Lenovo ThinkSystem SD650 V3 Neptune Liquid-Cooled Node

The Lenovo ThinkSystem SD650 V3 Neptune Liquid-Cooled Node is a high-density, direct water-cooled compute node engineered for demanding HPC and AI workloads. Built on the Intel Xeon Scalable (4th Gen, codenamed Sapphire Rapids) processor platform, the SD650 V3 uses Lenovo's Neptune direct water cooling technology to transfer heat directly away from processors and other critical components via a rear-door heat exchanger and cold-plate assembly, enabling sustained high-performance operation at significantly reduced energy consumption compared to air-cooled equivalents. The node is housed in a half-width, 2U tray form factor that fits within the Lenovo ThinkSystem DW612S 6U enclosure, allowing up to six nodes per enclosure for exceptional compute density per rack unit.

Each SD650 V3 node supports dual Intel Xeon processors and a high memory capacity configuration using DDR5 DIMMs across 32 DIMM slots, delivering the memory bandwidth required by in-memory analytics, simulation, and large-scale AI training workloads. The platform supports PCIe Gen 5 expansion, enabling integration with the latest high-bandwidth accelerators and networking adapters. Onboard storage options include support for NVMe drives, and the node provides flexible networking connectivity suited to high-performance fabrics including Ethernet and InfiniBand.

The ThinkSystem SD650 V3 is purpose-built for organizations running computational fluid dynamics, molecular dynamics, climate modeling, financial risk simulation, deep learning training, and other workloads where thermal density and performance-per-watt are critical design criteria. Managed through Lenovo XClarity Administrator and compatible with OpenHPC and other HPC software stacks, the SD650 V3 integrates into large-scale cluster deployments with centralized lifecycle management, firmware orchestration, and telemetry collection to support operational efficiency at scale.

Ideal for

• Large-scale HPC cluster deployments for scientific simulation including computational fluid dynamics, structural analysis, and climate modeling
• Deep learning and AI model training requiring sustained high CPU throughput paired with PCIe Gen 5 GPU accelerators
• Financial services risk modeling and quantitative analytics demanding high memory bandwidth and low-latency compute
• Genomics and life sciences workloads such as molecular dynamics simulations and genome sequencing pipelines
• Densely packed enterprise data center modernization where rack space and power budgets constrain traditional air-cooled server deployments
• Government and research laboratory supercomputing infrastructure requiring liquid-cooled high-density nodes under centralized cluster management

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.