Technical Papers Archive

DOLPHIN: Scalable Disk-RAM-GPU Pipelined Training for Massive Temporal GNNs


Authors: Zezhong Ding, Rui Guo, Wenbo Zhen, and Junlin Lv (University of Science and Technology of China, Jiangsu Provincial Key Laboratory of Multimodal Digital Twin Technology); Jianliang Xu (Hong Kong Baptist University); and Xike Xie (University of Science and Technology of China, Jiangsu Provincial Key Laboratory of Multimodal Digital Twin Technology)

Abstract: Temporal graph neural networks (T-GNNs) play a key role in modeling dynamic graphs by learning representations that capture evolving structures and interactions.

However, scaling T-GNNs remains challenging, as massive real-world graphs and complex temporal dependencies incur prohibitive runtime and memory costs. Existing systems often preload entire datasets into RAM and rely on RAM-GPU pipelines, suffering from batch-boundary pipeline bubbles, thereby limiting throughput.

In this paper, we present DOLPHIN, a scalable Disk-RAM-GPU pipelined system for massive T-GNNs. DOLPHIN groups training batches into chunks and incorporates two key techniques: an I/O-efficient feature window scheduler for just-in-time Disk-to-RAM chunk provisioning, and a dependency-graph–based parallel execution plan that improves GPU utilization while preserving model quality. Together, these techniques improve pipeline efficiency across the Disk-RAM-GPU memory hierarchy and computation phases.

Experiments on six real-world graphs show DOLPHIN delivers up to 21.07x faster training and reduces RAM usage by up to 92.04% compared to state-of-the-art baselines.



Presentation: file

Back to Technical Papers Archive Listing