DiskGNN: Bridging I/O Efficiency and Model Accuracy for Out-of-Core GNN Training

Summary: Offline sampling decouples graph sampling from computation, enabling batch features and avoiding accuracy loss. 4-level store, batched packing, and pipelined training exploit CPU/GPU hierarchy to cut I/O and deliver ~8x speed with identical accuracy. (summarized by gpt-5-nano on Feb 09 2026)

Paper ID: 7139
Venue: SIGMOD
Year: 2025
Pagerank: 5.8782861e-05
Overall Rank: 6,440 | 55.82%
DOI: 10.1145/3709738

Incoming Non-self Citations Over Time

Authors

1. Renjie Liu (Southern University of Science and Technology)
2. Yichuan Wang (University of California Berkeley)
3. Xiao Yan (Centre for Perceptual and Interactive Intelligence)
4. Haitian Jiang (New York University)
5. Zhenkun Cai (Amazon)
6. Minjie Wang (Amazon)
7. Bo Tang (Southern University of Science and Technology)
8. Jinyang Li (New York University)

BibTeX Citation

@inproceedings{liu_sigmod25,
        title = {{DiskGNN: Bridging I/O Efficiency and Model Accuracy for Out-of-Core GNN Training}},
        author = {Liu, Renjie and Wang, Yichuan and Yan, Xiao and Jiang, Haitian and Cai, Zhenkun and Wang, Minjie and Tang, Bo and Li, Jinyang},
        series = {{SIGMOD} '25},
        booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
        publisher = {Association for Computing Machinery},
        doi = {10.1145/3709738},
        url = {https://dl.acm.org/doi/10.1145/3709738},
        year = {2025}
}

Incoming Citations (Sorted by Pagerank)

Showing 5 of 5 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
9,879	Capsule*: An Out-of-Core Training Mechanism for Colossal GNNs	2025	SIGMOD	5.2040783e-05
10,207	Bw-Graph: An Efficient Graph Storage System Harmonizing Topology-Aware Tree with Paged CSR	2026	SIGMOD	5.093636e-05
10,251	GPS: Revisiting the Data Layout for Disk-based High-Dimensional Vector Search	2026	SIGMOD	5.093636e-05
10,373	Gem: Scalable Monotonic Graph Processing Beyond Billion-Scale on a Single Machine	2026	SIGMOD	5.093636e-05
10,899	Can Graph Reordering Speed Up Graph Neural Network Training? An Experimental Study	2025	VLDB	5.093636e-05

Outgoing Citations (Sorted by Pagerank)

Showing 4 of 4 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
1,511	Speedup Graph Processing by Graph Ordering	2016	SIGMOD	0.00010538011
2,953	Ginex: SSD-enabled Billion-scale Graph Neural Network Training on a Single Machine via Provably Optimal In-memory Caching	2022	VLDB	7.9237794e-05
5,597	Accelerating Sampling and Aggregation Operations in GNN Frameworks with GPU Initiated Direct Storage Accesses	2024	VLDB	6.1547003e-05
6,566	Play like a Vertex: A Stackelberg Game Approach for Streaming Graph Partitioning	2024	SIGMOD	5.8381749e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	1,234	Large Graph Convolutional Network Training with GPU-Oriented Data Communication Architecture	2021	VLDB
2	1,863	ByteGNN: Efficient Graph Neural Network Training at Large Scale	2022	VLDB
3	2,640	Scalable and Efficient Full-Graph GNN Training for Large Graphs	2023	SIGMOD
4	7,374	ADGNN: Towards Scalable GNN Training with Aggregation-Difference Aware Sampling	2023	SIGMOD
5	6,630	Efficient Training of Graph Neural Networks on Large Graphs	2024	VLDB
6	6,803	OUTRE: An OUT-of-core De-REdundancy GNN Training Framework for Massive Graphs within A Single Machine	2024	VLDB
7	5,243	FreshGNN: Reducing Memory Access via Stable Historical Embeddings for Graph Neural Network Training	2024	VLDB
8	5,316	Comprehensive Evaluation of GNN Training Systems: A Data Management Perspective	2024	VLDB
9	2,953	Ginex: SSD-enabled Billion-scale Graph Neural Network Training on a Single Machine via Provably Optimal In-memory Caching	2022	VLDB
10	5,597	Accelerating Sampling and Aggregation Operations in GNN Frameworks with GPU Initiated Direct Storage Accesses	2024	VLDB