Please ensure Javascript is enabled for purposes of website accessibility

Fabric-Scheduled Ethernet as an Effective Backend Interconnect for Large AI Compute Clusters

In this white paper, Fabric-Scheduled Ethernet as an Effective Backend Interconnect for Large AI Compute Clusters, we explore the many networking alternatives offered for interconnecting AI clusters, examining pros and cons and forced tradeoffs for the end user.

Fabric-Scheduled Ethernet appears to meet all the challenges of the AI-enabled data center infrastructure. Fabric-Scheduled Ethernet guarantees lossless connectivity for a large-scale server array running high-bandwidth workloads free of flow discrimination and with minimal impact from network failures, all in a traditional and easy-to- manage Clos-like topology.

In this white paper you will learn about:

  • Networking challenges for AI Compute Server Arrays
  • Networking solutions for AI Clusters
  • How DDC meets the challenges facing an AI-enabled Data Center Infrastructure


Download the White Paper

Continue reading

Videos

Bentaus cuts energy per token by 30% with AMD Instinct and DriveNets AI Fabric

Kevin Matcham, COO and co-founder of Bentaus shares how together with DriveNets and AMD allows them to scale up and sca ...

Read more
Optimal AI performance per watt for AMD Instinct clusters and DriveNets AI Fabric

Case Studies

NeoCloud Case Study: Optimal AI performance per watt for AMD Instinct clusters and DriveNets AI Fabric

How Bentaus built its AMD Instinct™ MI325X NeoCloud cluster on DriveNets AI Fabric and cut energy per token by 30%.

Read more
DriveNets and WhiteFiber Connect Two Distant Sites into One AI Supercluster

Videos

DriveNets and WhiteFiber Connect Two Distant Sites into One AI Supercluster

DriveNets, WhiteFiber and WEKA successfully connected GPU clusters at two geographically separated data centers to opera ...

Read more