Cornelis Networks Announces The Upcoming Release Of The Industry’s Highest Performing HPC Interconnect: Omni-Path Express

Cornelis Networks, a technology leader delivering purpose-built, high-performance fabrics for High Performance Computing (HPC), High Performance Data Analytics (HPDA), and Artificial Intelligence (AI), today announced the upcoming release of a new fabric solution, Cornelis Omni-Path Express, with availability beginning in the fourth quarter of this year.

“The company’s solutions have long enabled end-users to efficiently address their most challenging application needs associated with modeling and simulation, data analytics, and deep learning, and the company’s new offering is no exception,” said Phil Murphy, CEO, Cornelis Networks. “Cornelis Omni-Path Express, available later this year, cost-effectively delivers industry-leading network performance at the lowest software footprint through the optimized partitioning of network functions. Enabled by a hardware infrastructure well-matched to the needs of OpenFabrics Interfaces (OFI), this new solution provides end-users a wide spectrum of optimized capabilities.”

Omni-Path Express is powered by new, highly optimized host software supporting OFI, and is compatible with all existing Omni-Path deployments. OFI is developed under the auspices of the OpenFabrics Alliance and is focused on providing modern applications with a rich set of fabric capabilities. Cornelis Networks has worked closely with leading CPU and GPU manufacturers, as well as leading system solution providers, to ensure broad applicability and availability of its Omni-Path Express solution.

“The fields of analytics and machine learning have continued to drive the need for high-performance systems. Our surveys reveal that over 80% of HPC users have already adopted AI into their HPC environments or are actively working towards it within the next year,” said Addison Snell, CEO, Intersect360 Research. “High-performance fabric interconnects are an essential element of the HPC ecosystem. A few years ago, InfiniBand was the de facto open standard for scalable, low-latency systems. Cornelis’ investment in Omni-Path, combined with the NVIDIA acquisition of Mellanox, suddenly flips the conversation. Now it is InfiniBand that can potentially be viewed as proprietary to a single processor vendor, whereas Omni-Path Express is an open, multi-vendor solution.”

Omni-Path Express supports the market’s need for exceptional application performance in traditional simulation and modeling environments, but it also addresses the new challenges stemming from the convergence of HPC, HPDA, and AI infrastructures. The benefits of Omni-Path Express include:

  • Increased application-level performance through industry-leading fabric performance
  • Support for a wide variety of application-critical technologies

    ○        All popular MPI implementations

    ○        Multiple AI frameworks

    ○        Advanced object storage file systems

    ○        Portable GPU acceleration solutions
  • An aggressive pricing model that enhances performance-per-dollar

“For the past two decades NHR@ZIB has supported 200 research institutions and universities in their unwavering quest to understand processes in nature at the atomistic level, turbulence, climate change, and breakthroughs in drug design, among other discoveries. This places stringent requirements on the HPC systems we procure that drive low latency and high message rate – the foundations for application performance and scalability,” said Dr. Thomas Steinke, Head of Supercomputing, Zuse Institute, Berlin. “Having adopted Cornelis’ Omni-Path interconnect fabrics for several years has enabled us to support both traditional computational and emerging AI workloads. NHR@ZIB is excited to test drive the new Cornelis Omni-Path Express that promises dramatic performance improvements.”

In conjunction with its Omni-Path Express announcement, Cornelis Networks is also announcing the planned release of its 400 Gbps next generation fabric solution, available late in 2022. The new high performance scale-out interconnect leverages industry advancements in software infrastructure, process technology, advanced packaging, serial communication, and optical interfaces to cost-effectively meet the needs of the most demanding next-generation application environments. In addition to an optimal host interface, the efficiency of the switching infrastructure is crucial in avoiding congestion while routing packets to their destination. By supporting advanced topologies and their associated routing algorithms, dynamic adaptive routing, and state-of-the-art congestion control, Cornelis Networks furthers its leadership position in delivering performance and scalability for its users.

Application needs are evolving more quickly than ever with new algorithms combining techniques from modeling and simulation, data analytics, and deep learning, necessitating an interconnect with very low latency, very high message rate, and minimal software footprint. Cornelis Networks is addressing the needs of the most challenging environments now with its imminent release of Omni-Path Express and will continue its fabric leadership with the release of its next-generation 400 Gbps solution in late 2022.

Roboticulized digests/handpicks the latest news about the artificial intelligence/machine learning industry and serves them to you daily. We provide you with the latest news and press releases straight from the AI/ML industry.