L3: A GPU-Native Co-Designed Data Format for Learned Lossless Lightweight Compression
Summary: L3 co-designs a GPU-native learned lossless codec around SLAP’s lane-major residual layout, warp-cooperative decoding, and fully on-device compression. It delivers TB/s decompression, 1.2–2.6B random accesses/s, and strong compression, outperforming GPU baselines. (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. Youyang Xia (Renmin University of China)
- 2. Feng Zhang (Renmin University of China)
- 3. Junda Pan (Renmin University of China)
- 4. Yihao Liu (Tsinghua University)
- 5. Jiawei Guan (Renmin University of China)
- 6. Huanchen Zhang (Tsinghua University)
- 7. Xiaoyong Du (Renmin University of China)
BibTeX Citation
@inproceedings{xia_sigmod26,
title = {{L3: A GPU-Native Co-Designed Data Format for Learned Lossless Lightweight Compression}},
author = {Xia, Youyang and Zhang, Feng and Pan, Junda and Liu, Yihao and Guan, Jiawei and Zhang, Huanchen and Du, Xiaoyong},
series = {{SIGMOD} '26},
booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
publisher = {Association for Computing Machinery},
doi = {10.1145/3802078},
url = {https://dl.acm.org/doi/10.1145/3802078},
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 26 of 26 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 2,788 | BtrBlocks: Efficient Columnar Compression for Data Lakes | 2023 | SIGMOD |
| 2 | 4,157 | GPU-based Graph Traversal on Compressed Graphs | 2019 | SIGMOD |
| 3 | 4,065 | LeCo: Lightweight Compression via Learning Serial Correlations | 2024 | SIGMOD |
| 4 | 1,644 | Compressed Linear Algebra for Large-Scale Machine Learning | 2016 | VLDB |
| 5 | 3,592 | The FastLanes Compression Layout: Decoding >100 Billion Integers per Second with Scalar Code | 2023 | VLDB |
| 6 | 10,261 | LICO: An SIMD-Aware High-Performance Learned Inverted Index Compression Framework | 2026 | SIGMOD |
| 7 | 9,038 | The FastLanes File Format | 2025 | VLDB |
| 8 | 2,232 | Database Compression on Graphics Processors | 2010 | VLDB |
| 9 | 10,580 | GPU Acceleration of SQL Analytics on Compressed Data | 2026 | VLDB |
| 10 | 4,144 | Tile-based Lightweight Integer Compression in GPU | 2022 | SIGMOD |