Articles by tag "InfiniBand"

1. Optimizing AI Inference Clusters: MCX7 AAS-NEAT Adapters
   Selene Gong

   25
   As the demand for Large Language Models (LLMs) and generative AI surges across Singapore and APAC, GPU-as-a-Service (GPUaaS) providers and AI engineers face a critical challenge: your inference cluster is only as fast as its network. Microsecond delays in transferring KV caches or synchronizing ...
2. How to Choose Between NVIDIA InfiniBand and Ethernet for AI Infrastructure Deployment
   Selene Gong

   33
   When AI infrastructure projects move from concept to deployment, one of the most important design decisions is whether the network should follow an InfiniBand path or an Ethernet path. This is not just a hardware preference. It affects how the environment scales, how the architecture is planned, ...
3. How to Simplify Global AI Infrastructure Procurement with a One-Stop High-Performance Networking Supply Chain
   Selene Gong

   18
   ...
4. Troubleshooting InfiniBand Subnet Manager Failover High-Availability OpenSM Architecture for MQM8790 Unmanaged Clusters
   Selene Gong

   18
   Quick Take In deployments based on unmanaged MQM8790 switches, the entire InfiniBand fabric control plane relies on an external Subnet Manager (SM). Implementing a high-availability active/standby OpenSM architecture with distinct sm_priority values prevents full cluster outages during ...