---
title: "Workload-Static Optical Circuit Switching For Scalable LLM Inference Decode"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.05104v1"
permalink: "https://feed.astralor.com/items/itm_34680961e48f24a8d393a9bc"
published: "2026-11-03T16:00:07.000Z"
score: 56
channel: "tech"
tags: ["LLM","推理效率","Workload-Static Optical Circuit","Switching For Scalable","LLM Inference Decode","OCS","GXNET","MoE"]
---

# Workload-Static Optical Circuit Switching For Scalable LLM Inference Decode

## At a glance

LLM inference decode executes fine-grained collectives whose latency directly limits per-user interactivity, yet scale-up fabrics are still built from electronic packet switches designed for training. We observe that, f…

## Key points

- LLM inference decode executes fine-grained collectives whose latency directly limits per-user interactivity, yet scale-up fabrics are still built from electronic packet switches designed for training.
- We observe that, for a given model deployment and parallelization, the communication structure of decode is fixed, and propose workload-static optical circuit switching (OCS): circuits are configured once per deployment…
- We evaluate this design with GXNET, a graph-based workload emulator coupled to analytical and flit-level network models grounded in hardware measurements.

## Full text

LLM inference decode executes fine-grained collectives whose latency directly limits per-user interactivity, yet scale-up fabrics are still built from electronic packet switches designed for training. We observe that, for a given model deployment and parallelization, the communication structure of decode is fixed, and propose workload-static optical circuit switching (OCS): circuits are configured once per deployment, co-designed with model parallelism and with topology-aware collective schedules that turn spare accelerator ports into fewer communication phases or higher bandwidth. We evaluate this design with GXNET, a graph-based workload emulator coupled to analytical and flit-level network models grounded in hardware measurements. In a 576-accelerator scale-up domain serving 5T-parameter MoE models with up to 5M tokens of context, all-OCS fabrics improve interactivity by up to 65% and energy efficiency by up to 119% over 4:1 oversubscribed packet-switched fabrics.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_34680961e48f24a8d393a9bc