DynSpAttn: Efficient Attention via Dual-Side Dynamic Sparsity on Sparse Tensor Cores

Xiangrui Yu, Ruibo Fan, ZeYu Li, Weile Luo, Gu Gong, Xiaowen Chu 0001. DynSpAttn: Efficient Attention via Dual-Side Dynamic Sparsity on Sparse Tensor Cores. In Proceedings of the 40th ACM International Conference on Supercomputing, ICS 2026, Belfast, United Kingdom, July 6-9, 2026. pages 1053-1064, ACM, 2026. [doi]

@inproceedings{YuFLLGC26,
  title = {DynSpAttn: Efficient Attention via Dual-Side Dynamic Sparsity on Sparse Tensor Cores},
  author = {Xiangrui Yu and Ruibo Fan and ZeYu Li and Weile Luo and Gu Gong and Xiaowen Chu 0001},
  year = {2026},
  doi = {10.1145/3797905.3800548},
  url = {https://doi.org/10.1145/3797905.3800548},
  researchr = {https://researchr.org/publication/YuFLLGC26},
  cites = {0},
  citedby = {0},
  pages = {1053-1064},
  booktitle = {Proceedings of the 40th ACM International Conference on Supercomputing, ICS 2026, Belfast, United Kingdom, July 6-9, 2026},
  publisher = {ACM},
  isbn = {979-8-4007-2522-7},
}