Mode | Air cooling |
Fan Modules | 8 hot-swappable fan modules, supporting N + 1 redundancy |
Temperature | • Operating: 5℃ to 35℃ • Storage: –40°C to +70°C |
Dimensions (H x W x D) | 790 mm x 175 mm x 447 mm |
Resources