SNI-GNN: SmartNIC-Assisted Full-Graph GNN Training with In-Network Embedding Prediction

Researchers introduced SNI-GNN, a SmartNIC-assisted full-graph GNN training system that reduces communication by 21–45% and achieves 1.3–3.6× end-to-end speedups over BNS-GCN and up to 1.29× over SANCUS, with accuracy loss ≤0.01, by predicting remote embeddings in-network on NVIDIA BlueField-3 SmartNICs. The system scales to 16 GPUs on graphs with tens of millions of edges, demonstrating that in-network prediction complements partitioning and compression for efficient full-graph training.

arXiv:2608.06441v1 Announce Type: new Abstract: Full-graph GNN training delivers high accuracy but scales poorly on multi-server clusters due to heavy, irregular inter-node embedding exchanges. We present SNI-GNN, a SmartNIC-assisted full-graph training system that reduces communication while preserving accuracy by predicting remote embeddings in-network. SNI-GNN deploys a lightweight linear-trend predictor on SmartNICs to refine cached historical embeddings, coupled with an importance-based boundary-node sampling policy and an asynchronous DPU--GPU data pipeline with intermediate-result reuse. We provide error and convergence bounds showing that predictor bias remains controlled under bounded second-order dynamics and yields standard non-convex convergence with inexact gradients. Implemented on NVIDIA BlueField-3, SNI-GNN integrates with state-of-the-art full-graph systems, cuts communication by 21--45\%, achieves 1.3--3.6$\times$ end-to-end speedups over BNS-GCN and up to 1.29$\times$ over baseline SANCUS, with accuracy loss $\leq 0.01$, and scales efficiently to 16 GPUs on graphs with up to tens of millions of edges. These results indicate SmartNIC-based in-network prediction is a practical complement to partitioning and compression techniques for communication-efficient full-graph GNN training at scale.