AI Computing Servers
10-GPU High-Density Inference Server (Thor T5000)
A high-density dual-socket rack inference platform built for GPU density and low latency: up to ten dual-die Thor T5000 GPUs, ten PCIe 5.0 x16 dual-slot bays, and four CRPS PSUs in 2+2 / 3+1 redundancy.
10×GPU10× T5000High-density rackUp to 10 dual-die T5000 · 10× PCIe 5.0 x16 · 4× CRPS
Applications:LLM InferenceHigh-Concurrency API ServingEnterprise Agents & Knowledge Bases
Specifications
| Form Factor | High-density dual-socket rack |
|---|---|
| Processors | 2× Intel Xeon 4th/5th Gen |
| GPU Expansion | Up to 10 dual-die Thor T5000 GPUs |
| PCIe Slots | 10× PCIe 5.0 x16 dual-slot + 1× PCIe 5.0 x16 single-slot |
| Storage | 12 front 3.5/2.5-inch SAS/SATA bays; optional 4× U.2 |
| Power Supply | 4× standard CRPS PSUs with 2+2 / 3+1 redundancy |
| Management & I/O | Modular I/O: LAN, USB, IPMI, VGA, COM, UID |
| Positioning | High-performance, high-density dual-socket inference system |
* Configurations are indicative; final specs follow the project BOM, delivery specification and measured results.
Special Functions
Redundant CRPS PSUsSupported
BMC / IPMI remote managementSupported
Optional 4× U.2Supported
CPU/GPU cold-plate liquid coolingNot supported
Product Features
GPU topology, PCIe 5.0, power and airflow engineering
BIOS/BMC, driver, CUDA and firmware compatibility validation
Burn-in, NCCL / GPUDirect RDMA and multi-node stress testing
Tiered system/cache/data drives from our enterprise SSD and DDR5 RDIMM lines
Design, integration testing, on-site deployment and O&M in one delivery
Interested in 10-GPU · Thor T5000?
Contact us for the full datasheet, sample testing, and volume pricing.
Contact Us