Brand: HPE | Category: GPUs
SKU: R9H94A | Part #: R9H94A | MPN: R9H94A
Contact for Pricing — Request a Quote
The HPE NVIDIA GH200 Grace Hopper Superchip Compute Module (part number R9H94A) is an accelerator compute module that combines an ARM Grace CPU with an NVIDIA Hopper architecture GPU. The two are linked by NVIDIA NVLink-C2C, and the module is built as a compute module for HPE Cray systems.
The Hopper GPU component has 96 GB of HBM3 high bandwidth memory with 4.8 TB/s of memory bandwidth and approximately 67 TFLOPS of peak FP32 performance. Power consumption is approximately 500 W, and the specified operating temperature range is 0–50°C.
The module is aimed at AI, machine learning and high-performance computing workloads. It suits HPE Cray system deployments in data centre and research computing environments where large memory bandwidth and close CPU-GPU coupling support compute-intensive applications.
Send the part number through our RFQ form and our team will reply within 48 hours.
| Brand | HPE |
| Category | GPUs |
| SKU | R9H94A |
| Part Number | R9H94A |
| Condition | New |
| Manufacturer Part Number | R9H94A |
| Product Type | Accelerator Compute Module |
| GPU Architecture | NVIDIA Hopper |
| CPU Architecture | ARM Grace |
| Peak FP32 Performance | Approximately 67 TFLOPS (Hopper GPU component) |
| Memory Type | HBM3 (High Bandwidth Memory) |
| GPU Memory | 96 GB HBM3 |
| Memory Bandwidth | 4.8 TB/s |
| Form Factor | Compute Module for HPE Cray Systems |
| Interconnect | NVIDIA NVLink-C2C |
| Target Workloads | AI, machine learning, high-performance computing |
| Power Consumption | Approximately 500 W |
| Operating Temperature Range | 0–50°C |
The HPE NVIDIA GH200 Grace Hopper Superchip Compute Module accelerates AI/ML training, inference, scientific HPC, and virtualization (vGPU) workloads. Typical deployments include LLM training clusters, computer-vision pipelines, financial risk modeling, and rendering farms.
Key specifications for the HPE NVIDIA GH200 Grace Hopper Superchip Compute Module: new condition; manufacturer part number R9H94A; product type Accelerator Compute Module; gpu architecture NVIDIA Hopper; cpu architecture ARM Grace; peak fp32 performance Approximately 67 TFLOPS (Hopper GPU component); memory type HBM3 (High Bandwidth Memory). Manufacturer part number R9H94A. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.
The HPE NVIDIA GH200 Grace Hopper Superchip Compute Module requires a PCIe Gen4 or Gen5 x16 slot, server power adequate for the card's TDP, and CUDA/ROCm driver support in your hypervisor or bare-metal OS. Sales engineering will confirm chassis fit (1U/2U/4U), PCIe lane count, and PSU headroom before quoting.
