SG-Serve: Efficient Model Serving for Subgraph-based Graph Representation Learning
Summary: SG-Serve addresses SGRL’s skewed online workloads via parallel extraction, CPU work stealing, workload-aware GPU batching, and dual GPU processes that avoid head-of-line blocking. It cuts P99 latency 13× and improves throughput by over 33%.
(summarized by gpt-5.6-luna on Jul 26 2026)
@inproceedings{zhou_sigmod26,
title = {{SG-Serve: Efficient Model Serving for Subgraph-based Graph Representation Learning}},
author = {Zhou, Qihui and Yin, Peiqi and Yan, Xiao and Li, Changji and Cheng, James},
series = {{SIGMOD} '26},
booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
publisher = {Association for Computing Machinery},
doi = {10.1145/3786697},
url = {https://dl.acm.org/doi/10.1145/3786697},
year = {2026}
}
Incoming Citations (Sorted by Pagerank)
Showing 0 of 0 citing papers.
Rank
Citing Paper
Year
Venue
Pagerank
PreviousPage 1 / 1Next
Outgoing Citations (Sorted by Pagerank)
Showing 16 of 16 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.