Presented by
Tom Sanfilippo, CTO, WhiteFiber; Yossi Kikozashvili, Head of AI Infrastructure Product line, DriveNets
Insights from deploying an Ethernet-based GPU cluster fabric
About this talk
The NeoCloud WhiteFiber successfully deployed a large-scale AI cluster using DriveNets Network Cloud-AI for its back-end fabric and storage network.
- AI cluster performance network considerations and the benefits of fabric-scheduled Ethernet
- Moving from InfiniBand to Ethernet-based fabric without compromising Job Completion Time (JCT)
- Insights on how to accelerate AI fabric deployment time - lessons learned & best practices
DriveNets
370 subscribers3 talks
Highest performance AI Ethernet Fabric
Large AI systems support growing workloads running over individual compute and storage nodes that work together as a logical cluster and are connected via a high-capacity interconnect fabric, also known as AI networking. These networking solutions need to evolve to maximize the utilization of costly AI compute resources and support standard connectivity that enables vendor interoperability. DriveNets Network Cloud-AI offers the highest performance AI networking at scale – at the most optimal cost structure.