Lenovo ThinkSystem SD650 V4 Neptune Liquid-Cooled Server

The Lenovo ThinkSystem SD650 V4 Neptune Liquid-Cooled Server is a half-width, double-wide compute tray engineered for ultra-high-density data center deployments where thermal efficiency and compute density are paramount. Housed within the Lenovo Neptune DW612S enclosure—which accommodates up to six SD650 V4 trays—the system leverages direct water cooling (DWC) via warm-water loops to transfer heat directly off processors, memory, and key components without relying on conventional air cooling. This architecture allows facility inlet water temperatures of up to 45°C, dramatically reducing or even eliminating the need for mechanical refrigeration and lowering overall Power Usage Effectiveness (PUE) significantly compared to air-cooled alternatives. The SD650 V4 supports dual Intel Xeon 6 processors with P-cores or E-cores, delivering scalable performance across floating-point-intensive simulation workloads and throughput-oriented AI inference tasks within the same platform family.

Each SD650 V4 tray provides an extensive memory subsystem with up to 32 DDR5 DIMM slots supporting DDR5-6400 speeds and capacities up to 8 TB per tray using high-capacity 3DS RDIMMs. The platform integrates PCIe 5.0 connectivity throughout, supporting high-bandwidth NVMe storage via dual M.2 boot drives and direct GPU or accelerator attachment through the enclosure's expansion infrastructure. Network connectivity is delivered via onboard OCP 3.0 slots accommodating 25GbE, 100GbE, or InfiniBand HDR/NDR adapters, ensuring the tray integrates seamlessly into high-performance fabric topologies. Lenovo's XClarity Controller 3 (XCC3) provides out-of-band management with IPMI 2.0, Redfish API, and advanced telemetry collection for per-tray power and thermal monitoring.

The SD650 V4 Neptune is purpose-built for organizations constructing energy-efficient HPC clusters, large-scale AI training and inference farms, financial modeling environments, and computational fluid dynamics installations where rack density, total energy consumption, and sustained compute throughput are critical KPIs. The Neptune liquid cooling system enables sustained all-core turbo frequencies that air-cooled systems cannot maintain under heavy sustained loads, translating directly into improved time-to-solution for batch compute jobs. Lenovo's Open Cloud Automation (OCA) toolchain and integration with workload managers such as IBM Spectrum LSF, Altair PBS Professional, and SLURM further streamline cluster provisioning and lifecycle management at scale.

Ideal for

• High-performance computing (HPC) clusters running weather modeling, molecular dynamics, and computational fluid dynamics requiring sustained multi-node floating-point throughput
• Large-scale AI inference farms deploying dense accelerator-attached trays for low-latency model serving at hyperscale
• Energy research and national laboratory environments mandating maximum compute-per-watt efficiency with warm-water facility cooling integration
• Financial services quantitative analytics platforms executing Monte Carlo simulations and risk calculations across thousands of parallel threads
• Genomics and life sciences pipelines processing whole-genome sequencing and protein folding workloads that demand high memory bandwidth and capacity
• Colocation and enterprise data centers transitioning to liquid-cooled infrastructure to meet sustainability mandates and reduce cooling overhead OpEx

Technical specifications

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