Preprint
Jul 2026
FAST: A Holistic Framework for Optimizing Memory-I/O, Computation, and Sampling in Temporal GNN Training
FAST is presented, a holistic framework that accelerates end-to-end TGNN training by jointly optimizing sampling, memory I/O, and computation and designs thread-efficient graph operators tailored to sparse temporal subgraphs, improving GPU cache locality and reducing the latency of aggregation and edge softmax.
Yu Cai, Qingrui Zhu, Lei Liu et al.
· 0 citations