Date: October 2, 2026
Speaker: Eric Ding, Ph.D. Student, Cornell Bowers
Title: Photonic Rail-Optimized Fabric in ML Datacenters for Efficient LLM Training
Abstract: Rail-optimized network fabrics have become the de facto datacenter scale-out fabric for large-scale ML training. However, the use of high-radix electrical switches to provide all-to-all connectivity in rails imposes substantial power and cost. We propose a rethinking of the rail abstraction by retaining its communication semantics, but realizing it using optical circuit switches. The key challenge is that optical switches support one-to-one connectivity at a time, limiting the fan-out of traffic in ML workloads using hybrid parallelisms. We overcome this through parallelism-driven rail reconfiguration, which exploits the non-overlapping communication phases of different parallelism dimensions. This time-multiplexes a single set of physical ports across circuit configurations tailored to each phase within a training iteration. We design and implement Opus, a control plane that orchestrates this in-job reconfiguration of photonic rails at parallelism phase boundaries, and evaluate it on a physical OCS testbed, the Perlmutter supercomputer, and in simulation at up to 2,048 GPUs. Our results show that photonic rails can achieve over 23x network power reduction and 4x cost savings while incurring only modest training overhead at production-relevant OCS reconfiguration latencies.
Bio: Eric Ding is a Ph.D. student working with Prof. Rachee Singh. His research focuses on systems for AI and datacenter networking, with a particular interest in applying optics to improve interconnect efficiency in AI datacenters.
