WAYNE, PA — Cornelis has completed qualification and integration of its CN5000 Omni-Path networking fabric across Lenovo’s ThinkSystem V3 and V4 servers, establishing the high-performance interconnect as a fully supported option within Lenovo’s EveryScale Solution portfolio.
The collaboration targets one of the most persistent challenges in large-scale AI and high-performance computing (HPC): network congestion that blunts the impact of accelerated compute. By incorporating the CN5000 fabric, Lenovo customers can deploy tightly integrated AI and HPC systems with predictable, scalable performance and dramatically reduced latency.
The integration spans a range of Lenovo platforms, including air- and liquid-cooled ThinkSystem V4 servers powered by Intel Xeon 6 processors and ThinkSystem V3 systems equipped with AMD EPYC 9004 series CPUs and Intel’s 4th and 5th generation Xeon Scalable processors. These configurations support dense, accelerator-heavy deployments where communication efficiency is critical.
Scott Tease, Vice President and General Manager for AI and HPC at Lenovo, said the rapid shift toward large, distributed compute clusters demands a networking layer that can keep pace. He noted that pairing Lenovo servers with the Cornelis fabric removes communication bottlenecks, improves system utilization, and accelerates time-to-insight across workloads ranging from climate modeling and CFD to seismic analysis and scientific AI.
Cornelis’ CN5000 portfolio delivers 400 Gbps end-to-end throughput via a PCIe 5.0 SuperNIC and a 48-port non-blocking edge switch. Built for the communication patterns of massive training runs and complex simulations, the solution supports lossless transport, congestion awareness, adaptive routing, per-flow credit-based control, and in-network telemetry. An open-source software stack ensures interoperability across diverse architectures and cluster designs.
Lisa Spelman, CEO of Cornelis, said the integration represents a milestone for scalable accelerated computing. She emphasized that organizations are increasingly seeking fabrics that match the capabilities of modern compute hardware, adding that the CN5000’s design—combined with Lenovo’s energy-efficient platforms—unlocks the full performance potential of AI and HPC systems.
Both companies played a role in Lenovo’s EveryScale 25B 2.1 “Best Recipe,” which incorporates the CN5000 SuperNIC, switch, cabling, and storage components. GPU-accelerated configurations within ThinkSystem V3 and V4 machines are also fully validated.
Cornelis (booth #2009) and Lenovo (booth #3814) are showcasing the jointly qualified CN5000 solution this week at SuperComputing 2025 in St. Louis.
For the latest news on everything happening in Chester County and the surrounding area, be sure to follow MyChesCo on Google News and MSN.