This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>
This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>
Atlas 800 AI Training Server（Model：9000）
The Atlas 800 AI training server (model: 9000) is based on Huawei Kunpeng 920 and Ascend 910 processors. It features the industry's highest computing density, ultra-high energy efficiency, and high network bandwidth. The server is widely used in deep learning model training, and is an ideal option for computing power-demanding industries, such as Smart City, intelligent healthcare, astronomical exploration, and oil exploration.
• 2 PFLOPS FP16 computing power in a 4U chassis
• 2.5x the computing density of industry counterparts
• Air/Liquid cooling for a single server
• Ultra-high energy efficiency ratio of 2 PFLOPS/5.5 kW, 1.8x of the industry average
• 8 x 100G RoCE v2 high-speed interfaces, doubling the industry average bandwidth
• Cross-server chip interconnection latency slashed by 10% to 70%
| Parameters | Atlas 800 AI Training Server Model: 9000 |
| Form Factor | 4U 4-socket AI server |
| Processor | 4 Kunpeng 920 processors |
| Memory | • Up to 32 DDR4 DIMMs, supporting RDIMM • Maximum memory speed: 2933 MT/s • 32 GB or 64 GB per DIMM |
| AI Processor | 8 Ascend 910 |
| HBM | 32 GB, 1200 Gbit/s |
| NPU Memory | 16 GB, total bandwidth 38.4 Gbit/s |
| AI Computing Power | 2 PFLOPS FP16 |
| Local Storage | • 2 x 2.5'' SAS/SATA HDDs or SSDs + 3 x 2.5'' NVMe SSDs • 2 x 2.5'' SATA HDDs or SSDs + 3 x 2.5'' NVMe SSDs • 2 x 2.5'' SAS/SATA HDDs or SSDs + 6 x 2.5'' NVMe SSDs • 2 x 2.5'' SATA HDDs or SSDs + 6 x 2.5'' NVMe SSDs • 2 x 2.5'' SATA HDDs or SSDs + 8 x 2.5'' SAS/SATA HDDs or SSDs |
| RAID | RAID 0, 1, 10, 5, 50, 6, and 60 |
| Network | 8 x 100GE + 4 x 25GE/2 x 100GE |
| PCIe Expansion | Up to 2 PCIe 4.0 slots |
| PSU | 4 hot-swappable 3 kW AC PSUs in 2+2 redundancy mode |
| Power Supply | • 200 V to 240 V AC • 240 V DC |
| Power Consumption | Maximum: 5.5 kW |
| Cooling Mode | Air/Liquid cooling |
| Fan Module | 8 hot-swappable fan modules in N+1 redundancy mode |
| Temperature | • Operating temperature: 5°C to 35°C (41°F to 95°F) • Storage temperature: -40°C to +70°C (40°F to +158°F) |
| Dimensions | 175 mm x 447 mm x 790 mm |
| Codec Capability | 32 built-in hardware decoders decode up to 16384 1080p images per second, which can be processed in parallel with training. |
If you are already a Partner, please click here to get more marketing resources.
Click here to visit the Partner zone to check enquiry status, manage orders, get support, or learn more about Huawei Partners.
