HPE Cray XD670 (8x NVIDIA H200 SXM, Dual 5th Gen Intel Xeon)
Flagship 8-GPU node with 1.1TB HBM3e for frontier LLM training.
We help you choose, configure, and deliver the right system — no obligation.
Configuration at a Glance
Tailored per engagement. Full technical overview below.
Configuration Options
Core specifications for this system. Every component is configurable to your workload — request a quote for a tailored build.
8x NVIDIA H200 SXM, 141GB HBM3e each (1.1TB total)
Dual 5th Gen Intel Xeon Scalable, up to 64 cores each
Up to 32x DDR5-5600 DIMMs (8-channel per socket)
Up to 8x SFF NVMe U.2/U.3 + 2x M.2 RAID
Overview
The HPE Cray XD670 in this configuration pairs eight NVIDIA H200 SXM GPUs on an NVLink-connected HGX baseboard with dual 5th Gen Intel Xeon processors in a single 5U node. Nexus Compute delivers it specified, integrated, and burn-in tested with direct liquid cooling, backed by HPE warranty and sourced through authorized channels.
Who This Solution Is For
Business Benefits
Frontier-scale training
Eight NVLink-coupled H200 GPUs with 1.1TB aggregate HBM3e shorten time-to-train for the largest models.
Bigger models per node
141GB per GPU lets larger context windows and parameter counts fit before sharding across nodes.
Deployment-ready integration
Each node arrives configured, liquid-cooled, and tested so your team racks and trains rather than builds.
Typical Business Use Cases
Large language model pre-training and fine-tuning
Multimodal and generative AI training
Mixed AI-plus-HPC simulation pipelines
Foundation-model research clusters
Industry Applications
Technical Overview
Built on the NVIDIA HGX H200 8-GPU baseboard with fourth-generation NVLink and NVSwitch, the node provides eight PCIe Gen5 fabric slots for InfiniBand NDR, Slingshot 11, or Ethernet alongside dual 5th Gen Intel Xeon Scalable CPUs. The 5U chassis supports direct liquid cooling to sustain the 700W GPU TDP under continuous training load.
| GPU / Accelerator | 8x NVIDIA H200 SXM, 141GB HBM3e each (1.1TB total) |
| GPU Interconnect | NVIDIA NVLink + NVSwitch (HGX baseboard) |
| CPU | Dual 5th Gen Intel Xeon Scalable, up to 64 cores each |
| Memory | Up to 32x DDR5-5600 DIMMs (8-channel per socket) |
| Storage | Up to 8x SFF NVMe U.2/U.3 + 2x M.2 RAID |
| Networking / Fabric | 8x PCIe Gen5 slots: InfiniBand NDR / Slingshot 11 / Ethernet |
| Form Factor | 5U single dual-CPU node |
| Cooling | Direct liquid cooling (air-cooled option) |
| Warranty | HPE support, configurable terms |
Specifications are indicative and configured to each engagement. Request a quote for a configuration tailored to your requirements.
Warranty, Support & Fulfillment
Every system ships from an authorized channel, configured and tested, with the documentation enterprise buyers need — backed by warranty and a dedicated account team.
Enterprise Warranty
Full manufacturer warranty with optional on-site, next-business-day support and extended coverage.
Authorized Channel
Sourced through Tier-1 distribution and OEM partners — never grey market. Asset & warranty records included.
Lead Time & Deployment
48-hour quotes, then configured, burn-in tested, and delivered on a committed schedule.
Nationwide Fulfillment
Coordinated logistics, rack-and-stack, and delivery wherever your infrastructure lives.
Frequently Asked Questions
How many XD670 H200 nodes do we need for our model size?
Sizing depends on parameter count, sequence length, and target throughput; a single node handles many fine-tuning and mid-scale training jobs, while pre-training large models scales across NVLink nodes joined by InfiniBand NDR. We size the node and fabric count to your workload.
Why choose H200 over H100 here?
H200 raises per-GPU memory to 141GB HBM3e with higher bandwidth, so larger models and contexts fit per node and memory-bound training and inference run faster. Choose it when capacity and bandwidth, not just compute, gate your work.
What facility requirements come with direct liquid cooling?
DLC nodes connect to facility water or a CDU and need appropriate flow, temperature, and rack power provisioning. We confirm thermal and power requirements as part of the configuration.
Hardware Assistance
Configure the HPE Cray XD670 (8x NVIDIA H200 SXM, Dual 5th Gen Intel Xeon) with Nexus Compute
Tell us your requirements and a hardware specialist will help you specify, configure, and quote the right system — typically within two business days. No obligation.