Deep Graph Library: A Graph-Centric, Highly-Performant Package for Graph Neural Networks

TL;DR

Deep Graph Library (DGL) is a highly efficient, framework-neutral tool that enables scalable graph neural network computations through generalized sparse tensor operations, outperforming existing frameworks in speed and memory efficiency.

Contribution

DGL introduces a graph-centric design with generalized sparse tensor operations, supporting multiple frameworks and optimizing GNN computations for speed and memory.

Findings

DGL significantly outperforms other frameworks in speed.

DGL reduces memory consumption.

DGL has minimal overhead for small workloads.

Abstract

Advancing research in the emerging field of deep graph learning requires new tools to support tensor computation over graphs. In this paper, we present the design principles and implementation of Deep Graph Library (DGL). DGL distills the computational patterns of GNNs into a few generalized sparse tensor operations suitable for extensive parallelization. By advocating graph as the central programming abstraction, DGL can perform optimizations transparently. By cautiously adopting a framework-neutral design, DGL allows users to easily port and leverage the existing components across multiple deep learning frameworks. Our evaluation shows that DGL significantly outperforms other popular GNN-oriented frameworks in both speed and memory consumption over a variety of benchmarks and has little overhead for small scale workloads.