With the advent of Gen AI, the explosion of Edge computing and high-performance computing (HPC) applications, hyperscalers are rushing to optimize datacenters for AI workloads. To connect these workloads, new hardware architecture is required for accelerated computing with very low latency and almost no tolerance for data loss. The network must provide seamless, ultra-reliable connectivity to every part of the AI infrastructure.
Hyve’s custom designed network switches leverage the Broadcom Tomahawk™ 5 chipset to provide 800Gb and 400Gb in a 2U compact form factor. It can deliver up to 51.2 Tbps switching capacity ideal for AI and HPC applications. It is built to meet the needs of the advanced AI and ML/DL training and inference applications by providing 64 ports of 800GB/400GB. Hyve’s designed switches come optimized for L2 and L3 networking at line rate with adaptive routing and dynamic load balancing capabilities in the hardware.
Hyve’s switches offer increased flexibility and modularity to meet the specific needs of hyperscalers. In addition, there is a potential costs savings by eliminating unnecessary features.
Hyve’s network switches arrive “rack-ready,” meaning that what’s tested at Hyve, whether it’s a single switch or a large cluster of switches at scale, is precisely what’s deployed. This helps eliminate unexpected complications. Coupled with Hyve’s Unified Global Services, this platform offers best-of-breed, networking solutions for your AI infrastructure strategy.
Want to learn about Hyve’s Unified Global Services?
