Moore Threads · Compute Server
MCCX D800 X1
(Translation draft; no standalone English page) The MCCX D800 X1 is a server purpose-built to unleash the training and inference performance of the MTT S4000. Based on latest-generation Intel Xeon Scalable processors with DDR5 memory and PCIe Gen5, upgraded with 200G InfiniBand/Ethernet networking, it powers S4000 thousand-card LLM clusters under MTLink and the MCC Platform cluster software.
Key Highlights
- (Translation draft) 4U server
- 8× MTT S4000 (PCIe Gen5, 48GB each)
- 1TB DDR5 memory
- MT-Link 1.0 plus PCIe Gen5 P2P
- 200G InfiniBand/Ethernet adapters
- four 2400W platinum redundant PSUs
- rated power 6000W
- standard rails.
Technical Specifications
| 形态/Form Factor | 4U server |
|---|---|
| CPU | 2× Intel Xeon Gold 6430 (2.1GHz/32C) |
| GPU | 8× MTT S4000 (PCIe Gen5, 48GB) |
| 内存/Memory | 1TB (16×64GB DDR5-4800 RDIMM) |
| 系统盘/Boot | 2× 480GB SATA SSD |
| 数据盘/Data | 4× 3.84TB Gen4 NVMe SSD |
| 互联/Interconnect | MT-Link 1.0 + PCIe Gen5 P2P |
| 网卡/Network | 2× single-port 200G IB NDR/Eth CX7 + 2× dual-port 25Gb optical |
| 电源/PSU | 4× 2400W hot-swap N+M platinum |
| 额定功率/Rated Power | 6000W |
Typical Use Cases
(Translation draft) Turnkey LLM training/fine-tuning and inference appliancebuilding block of thousand-card LLM clustersrecommendation systems, data analytics and NLP workloads.
英文为翻译稿，官网无独立英文页；规格数字来自官网中文 D800 页面规格表；图片为官网 related/D800.jpg 官方配图（d800/s4000.jpg 为 S4000 单卡图、主体不符，未采用）。
Related Products
Moore Threads
MTT S5000
Universal GPU for AI Training and Inference View details →
Moore Threads
MTT S4000
Large-model AI inference and training GPU with Tensor Cores View details →
Moore Threads
MTT S3000
Full-featured server-class GPU View details →
Moore Threads