NeutronOrch: Rethinking Sample-based GNN Training under CPU-GPU Heterogeneous Environments
Summary: Diagnoses suboptimal CPU–GPU orchestration in sample-based GNN training and introduces layer-decoupled execution that pushes bottom-layer training to CPU to shrink GPU compute and memory footprint. NeutronOrch offloads only frequently accessed vertices with bounded-staleness embedding reuse and a fine-grained pipeline, achieving up to 11.51× speedup over prior systems. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Xin Ai (Northeastern University)
- 2. Qiange Wang (National University of Singapore)
- 3. Chunyu Cao (Northeastern University)
- 4. Yanfeng Zhang (Northeastern University)
- 5. Chaoyi Chen (Northeastern University)
- 6. Hao Yuan (Northeastern University)
- 7. Yu Gu (Northeastern University)
- 8. Ge Yu (Northeastern University)
BibTeX Citation
@article{ai_vldb24,
title = {{NeutronOrch: Rethinking Sample-based GNN Training under CPU-GPU Heterogeneous Environments}},
author = {Ai, Xin and Wang, Qiange and Cao, Chunyu and Zhang, Yanfeng and Chen, Chaoyi and Yuan, Hao and Gu, Yu and Yu, Ge},
journal = {PVLDB},
series = {{VLDB} '24},
volume = {17},
number = {8},
pages = {1995--2008},
doi = {10.14778/3659437.3659453},
url = {https://doi.org/10.14778/3659437.3659453},
year = {2024}
}Incoming Citations (Sorted by Pagerank)
Showing 7 of 7 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 6,630 | Efficient Training of Graph Neural Networks on Large Graphs | 2024 | VLDB | 5.8192133e-05 |
| 10,309 | A Comprehensive Benchmark on Spectral GNNs: The Impact on Efficiency, Memory, and Effectiveness | 2026 | SIGMOD | 5.093636e-05 |
| 10,323 | NeutronHeter: Optimizing Distributed Graph Neural Network Training for Heterogeneous Clusters | 2026 | SIGMOD | 5.093636e-05 |
| 10,521 | Efficient GNN Training on Giant Graphs with Collective Batching and Scheduling | 2026 | VLDB | 5.093636e-05 |
| 10,811 | Graph Neural Network Training Systems: A Performance Comparison of Full-Graph and Mini-Batch | 2025 | VLDB | 5.093636e-05 |
| 10,835 | NeutronTask: Scalable and Efficient Multi-GPU GNN Training with Task Parallelism | 2025 | VLDB | 5.093636e-05 |
| 10,975 | Faster Convergence in Mini-batch Graph Neural Networks Training with Pseudo Full Neighborhood Compensation | 2025 | VLDB | 5.093636e-05 |
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 6 of 6 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 1,132 | SANCUS: Staleness-Aware Communication-Avoiding Full-Graph Decentralized Training in Large-Scale Graph Neural Networks | 2022 | VLDB | 0.00012041292 |
| 1,234 | Large Graph Convolutional Network Training with GPU-Oriented Data Communication Architecture | 2021 | VLDB | 0.00011549432 |
| 2,668 | DUCATI: A Dual-Cache Training System for Graph Neural Networks on Giant Graphs with the GPU | 2023 | SIGMOD | 8.2750247e-05 |
| 2,695 | NeutronStar: Distributed GNN Training with Hybrid Dependency Management | 2022 | SIGMOD | 8.2468134e-05 |
| 3,304 | Saga: A Platform for Continuous Construction and Serving of Knowledge At Scale | 2022 | SIGMOD | 7.5404346e-05 |
| 5,243 | FreshGNN: Reducing Memory Access via Stable Historical Embeddings for Graph Neural Network Training | 2024 | VLDB | 6.3018825e-05 |
Previous
Page 1 / 1
Next