ATLAS: Efficient Out-of-Core Inference for Billion-Scale Graph Neural Networks

TL;DR

ATLAS is a disk-based GNN inference framework that enables efficient full-graph, layer-wise inference on billion-scale graphs exceeding memory capacity, achieving 12-30x speedup over state-of-the-art methods.

Contribution

ATLAS introduces a broadcast-based execution model and a tiered memory-disk hierarchy to significantly improve out-of-core GNN inference efficiency.

Findings

Achieves 12-30x faster inference than existing out-of-core baselines.

Supports graphs with up to 4 billion edges and 550 GiB features.

Maintains high throughput with only 128 GiB RAM and 2 TiB SSD.

Abstract

Graph Neural Network (GNN) inference on billion-scale graphs is critical for domains like fintech and recommendation systems. Full-graph inference on these large graphs can be challenging due to high communication costs in distributed settings and high I/O costs in disk-backed Out-of-Core (OOC) settings. Existing OOC systems, operating across disk and memory, primarily focus on GNN training and perform poorly for full-graph inference due to massive read amplification, irregular I/O, and memory pressure. We present ATLAS, a disk-based GNN inference framework that enables efficient full-graph, layer-wise inference on graphs whose topologies, features and intermediate embeddings exceed the available memory on single machines. ATLAS replaces gather-based execution with a broadcast-based model that enables sequential, single-pass streaming reads of features and embeddings per layer. A tiered memory-disk hierarchy with minimum-pending-message eviction, graph reordering and a GPU-accelerated pipeline sustains high throughput within $128$ GiB RAM and $2$ TiB SSD. Across out-of-core graphs with up to $4$B edges and $550$ GiB features and multiple GNN architectures, ATLAS improves end-to-end inference time by $\approx12$--$30\times$ over State-of-the-Art (SOTA) OOC baselines on a single workstation, while remaining within $\approx5\%$ when features fit in memory.