CMANNS: GPU-Accelerated Graph Index Construction for ANNS via Compute-Memory Disaggregation
Summary: CMANNS accelerates NSG/HNSW construction by disaggregating dense distance computation (Tensor-Core GEMMs) from irregular graph memory operations. Hot-set locality and overlapped shard streaming scale beyond HBM, yielding up to 13.05× faster builds without changing query recall/latency. (summarized by gpt-5.6-luna on Jul 26 2026)
Incoming Non-self Citations Over Time
No non-self incoming citations found for this paper in this database.
Authors
- 1. Chengying Huan (Nanjing University)
- 2. Renjie Yao (Nanjing University)
- 3. Shaonan Ma (Qiyuan Lab)
- 4. Rong Gu (Nanjing University)
- 5. Zhengyi Yang (University of New South Wales)
- 6. Lizheng Chen (Nanjing University)
- 7. Zhibin Wang (Nanjing University)
- 8. Mingxing Zhang (Tsinghua University)
- 9. Fang Xi (Qiyuan Lab)
- 10. Guihai Chen (Nanjing University)
- 11. Chen Tian (Nanjing University)
BibTeX Citation
@inproceedings{huan_sigmod26,
title = {{CMANNS: GPU-Accelerated Graph Index Construction for ANNS via Compute-Memory Disaggregation}},
author = {Huan, Chengying and Yao, Renjie and Ma, Shaonan and Gu, Rong and Yang, Zhengyi and Chen, Lizheng and Wang, Zhibin and Zhang, Mingxing and Xi, Fang and Chen, Guihai and Tian, Chen},
series = {{SIGMOD} '26},
booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
publisher = {Association for Computing Machinery},
doi = {10.1145/3802027},
url = {https://dl.acm.org/doi/10.1145/3802027},
year = {2026}
}Incoming Citations (Sorted by Pagerank)
Showing 0 of 0 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 18 of 18 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
Previous
Page 1 / 1
Next