Sean Fraser·Oct 1InfiniBand vs RoCEv2 for Inference: Start With the Traffic, Not the LogoMy current inference machine is one Threadripper PRO 7960X with one RTX PRO 6000. It is not a cluster, and I have not run an…
AICPLIGHT·Sep 22Spectrum-X vs Quantum-X: Which Network Architecture Fits Your AI Data Center?As AI clusters grow larger, networking is becoming one of the most critical components of AI infrastructure.
Jaya Sandeep Ketha·Sep 20How I Got Hands-On Access to a Real InfiniBand HPC Environment with NVIDIA/Mellanox NetworkingI wanted hands-on access to a real HPC environment, particularly to learn more about Linux-based HPC systems, high-speed networking, and…
Jaya Sandeep Ketha·Sep 13Why AI and HPC Need High-Speed Networking: Before Learning InfiniBandAI Networking from the Wire to the GPU - Part 1
gengwg·Aug 25The IOMMU Was Eating 40% of Our B200 InfiniBand BandwidthA brand-new eight-rail B200 cluster was delivering roughly 229 GB/s on a multi-node NCCL all-reduce. It should have been doing 350 to 390…
AICPLIGHT·Jul 7InfiniBand Network Solutions: Transceiver & Cable Deployment Guide for AI Data CentersBuilding next-generation AI infrastructure is no longer just about compute power; it is about eliminating the network bottlenecks that…
AICPLIGHT·May 19InfiniBand XDR vs 800G RoCE: Which is Better for AI Clusters and Tail Latency?In the frantic race to build the next generation of AI superclusters, the spotlight often shines brightest on GPUs like NVIDIA’s Blackwell…
Inmy_aimlbySanjay Basu, PhD·Feb 16The Backbone Nobody SeesA Deep Dive into HPC Communication Protocols and the Networks That Make Everything Possible
Kannan Venugopal·Jan 26The 8-Lane Highway (Why Bandwidth without Rails is Worthless)1. The “Pooling” Trap
