Moore Threads · Compute Server
MTT SGX5000
(Translation draft; official site has no standalone English SGX5000 page) The MTT SGX5000 is an integrated AI server equipped with eight MTT S5000 OAM compute modules. Optimized for cooling, power delivery and I/O expansion with air- and liquid-cooled SKUs, it pairs high-end compute, storage and networking to run hundred-billion- and trillion-parameter models, and can ship preloaded with optimized training and inference software for out-of-box use.
Key Highlights
- (Translation draft) 8U rack server with eight S5000 OAM modules
- up to 2TB DDR5-5600 RDIMM
- eight single-port 400Gb InfiniBand HDR/Ethernet CX7 adapters
- native FP8 engine for DeepSeek/Qwen-class architectures
- air- or liquid-cooled SKUs
- preinstalled training/inference software stack available.
Technical Specifications
| 形态/Form Factor | 8U rack server |
|---|---|
| GPU | 8× S5000 OAM modules |
| CPU | 2× x86 or C86 processor |
| 内存/Memory | up to 2TB DDR5-5600 RDIMM (training config 32×64GB or 24×96GB) |
| 系统盘/Boot | 2× 960GB SATA RI SSD |
| 数据盘/Data | 4× 3.84TB PCIe Gen4 NVMe U.2 |
| GPU 网络/GPU Net | 8× 400Gb IB HDR/Ethernet CX7 |
| 存储网/Storage Net | 1× dual-port 100Gb RoCE |
| 业务网/Front Net | 1× dual-port 25Gb optical |
| 额定功耗/Rated Power | 11.4kW |
| 重量/Weight | 100kg |
Typical Use Cases
(Translation draft) Training and inference of hundred-billion/trillion-parameter LLMslarge-scale AI computing centersGPU server systems compatible with mainstream software and hardware ecosystems.
官网无独立英文产品页（en 路由回退为英文首页），英文列为翻译稿；规格数字来自官网中文规格表；图片采用官网 row-3-3.jpg（图注 MTT SGX5000 的官方整机渲染，原 related/SGX5000.jpg 为同系列低分辨率小图未采用）；页面前半部分复用 S5000 宣传文案，其中竞品对比基准未录入本机规格。
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