Distributed GNN training paper is accepted by ICCD 2025!

👏 Paper title: Towards Affordable, Adaptive and Automatic GNN Training on CPU-GPU Heterogeneous Platforms. In this paper, we introduces A3GNN, a framework for Affordable, Adaptive, and Automatic GNN training on heterogeneous CPU-GPU platforms. It improves resource usage through locality-aware sampling and fine-grained parallelism scheduling. Moreover, it leverages reinforcement learning to explore the design space and achieve pareto-optimal trade-offs among throughput, memory footprint, and accuracy. [related project]