HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration)
Brand: HPE | Category: GPUs
SKU: HPE-R9W12A | Part #: R9W12A | MPN: R9W12A
Contact for Pricing — Request a Quote
About the HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration)
The HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration) from HPE is enterprise-grade GPUs hardware built for data centers. It features Rack-Mount form factor, PCIe Gen5 interface. The HPE Cray XD680 in the NVIDIA Blackwell GB200 NVL72 configuration is a purpose-built, liquid-cooled AI supercomputing node designed to deliver extreme-scale AI training and inference performance within HPE's Cray XD server family. At its core, the system integrates NVIDIA's GB200 NVL72 architecture, which combines 36 Grace CPU modules and 72 Blackwell B200 GPU dies interconnected via fifth-generation NVLink at 1.8 TB/s aggregate GPU-to-GPU bandwidth, forming a single coher Available in brand new condition. HPE gpus products are trusted by enterprises and data centers worldwide. Available from Omnixon Global. Contact our sales team to request a quote.
Technical Specifications
| Brand | HPE |
| Category | GPUs |
| SKU | HPE-R9W12A |
| Part Number | R9W12A |
| Condition | New |
| Form Factor | Rack-Mount |
| Interface | PCIe Gen5 |
| Manufacturer Part Number | R9W12A |
| Product Line | HPE Cray XD Series |
| Model | HPE Cray XD680 |
| GPU Architecture | NVIDIA Blackwell |
| GPU Configuration | GB200 NVL72 (72x Blackwell B200 GPU dies + 36x Grace CPU modules) |
| Total GPUs per NVL72 Module | 72 Blackwell B200 GPU dies |
| Total CPUs per NVL72 Module | 36 NVIDIA Grace (Arm Neoverse V2) |
| GPU Interconnect | NVLink 5th Generation, 1.8 TB/s aggregate GPU-to-GPU bandwidth |
| CPU-GPU Interconnect | NVLink-C2C (chip-to-chip), 900 GB/s per Grace-Blackwell pair |
| Total HBM Memory (NVL72) | 720 GB HBM3e |
| AI Compute Performance (FP4) | Up to 1.4 ExaFLOPS (NVL72 rack scale) |
| CPU Memory Type | LPDDR5X (per Grace module) |
| Cooling | Direct Liquid Cooling (DLC) |
| Network Fabric Support | HPE Slingshot 200Gb/s high-speed interconnect |
| Management | HPE iLO |
| Form Factor | Rack-scale module (fits HPE Cray XD rack infrastructure) |
| Operating System Support | Linux (RHEL, Ubuntu); NVIDIA AI Enterprise stack |
| CUDA Support | CUDA 12.x and later |
Frequently Asked Questions about HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration)
What does the HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration) do?
The HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration) 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.
What are the headline specs of the HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration)?
Key specifications for the HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration): form factor Rack-Mount; interface PCIe Gen5; 3 years support; new condition; support 3 years standard (manufacturer); gpu support NVIDIA GB200; ai optimized Yes. Manufacturer part number R9W12A. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.
Is the HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration) compatible with my infrastructure?
The HPE Cray XD680 (NVIDIA Blackwell GB200 NVL72 Configuration) 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.