NVIDIA GH200 Grace Hopper Superchip Infrastructure & Deployment_
Grace CPU and Hopper GPU share a coherent memory architecture through NVLink-C2C. Confirm the exact GH200 server variant and its rack requirements before planning physical installation.
What Is the NVIDIA GH200?_
The NVIDIA GH200 Grace Hopper Superchip combines a 72-core Arm-based Grace CPU with a Hopper GPU through a 900 GB/s NVLink-C2C interface. The 96 GB HBM3 variant and later HBM3e systems must be identified separately; memory and OEM server requirements are not interchangeable.
Coherent CPU/GPU memory lets supported software access data beyond GPU memory alone. Physical installation still depends on the OEM rack elevation, rails, cable plan and site approvals. Leviathan can scope assembly, cabling and documented physical checks; liquid-side, facility electrical and software work are outside that scope.
Technical Specifications_
| Specification | GH200 |
|---|---|
| Architecture | Grace Hopper Superchip |
| GPU Memory | 96 GB HBM3 variant; check exact GH200 SKU |
| CPU | 72-core ARM Grace CPU |
| CPU-GPU Link | NVLink-C2C (900 GB/s) |
| Interconnect | NVLink 4.0, InfiniBand NDR |
| Networking | OEM-selected Ethernet or InfiniBand adapters |
| Cooling | Exact OEM cooling configuration |
| Key Feature | Unified CPU+GPU memory space |
Related Resources
Discuss Your Infrastructure Project_
Tell us about your project, and our team will follow up to discuss the scope, requirements and availability.
Discuss a ProjectEmail Project Scope