STile: Searching Hybrid Sparse Formats for Sparse Deep Learning Operators Automatically
Summary: STile enlarges the sparse-format search space with flexible tensor transforms and multi-level decomposition; formalizes the NP-hard multi-level sparse-format decomposition problem. Greedy, cost-driven search yields 2.1–18.0x speedups for SpMM (cuSPARSE) and 1.5–6.9x for SDDMM (DGL), with sub-hour search time amortized. (summarized by gpt-5-nano on Feb 09 2026)
Incoming Non-self Citations Over Time
No non-self incoming citations found for this paper in this database.
Authors
- 1. Jingzhi Fang (Hong Kong University of Science and Technology)
- 2. Yanyan Shen (Shanghai Jiao Tong University)
- 3. Yue Wang (Shenzhen University)
- 4. Lei Chen (Hong Kong University of Science and Technology)
BibTeX Citation
@inproceedings{fang_sigmod24,
title = {{STile: Searching Hybrid Sparse Formats for Sparse Deep Learning Operators Automatically}},
author = {Fang, Jingzhi and Shen, Yanyan and Wang, Yue and Chen, Lei},
series = {{SIGMOD} '24},
booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
publisher = {Association for Computing Machinery},
doi = {10.1145/3639323},
url = {https://dl.acm.org/doi/10.1145/3639323},
year = {2024}
}Incoming Citations (Sorted by Pagerank)
Showing 3 of 3 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 6,166 | Apt-Serve: Adaptive Request Scheduling on Hybrid Cache for Scalable LLM Inference Serving | 2025 | SIGMOD | 5.8631131e-05 |
| 6,669 | Efficient Training of Graph Neural Networks on Large Graphs | 2024 | VLDB | 5.7147311e-05 |
| 10,706 | Efficient GNN Training on Giant Graphs with Collective Batching and Scheduling | 2026 | VLDB | 4.9793485e-05 |
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 0 of 0 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 9,944 | On Efficient Large Sparse Matrix Chain Multiplication | 2024 | SIGMOD |
| 2 | 3,330 | SPOOF: Sum-Product Optimization and Operator Fusion for Large-Scale Machine Learning | 2017 | CIDR |
| 3 | 1,614 | Compressed Linear Algebra for Large-Scale Machine Learning | 2016 | VLDB |
| 4 | 10,013 | ETO: Accelerating Optimization of DNN Operators by High-Performance Tensor Program Reuse | 2022 | VLDB |
| 5 | 10,838 | Unified Static–Dynamic Pruning for Efficient LLM Inference | 2026 | VLDB |
| 6 | 4,330 | MNC: Structure-Exploiting Sparsity Estimation for Matrix Expressions | 2019 | SIGMOD |
| 7 | 5,439 | Optimizing Tensor Programs on Flexible Storage | 2023 | SIGMOD |
| 8 | 4,974 | SPORES: Sum-Product Optimization via Relational Equality Saturation for Large Scale Linear Algebra | 2020 | VLDB |
| 9 | 6,169 | Automatic Optimization of Matrix Implementations for Distributed Machine Learning and Linear Algebra | 2021 | SIGMOD |
| 10 | 3,101 | On Optimizing Operator Fusion Plans for Large-Scale Machine Learning in SystemML | 2018 | VLDB |