Lenovo ThinkSystem SR655 V3 GPU-Ready Server

The Lenovo ThinkSystem SR655 V3 (7D9EA01GNA) is a 2U single-socket rack server built on the AMD EPYC 9004 series (Genoa) processor platform, delivering exceptional core density and memory bandwidth for compute-intensive enterprise workloads. The architecture supports up to 96 cores per socket, PCIe 5.0 connectivity, and CXL 1.1 memory expansion, making it well-suited for modern AI inference pipelines, high-throughput data analytics, and virtualized infrastructure that demands both raw compute power and flexible I/O.

The SR655 V3 is engineered for GPU-accelerated computing, accommodating multiple NVIDIA or AMD GPUs in a 2U form factor through its flexible PCIe expansion design. Up to 12 DIMM slots support DDR5 memory, enabling high-bandwidth data movement required by machine learning inference engines and real-time analytics applications. The server's Lenovo XClarity Controller 2 (XCC2) provides out-of-band management with IPMI 2.0 and Redfish API support, integrating cleanly into enterprise automation frameworks including Lenovo XClarity Administrator.

Targeted at organizations deploying AI at the edge, consolidating virtualized workloads, or running memory-intensive databases, the SR655 V3 balances thermal efficiency with computational density. Its 2U chassis offers front-accessible drive bays supporting NVMe, SAS, and SATA configurations, while the single-socket design reduces software licensing overhead compared to dual-socket alternatives—a meaningful operational consideration for per-core or per-socket licensed workloads in enterprise environments.

Ideal for

• AI and machine learning inference at the enterprise edge, leveraging GPU acceleration for real-time model scoring
• Edge analytics for manufacturing, retail, or telecommunications environments requiring high core count in compact 2U deployments
• Virtualized infrastructure consolidation using VMware vSphere or Red Hat OpenShift on AMD EPYC's high core density
• High-performance database deployments such as SAP HANA, PostgreSQL, or in-memory caching requiring large DDR5 memory footprints
• Media transcoding and video analytics workloads benefiting from GPU offload and PCIe 5.0 bandwidth
• HPC departmental clusters and simulation workloads where single-socket licensing economics reduce total software expenditure

Technical specifications

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