HGL: Accelerating Heterogeneous GNN Training with Holistic Representation and Optimization
DescriptionGraph neural networks (GNNs) have shown to significantly improve graph analytics. Existing systems for GNN training are primarily designed for homogeneous graphs. In industry, however, most graphs are actually heterogeneous in nature (i.e., having multiple types of nodes and edges). Existing systems train a heterogeneous GNN (HetGNN) as a composition of homogeneous GNNs and thus suffer from critical limitations such as lack of memory optimization and limited operator parallelism. To address these limitations, we propose HGL– a heterogeneity-aware system for GNN training. At the core of HGL is an intermediate representation, called HIR, which provides a holistic representation for GNNs and enables cross-relation optimization for HetGNN training. We devise tailored optimizations on HIR, including graph stitching, operator fusion, and operator bundling. Experimental results verify that HGL significantly outperforms DGL and PyG.
Event Type
Paper
TimeThursday, 17 November 20222:30pm - 3pm CST
LocationC140-142
TP
Machine Learning and Artificial Intelligence
Recorded
Archive
view