HyGra: Accelerating Network-State Simulation for LLM Training in DCNs via Adaptive Packet-Flow Granularity

TL;DR

HyGra is a hybrid-granularity network simulator that adaptively switches between packet-level and flow-level simulation to efficiently and accurately emulate data center network dynamics during large language model training.

Contribution

It introduces HyGra, a novel adaptive simulation approach that significantly accelerates network-state simulation for LLM training without sacrificing fidelity.

Findings

Achieves up to 15.4x speedup in simulations.

Maintains high accuracy during LLM workload emulation.

Supports existing simulators without specialized hardware.

Abstract

In recent years, large language models (LLMs) have driven substantial intelligent transformation across diverse industries. Commercial LLM training is typically performed over data center networks (DCNs) comprising hundreds to thousands of GPUs, with multiple devices collocated per node. As network scale expands, inter-node communication becomes a primary bottleneck to training efficiency. Network-state simulators therefore play a crucial role by enabling cost-effective evaluation of network configurations and parallelization strategies through faithful emulation of DCN dynamics during LLM training. However, existing simulators are constrained by a efficiency-fidelity tradeoff, as packet-level simulators (PLSs) incur prohibitive runtime overhead, whereas flow-level simulators (FLSs) compromise essential modeling accuracy. In this paper, we develop \texttt{HyGra}, a hybrid-granularity network-state simulator that exploits intrinsic network dynamics in LLM training to adaptively switch simulation granularity. Specifically, \texttt{HyGra} employs packet-level simulation during non-steady phases with transient fluctuations and flow-level simulation during steady phases with periodic patterns, thereby accelerating execution while preserving high fidelity. Moreover, it requires no specialized hardware, supports single-machine deployment, and is compatible with existing simulators. Experiments based representative commercial LLM workloads, including ChatGPT, DeepSeek, and Qwen, show that \texttt{HyGra} achieves up to 15.4$\times$ speedup under single parallelization strategy and 7.8$\times$ under hybrid parallelization strategies while maintaining high accuracy.