# Scheduling Parallel Optical Circuit Switches for AI Training

arXiv: https://arxiv.org/abs/2603.07373 · HTML: https://mlsanity.com/paper/2603.07373/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Efficient MoE training and inference systems, one of 12 directions in Mixture-of-experts.
- Builds on 5 of its 25 references that are in the map.
- First posted on arXiv 7 Mar 2026.

## Placed in

- Mixture-of-experts / [Efficient MoE training and inference systems](https://mlsanity.com/moe/efficient-moe-training-and-inference-systems/)

## Builds on (most cited first, 5 of 5)

- [MixNet: A Runtime Reconfigurable Optical-Electrical Fabric for Distributed Mixture-of-Experts Training](https://mlsanity.com/paper/2501.03905/) (2025)
- [Approximation algorithms for Job Scheduling with reconfigurable resources](https://mlsanity.com/paper/2401.00419/) (2023)
- [NegotiaToR: Towards A Simple Yet Effective On-demand Reconfigurable Datacenter Network](https://mlsanity.com/paper/2407.20045/) (2024)
- [Vermilion: A Traffic-Aware Reconfigurable Optical Interconnect with Formal Throughput Guarantees](https://mlsanity.com/paper/2504.09892/) (2025)
- [RailX: A Flexible, Scalable, and Low-Cost Network Architecture for Hyper-Scale LLM Training Systems](https://mlsanity.com/paper/2507.18889/) (2025)

## Abstract (arXiv)

The rapid growth of AI training has dramatically increased datacenter traffic demand and energy consumption, which has motivated renewed interest in optical circuit switches (OCSes) as a high-bandwidth, energy-efficient alternative for AI fabrics. Deploying multiple parallel OCSes is a leading alternative. However, efficiently scheduling time-varying traffic matrices across parallel optical switches with non-negligible reconfiguration delays remains an open challenge. We consider the problem of scheduling a single AI traffic demand matrix D over s parallel OCSes while minimizing the makespan under reconfiguration delay δ. Our algorithm Spectra relies on a three-step approach: Decompose D into a minimal set of weighted permutations; Schedule these permutations across parallel switches using load-aware assignment; then Equalize the imbalanced loads on the switches via controlled permutation splitting. Evaluated on realistic AI training workloads (GPT model and Qwen MoE expert routing) as well as standard benchmarks, Spectra vastly outperforms a baseline based on state-of-the-art algorithms, reducing schedule makespan by an average factor of 1.4× on GPT AI workloads, 2.7× on MoE AI workloads, and 2.4× on standard benchmarks. Further, the makespans achieved by Spectra consistently approach newly derived lower bounds.