ThunderGNN: Unlocking Tensor Cores for Graph Neural Networks

Summary: ThunderGNN co-designs row reordering, zero-padding-free CBA storage, and compressed-block streaming to map irregular sparse GNN workloads onto NVIDIA Tensor Cores. It delivers 1.89×/2.59× geometric-mean speedups over DGL/PyG on A100s. (summarized by gpt-5.6-luna on Aug 28 2026)

Paper ID: he10556dc8b70899e
Venue: VLDB
Year: 2026
Pagerank: 4.9793485e-05
Overall Rank: 10,818 | 27.27%
DOI: 10.14778/3828612.3828625

Incoming Non-self Citations Over Time

No non-self incoming citations found for this paper in this database.

Authors

1. YuAng Chen (Chinese University of Hong Kong)
2. Siyi Teng (Chinese University of Hong Kong)
3. Wenqi Zeng (Hong Kong University of Science and Technology)
4. Jeffrey Xu Yu (Hong Kong University of Science and Technology)

BibTeX Citation

@article{chen_vldb26,
        title = {{ThunderGNN: Unlocking Tensor Cores for Graph Neural Networks}},
        author = {Chen, YuAng and Teng, Siyi and Zeng, Wenqi and Yu, Jeffrey Xu},
        journal = {PVLDB},
        series = {{VLDB} '26},
        volume = {19},
        number = {10},
        pages = {2699--2712},
        doi = {10.14778/3828612.3828625},
        url = {https://doi.org/10.14778/3828612.3828625},
        year = {2026}
}

Incoming Citations (Sorted by Pagerank)

Showing 0 of 0 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank

Outgoing Citations (Sorted by Pagerank)

Showing 10 of 10 cited papers.

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

Rank	Cited Paper	Year	Venue	Pagerank
1,440	Speedup Graph Processing by Graph Ordering	2016	SIGMOD	0.00010641888
2,506	DUCATI: A Dual-Cache Training System for Graph Neural Networks on Giant Graphs with the GPU	2023	SIGMOD	8.3774747e-05
2,774	GraphScope: A Unified Engine For Big Graph Processing	2021	VLDB	8.0327008e-05
4,351	Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity	2024	VLDB	6.642803e-05
4,825	DAHA: Accelerating GNN Training with Data and Hardware Aware Execution Planning	2024	VLDB	6.3923554e-05
4,826	Accelerating Sampling and Aggregation Operations in GNN Frameworks with GPU Initiated Direct Storage Accesses	2024	VLDB	6.391978e-05
4,898	Comprehensive Evaluation of GNN Training Systems: A Data Management Perspective	2024	VLDB	6.365824e-05
5,124	SCARA: Scalable Graph Neural Networks with Feature-Oriented Optimization	2022	VLDB	6.2624516e-05
5,899	Lotan: Bridging the Gap between GNNs and Scalable Graph Analytics Engines	2023	VLDB	5.9545431e-05
9,233	Can Graph Reordering Speed Up Graph Neural Network Training? An Experimental Study	2025	VLDB	5.2056825e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	2,596	Scalable and Efficient Full-Graph GNN Training for Large Graphs	2023	SIGMOD
2	9,233	Can Graph Reordering Speed Up Graph Neural Network Training? An Experimental Study	2025	VLDB
3	1,224	Large Graph Convolutional Network Training with GPU-Oriented Data Communication Architecture	2021	VLDB
4	6,826	SIMPLE: Efficient Temporal Graph Neural Network Training at Scale with Dynamic Data Placement	2024	SIGMOD
5	6,793	XGNN: Boosting Multi-GPU GNN Training via Global GNN Memory Store	2024	VLDB
6	11,459	Towards Ideal Temporal Graph Neural Networks: Evaluations and Conclusions after 10,000 GPU Hours	2025	VLDB
7	1,364	TGL: A General Framework for Temporal GNN Training on Billion-Scale Graphs	2022	VLDB
8	4,826	Accelerating Sampling and Aggregation Operations in GNN Frameworks with GPU Initiated Direct Storage Accesses	2024	VLDB
9	3,752	G3: When Graph Neural Networks Meet Parallel Graph Processing Systems on GPUs	2020	VLDB
10	9,032	TGraph: A Tensor-centric Graph Processing Framework	2025	SIGMOD