SERVER
8U Flagship AI Server
R985
The R985 is a high-end 8U rack-mounted server based on 4th/5th generation Intel® Xeon® Scalable processors. Equipped with NVLink professional GPU AI acceleration cards, it effectively meets the demands of ultra-high-density AI computing applications and is well-suited for large-scale cluster deployments, delivering stable and reliable performance for intensive AI workloads.
Products Specification
| Model | R985 |
| Processor | Supports 2*4th/5th Gen Intel Xeon® Scalable Processors, up to 350W |
| Memory | Up to 32*DDR5 slots, up to 5600MHz |
| Storage Controller | The onboard RAID controller features cache supercapacitor protection, supporting RAID level migration, disk roaming, self-diagnosis, web remote management and RAID secure boot |
| Storage | Supports up to 12*3.5 inchesSupports 2.5-inch U.2 hard drives; supports 4 or 8 hard drivesBuilt-in: 2*SATA M.2 |
| Network Interfaces | 1*OCP3.0 network card with NCSIOptional OCP 3.0 interfaces: 2*GE, 2*10GE, 2*25GE, 2*100GE, 2*200GE |
| PCIe Expansion | 9*PCIe 5.0 slots + 1*OCP slot |
| GPU | Supports 8*NVLink GPU |
| Additional Interfaces | Rear: 1*RJ-45 management port, 2*USB 3.0, 1*VGAFront: 2*USB 3.0+1*VGA |
| Power Supply | Optional 3300W 54V rear in and out power supply module, support 3+3 redundancySupports 2700W CRPS Platinum PSU (hot-swappable, 1+1 redundant)Voltage: 100~240V AC (refer to nameplate) |
| OS Support | Windows Server, RHEL, SLES, CentOS, Ubuntu, VMware ESXi, etc. |
| Operating Temperature | 5°C ~ 35°C |
Products View
Products Features
- Ultimate Performance and Flexible Scalability
- Powered by 4th/5th generation Intel® Xeon® Scalable processors, equipped to support up to 8*double-slot, full-height, and full-length professional GPU AI accelerator cards, it combines powerful hardware configuration with flexible computing matching, tailored specifically to meet the demands of ultra-high-density AI computing applications.
- Streamlined Operations and Intelligent Management
- It features streamlined operations and intelligent management capabilities to reduce operational complexity and improve maintenance efficiency. The out-of-band visualization management enables remote and precise location tracking of physical devices, further optimizing operational efficiency. It supports automatic crash logging, with online viewing and download capabilities for crash logs, enabling precise fault diagnosis and reducing troubleshooting time. Additionally, it provides real-time health monitoring and reporting for critical components, delivering comprehensive intelligent services throughout the entire product lifecycle.
- Extremely Quiet and Energy-Efficient
- It features extremely quiet operation and excellent energy efficiency, balancing high performance with low energy consumption and low noise pollution. It supports high-performance online inference with 8*GPUs, realizing seamless interconnection of 8*professional-grade GPU cards to ensure stable and efficient data transmission. Meanwhile, it optimizes GPU Direct efficiency specifically for deep learning training scenarios, effectively reducing data transmission latency and enhancing the overall training efficiency of deep learning models.