TL;DR:
- DriveNets joins Ultra Ethernet Consortium for open AI networking solutions.
- DriveNets Network Cloud-AI offers high-performance AI networking.
- DriveNets leads in Ethernet-based AI networking performance.
- Testing by Scala Computing validates DriveNets’ exceptional results.
- DriveNets contributes to Open Compute Project (OCP) standards.
- This reinforces DriveNets’ role in advancing high-performance Ethernet solutions.
Main AI News:
DriveNets, a trailblazer in cutting-edge networking solutions, has taken a significant stride by joining the Ultra Ethernet Consortium (UEC). This move underscores DriveNets’ commitment to shaping an open, interoperable Ethernet architecture, catering to the demands of high-performance networking for AI and HPC workloads. The inclusion of DriveNets within the consortium underscores the company’s position as a trailblazer and innovator in AI networking solutions.
The DriveNets Network Cloud-AI, introduced in May and currently available for order alongside Broadcom Jericho 3-AI based white boxes, stands as the premier AI networking solution in the market today. This achievement is based on rigorous trials conducted with leading hyperscalers and independent testing conducted by Scala Computing. Irrespective of the GPU or AI accelerator used, DriveNets Network Cloud-AI significantly enhances the performance of large-scale AI workloads, empowering users to establish open, multi-vendor AI infrastructures.
DriveNets joins the ranks of esteemed founding members of the UEC, which includes tech giants like Microsoft, Meta, Broadcom, AMD, Arista, Cisco, Oracle, HPE, Intel, and Eviden. Ido Susan, the founder and CEO of DriveNets, asserts, “AI is poised to permeate cloud and data center infrastructure, demanding technology that can effectively minimize GPU idle cycles in high-performance AI workloads.” Addressing this challenge necessitates an Ethernet-based fabric that champions open, standards-based architectures, eradicating vendor lock-in. The Ultra Ethernet Consortium is unwavering in its commitment to advancing open, interoperable Ethernet-based solutions, and DriveNets takes pride in bolstering these efforts to refine the Ethernet standard.
Notably, DriveNets Network Cloud-AI solution maintains an impressive lead, boasting 18-24 months of advancement over the rest of the AI networking industry in terms of performance. This is achieved through the implementation of a high-scale, cell-based fabric with cell-based load balancing and end-to-end scheduling. Leveraging the highest-scale router technology, DriveNets Network Cloud-AI extends its support to AI back-end and front-end networks, offering a comprehensive solution.
Scala Computing’s rigorous testing has demonstrated the exceptional job completion time (JCT) results achieved by DriveNets Network Cloud-AI Ethernet solution, all while remaining agnostic to specific hardware configurations. This robust support for an open and flexible ecosystem of ASICs, ODMs, and optical vendors further sets DriveNets apart.
Additionally, DriveNets remains committed to contributing to the Open Compute Project (OCP) distributed disaggregated chassis (DDC) standard, which is gaining prominence as the leading open networking standard. Collaborating with the UEC, OCP seeks to enhance Ethernet performance for AI clusters and HPC. DriveNets’ involvement in the UEC, combined with the industry-leading Ethernet-based AI networking performance of Network Cloud-AI, solidifies the company’s pivotal role in ushering high-performance Ethernet solutions into the industry.
Conclusion:
DriveNets’ strategic move to join the Ultra Ethernet Consortium and its continued leadership in AI networking solutions positions the company to drive the development of open, high-performance Ethernet solutions for the market. This move emphasizes the importance of open, interoperable architectures in the growing AI and HPC networking sector, setting the stage for greater innovation and collaboration in the industry.
