UB-Mesh proposes a hierarchical full-mesh network for AI training clusters
bookwormengr · x · 2026-07-22
This reply points to a paper on UB-Mesh, a hierarchical full-mesh datacenter network for AI training.
The core idea is that every NPU can also act as a router, so traffic can take multiple paths instead of relying on a single non-blocking flat fabric. The attached diagram contrasts shortest-path-first routing with all-path routing to show why the design matters for large training systems.
More from Infra
- SF Compute founder: buying compute is 'an absolutely awful experience' right now — IgorCarron · 2026-09-11
- SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11
- PyTorch Day Korea 2026 launches first offline conf, CFP closes Sept 13 — PyTorch · 2026-09-11
- Local LLM server dilemma: 4x CMP-170HX (price up 53% in 20 days) vs Mac Studio M5 Ultra — rumboll · 2026-09-11
- llama.cpp lands Flash Attention tuning for RDNA4, big prefill gains on AMD — pmttyji · 2026-09-11
- Your p99 latency benchmark may be lying: a deep dive into coordinated omission — Franc0Fernand0 · 2026-09-11