Saturn: An Optimized Data System for Multi-Large-Model Deep Learning Workloads
Summary: Formalizes SPASE: jointly select parallelism, allocate GPUs, and schedule multi-model DL experiments via a template plus an empirical profiler. Solves SPASE as an MILP (beats heuristics) with introspective scheduling in Saturn, reducing model-selection runtime by 39–49%. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Kabir Nagrecha (University of California San Diego)
- 2. Arun Kumar (University of California San Diego)
BibTeX Citation
@article{nagrecha_vldb24,
title = {{Saturn: An Optimized Data System for Multi-Large-Model Deep Learning Workloads}},
author = {Nagrecha, Kabir and Kumar, Arun},
journal = {PVLDB},
series = {{VLDB} '24},
volume = {17},
number = {4},
pages = {712--725},
doi = {10.14778/3636218.3636227},
url = {https://doi.org/10.14778/3636218.3636227},
year = {2024}
}Incoming Citations (Sorted by Pagerank)
Showing 3 of 3 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 5,899 | Lotan: Bridging the Gap between GNNs and Scalable Graph Analytics Engines | 2023 | VLDB | 5.9545431e-05 |
| 10,668 | Performant Synchronization in Geo-Distributed Databases | 2026 | SIGMOD | 4.9793485e-05 |
| 10,757 | Mil: Cost-guided Minimum Makespan Scheduling for Applications of Multiple LLMs | 2026 | VLDB | 4.9793485e-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 |
|---|---|---|---|---|
| 78 | Automatic Database Management System Tuning Through Large-scale Machine Learning | 2017 | SIGMOD | 0.00036684414 |
| 522 | PyTorch Distributed: Experiences on Accelerating Data Parallel Training | 2020 | VLDB | 0.00016912378 |
| 940 | Starfish: A Self-tuning System for Big Data Analytics | 2011 | CIDR | 0.00012964445 |
| 5,633 | NuPS: A Parameter Server for Machine Learning with Non-Uniform Parameter Access | 2022 | SIGMOD | 6.0578661e-05 |
| 9,144 | Cerebro: A Layered Data Platform for Scalable Deep Learning | 2021 | CIDR | 5.2201471e-05 |
| 9,595 | Model-Parallel Model Selection for Deep Learning Systems | 2021 | SIGMOD | 5.1560462e-05 |
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 1,081 | Hybrid Parallelization Strategies for Large-Scale Machine Learning in SystemML | 2014 | VLDB |
| 2 | 7,809 | Nautilus: An Optimized System for Deep Transfer Learning over Evolving Training Datasets | 2022 | SIGMOD |
| 3 | 9,595 | Model-Parallel Model Selection for Deep Learning Systems | 2021 | SIGMOD |
| 4 | 9,556 | Towards an Optimized GROUP BY Abstraction for Large-Scale Machine Learning | 2021 | VLDB |
| 5 | 1,152 | Cerebro: A Data System for Optimized Deep Learning Model Selection | 2020 | VLDB |
| 6 | 10,435 | DFLOP: A Data-driven Framework for Multimodal LLM Training Pipeline Optimization | 2026 | SIGMOD |
| 7 | 4,095 | Distributed Deep Learning on Data Systems: A Comparative Analysis of Approaches | 2021 | VLDB |
| 8 | 10,757 | Mil: Cost-guided Minimum Makespan Scheduling for Applications of Multiple LLMs | 2026 | VLDB |
| 9 | 13,689 | Reimagining Deep Learning Systems Through the Lens of Data Systems | 2024 | VLDB |
| 10 | 5,009 | Galvatron: Efficient Transformer Training over Multiple GPUs Using Automatic Parallelism | 2023 | VLDB |