Flux: Optimal Scheduling of Optical Circuit Switches for LLM Training

2026-09-22

AI Summary

本文提出Flux调度器，通过基于整个工作负载结构优化光学电路开关调度，解决LLM训练中因忽视计算时间线导致的通信停滞问题，提高带宽利用率并降低缓冲需求。

Abstract

Optical Circuit Switching (OCS) offers high bandwidth density and energy efficiency for LLM training, but incurs a non-negligible reconfiguration delay. Prior work typically schedules optical circuit switches independently of compute, using aggregate traffic demand to determine which circuits to provision and when. We argue that this separation creates a fundamental inefficiency: reconfigurations that ignore the compute timeline can stall communication, resulting in low circuit utilization and large buffer requirements. In this paper, we present Flux, a scheduler that optimally schedules optical circuit switches based on the structure of the entire workload. Flux remains effective across a wide range of switching speeds by reusing circuits and amortizing reconfiguration delay behind compute and communication. We show that Flux reduces training iteration time by up to $10\times$ and peak NIC buffer requirements by more than three orders of magnitude compared to traditional periodic schedulers.

Problem

• Optical Circuit Switching
• Reconfiguration Delay
• Compute Timeline
• Circuit Utilization
• Buffer Requirements

Innovation

• Optical Circuit Switching
• Scheduler
• Workload Structure
• Reconfiguration Delay
• Buffer Requirements