Cornelis Networks has achieved a new industry benchmark for Ansys Fluent performance, powered by AMD EPYC 9755 processors and its CN5000 Omni-Path interconnect. The system delivered a record-breaking score of 1704 on the Ansys Fluent f1_racecar_140m computational fluid dynamics (CFD) benchmark—more than double the previous record of 830—using 1024 CPU cores across an eight-node cluster. This milestone highlights the performance gains possible when high-speed networking is optimized for compute-intensive engineering simulations. The CN5000, featuring a 400Gbps PCIe 5.0 SuperNIC and a 48-port full non-blocking edge switch, achieved nearly 99% strong-scaling efficiency, enabling engineers to run twice as many design studies in the same timeframe. Lisa Spelman, CEO of Cornelis, emphasized the impact of eliminating network bottlenecks. “When the network doesn’t slow you down, innovation accelerates,” she said. “This collaboration with AMD and Synopsys empowers engineers in automotive, aerospace, and manufacturing to explore more ideas, test faster, and bring better products to market.” Robert Hormuth, Corporate Vice President of Architecture and Strategy at AMD, noted that the EPYC 9755 processors with Zen 5 architecture are built for scalable, high-throughput workloads. “This result proves that EPYC platforms deliver the consistent performance engineers need to push the boundaries of simulation,” he said. For Synopsys, which acquired Ansys Fluent, the achievement underscores the value of open, standards-based collaboration. Wim Slagter, Senior Director of Partner Programs at Synopsys, said the CN5000 enables “near-linear scaling across multi-node systems,” allowing teams to run higher-fidelity models and make data-driven decisions faster. “This means better products, faster time-to-insight, and reduced risk of costly design flaws,” he added. The f1_racecar_140m benchmark is one of the most demanding public CFD tests, widely used in automotive and aerospace industries to evaluate simulation efficiency. The new record demonstrates the effectiveness of Cornelis’ congestion-free, lossless networking architecture when paired with AMD’s latest processors. This advancement is critical for industries where simulation speed directly impacts safety, reliability, and cost. Faster simulations mean more iterations, improved designs, and accelerated product development—all within budget and schedule constraints. The CN5000 portfolio is now the preferred interconnect for AMD EPYC-based clusters in high-performance computing environments requiring both computational power and seamless communication across nodes. Cornelis Networks continues to drive innovation in AI and HPC by delivering scalable, efficient networking solutions that maximize compute utilization and reduce time-to-insight across diverse sectors.