{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,5]],"date-time":"2026-06-05T13:04:48Z","timestamp":1780664688910,"version":"3.54.1"},"publisher-location":"New York, NY, USA","reference-count":39,"publisher":"ACM","license":[{"start":{"date-parts":[[2026,4,26]],"date-time":"2026-04-26T00:00:00Z","timestamp":1777161600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2026,4,27]]},"DOI":"10.1145\/3767295.3769364","type":"proceedings-article","created":{"date-parts":[[2026,4,24]],"date-time":"2026-04-24T20:20:04Z","timestamp":1777062004000},"page":"1707-1721","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["SAS: Sparse Attention Synthesizer for Efficient Language Model Inference"],"prefix":"10.1145","author":[{"ORCID":"https:\/\/orcid.org\/0009-0002-2144-8931","authenticated-orcid":false,"given":"Yuan","family":"Zhou","sequence":"first","affiliation":[{"name":"Amazon, Santa Clara, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6901-8837","authenticated-orcid":false,"given":"Shaojie","family":"Xiang","sequence":"additional","affiliation":[{"name":"Amazon, Santa Clara, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-7711-263X","authenticated-orcid":false,"given":"Lingfan","family":"Yu","sequence":"additional","affiliation":[{"name":"Amazon, Santa Clara, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2386-2271","authenticated-orcid":false,"given":"Zhenyu","family":"Song","sequence":"additional","affiliation":[{"name":"Amazon, Santa Clara, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8140-2321","authenticated-orcid":false,"given":"Charith","family":"Mendis","sequence":"additional","affiliation":[{"name":"Amazon, Santa Clara, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8165-840X","authenticated-orcid":false,"given":"Yida","family":"Wang","sequence":"additional","affiliation":[{"name":"Amazon, Santa Clara, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2026,4,26]]},"reference":[{"key":"e_1_3_2_1_1_1","volume-title":"Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774","author":"Achiam Josh","year":"2023","unstructured":"Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023."},{"key":"e_1_3_2_1_2_1","volume-title":"Sarathi: Efficient llm inference by piggybacking decodes with chunked prefills. arXiv preprint arXiv:2308.16369","author":"Agrawal Amey","year":"2023","unstructured":"Amey Agrawal, Ashish Panwar, Jayashree Mohan, Nipun Kwatra, Bhargav S Gulavani, and Ramachandran Ramjee. Sarathi: Efficient llm inference by piggybacking decodes with chunked prefills. arXiv preprint arXiv:2308.16369, 2023."},{"key":"e_1_3_2_1_3_1","volume-title":"Gqa: Training generalized multi-query transformer models from multi-head checkpoints. arXiv preprint arXiv:2305.13245","author":"Ainslie Joshua","year":"2023","unstructured":"Joshua Ainslie, James Lee-Thorp, Michiel de Jong, Yury Zemlyanskiy, Federico Lebr\u00f3n, and Sumit Sanghai. Gqa: Training generalized multi-query transformer models from multi-head checkpoints. arXiv preprint arXiv:2305.13245, 2023."},{"key":"e_1_3_2_1_4_1","volume-title":"Neuron compiler. https:\/\/awsdocs-neuron.readthedocs-hosted.com\/en\/latest\/compiler\/index.html","author":"Services Amazon Web","year":"2024","unstructured":"Amazon Web Services. Neuron compiler. https:\/\/awsdocs-neuron.readthedocs-hosted.com\/en\/latest\/compiler\/index.html, 2024."},{"key":"e_1_3_2_1_5_1","volume-title":"Neuron Kernel Interface (NKI) - Beta. https:\/\/awsdocs-neuron.readthedocs-hosted.com\/en\/latest\/general\/nki\/index.html","author":"Services Amazon Web","year":"2025","unstructured":"Amazon Web Services. Neuron Kernel Interface (NKI) - Beta. https:\/\/awsdocs-neuron.readthedocs-hosted.com\/en\/latest\/general\/nki\/index.html, 2025."},{"key":"e_1_3_2_1_6_1","volume-title":"Longformer: The long-document transformer. arXiv preprint arXiv:2004.05150","author":"Beltagy Iz","year":"2020","unstructured":"Iz Beltagy, Matthew E Peters, and Arman Cohan. Longformer: The long-document transformer. arXiv preprint arXiv:2004.05150, 2020."},{"key":"e_1_3_2_1_7_1","volume-title":"et al. Pyramidkv: Dynamic kv cache compression based on pyramidal information funneling. arXiv preprint arXiv:2406.02069","author":"Cai Zefan","year":"2024","unstructured":"Zefan Cai, Yichi Zhang, Bofei Gao, Yuliang Liu, Tianyu Liu, Keming Lu, Wayne Xiong, Yue Dong, Baobao Chang, Junjie Hu, et al. Pyramidkv: Dynamic kv cache compression based on pyramidal information funneling. arXiv preprint arXiv:2406.02069, 2024."},{"key":"e_1_3_2_1_8_1","volume-title":"et al. Magicpig: Lsh sampling for efficient llm generation. arXiv preprint arXiv:2410.16179","author":"Chen Zhuoming","year":"2024","unstructured":"Zhuoming Chen, Ranajoy Sadhukhan, Zihao Ye, Yang Zhou, Jianyu Zhang, Niklas Nolte, Yuandong Tian, Matthijs Douze, Leon Bottou, Zhihao Jia, et al. Magicpig: Lsh sampling for efficient llm generation. arXiv preprint arXiv:2410.16179, 2024."},{"key":"e_1_3_2_1_9_1","volume-title":"Generating long sequences with sparse transformers. arxiv","author":"Child Rewon","year":"2019","unstructured":"Rewon Child, Scott Gray, Alec Radford, and Ilya Sutskever. Generating long sequences with sparse transformers. arxiv 2019. arXiv preprint arXiv:1904.10509, 1904."},{"key":"e_1_3_2_1_10_1","volume-title":"Cuda c++ programming guide. https:\/\/docs.nvidia.com\/cuda\/cuda-c-programming-guide\/","author":"Nvidia Corporation","year":"2025","unstructured":"Nvidia Corporation. Cuda c++ programming guide. https:\/\/docs.nvidia.com\/cuda\/cuda-c-programming-guide\/, 2025."},{"key":"e_1_3_2_1_11_1","volume-title":"Flashattention-2: Faster attention with better parallelism and work partitioning. arXiv preprint arXiv:2307.08691","author":"Dao Tri","year":"2023","unstructured":"Tri Dao. Flashattention-2: Faster attention with better parallelism and work partitioning. arXiv preprint arXiv:2307.08691, 2023."},{"key":"e_1_3_2_1_12_1","volume-title":"Longnet: Scaling transformers to 1,000,000,000 tokens. arXiv preprint arXiv:2307.02486","author":"Ding Jiayu","year":"2023","unstructured":"Jiayu Ding, Shuming Ma, Li Dong, Xingxing Zhang, Shaohan Huang, Wenhui Wang, Nanning Zheng, and Furu Wei. Longnet: Scaling transformers to 1,000,000,000 tokens. arXiv preprint arXiv:2307.02486, 2023."},{"key":"e_1_3_2_1_13_1","volume-title":"Get more with less: Synthesizing recurrence with kv cache compression for efficient llm inference. arXiv preprint arXiv:2402.09398","author":"Dong Harry","year":"2024","unstructured":"Harry Dong, Xinyu Yang, Zhenyu Zhang, Zhangyang Wang, Yuejie Chi, and Beidi Chen. Get more with less: Synthesizing recurrence with kv cache compression for efficient llm inference. arXiv preprint arXiv:2402.09398, 2024."},{"key":"e_1_3_2_1_14_1","volume-title":"Flex attention: A programming model for generating optimized attention kernels. arXiv preprint arXiv:2412.05496","author":"Dong Juechu","year":"2024","unstructured":"Juechu Dong, Boyuan Feng, Driss Guessous, Yanbo Liang, and Horace He. Flex attention: A programming model for generating optimized attention kernels. arXiv preprint arXiv:2412.05496, 2024."},{"key":"e_1_3_2_1_15_1","volume-title":"The llama 3 herd of models. arXiv preprint arXiv:2407.21783","author":"Dubey Abhimanyu","year":"2024","unstructured":"Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. The llama 3 herd of models. arXiv preprint arXiv:2407.21783, 2024."},{"key":"e_1_3_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1109\/BigData62323.2024.10825098"},{"key":"e_1_3_2_1_17_1","doi-asserted-by":"publisher","DOI":"10.1145\/3698038.3698535"},{"key":"e_1_3_2_1_18_1","volume-title":"Splat: A framework for optimised gpu code-generation for sparse regular attention. arXiv preprint arXiv:2407.16847","author":"Gupta Ahan","year":"2024","unstructured":"Ahan Gupta, Yueming Yuan, Devansh Jain, Yuhao Ge, David Aponte, Yanqi Zhou, and Charith Mendis. Splat: A framework for optimised gpu code-generation for sparse regular attention. arXiv preprint arXiv:2407.16847, 2024."},{"key":"e_1_3_2_1_19_1","volume-title":"Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, et al. Mistral 7b. arXiv preprint arXiv:2310.06825","author":"Jiang Albert Q","year":"2023","unstructured":"Albert Q Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, et al. Mistral 7b. arXiv preprint arXiv:2310.06825, 2023."},{"key":"e_1_3_2_1_20_1","doi-asserted-by":"publisher","DOI":"10.1145\/3731569.3764849"},{"key":"e_1_3_2_1_21_1","first-page":"52481","article-title":"Minference 1.0: Accelerating pre-filling for long-context llms via dynamic sparse attention","volume":"37","author":"Jiang Huiqiang","year":"2024","unstructured":"Huiqiang Jiang, Yucheng Li, Chengruidong Zhang, Qianhui Wu, Xufang Luo, Surin Ahn, Zhenhua Han, Amir Abdi, Dongsheng Li, Chin-Yew Lin, et al. Minference 1.0: Accelerating pre-filling for long-context llms via dynamic sparse attention. Advances in Neural Information Processing Systems, 37:52481\u201352515, 2024.","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_3_2_1_22_1","volume-title":"Reformer: The efficient transformer. arXiv preprint arXiv:2001.04451","author":"Kitaev Nikita","year":"2020","unstructured":"Nikita Kitaev, \u0141ukasz Kaiser, and Anselm Levskaya. Reformer: The efficient transformer. arXiv preprint arXiv:2001.04451, 2020."},{"key":"e_1_3_2_1_23_1","first-page":"22947","article-title":"Llm knows what you are looking for before generation","volume":"37","author":"Li Yuhong","year":"2024","unstructured":"Yuhong Li, Yingbing Huang, Bowen Yang, Bharat Venkitesh, Acyr Locatelli, Hanchen Ye, Tianle Cai, Patrick Lewis, and Deming Chen. Snapkv: Llm knows what you are looking for before generation. Advances in Neural Information Processing Systems, 37:22947\u201322970, 2024.","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_3_2_1_24_1","volume-title":"Hackable and optimized transformers building blocks, supporting a composable construction. https:\/\/github.com\/facebookresearch\/xformers","year":"2024","unstructured":"Meta. Hackable and optimized transformers building blocks, supporting a composable construction. https:\/\/github.com\/facebookresearch\/xformers, 2024."},{"key":"e_1_3_2_1_25_1","volume-title":"Faster causal attention over large sequences through sparse flash attention. arXiv preprint arXiv:2306.01160","author":"Pagliardini Matteo","year":"2023","unstructured":"Matteo Pagliardini, Daniele Paliotta, Martin Jaggi, and Fran\u00e7ois Fleuret. Faster causal attention over large sequences through sparse flash attention. arXiv preprint arXiv:2306.01160, 2023."},{"key":"e_1_3_2_1_26_1","volume-title":"et al. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32","author":"Paszke Adam","year":"2019","unstructured":"Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32, 2019."},{"key":"e_1_3_2_1_27_1","volume-title":"https:\/\/triton-lang.org\/main\/getting-started\/tutorials\/06-fused-attention.html","author":"Tillet Philippe","year":"2025","unstructured":"Philippe Tillet. Fused Attention. https:\/\/triton-lang.org\/main\/getting-started\/tutorials\/06-fused-attention.html, 2025."},{"key":"e_1_3_2_1_28_1","doi-asserted-by":"publisher","DOI":"10.1145\/3394486.3406703"},{"key":"e_1_3_2_1_29_1","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00353"},{"key":"e_1_3_2_1_30_1","volume-title":"Aws trainium: Get high performance for deep learning and generative ai training while lowering costs. https:\/\/aws.amazon.com\/machine-learning\/trainium\/","author":"Services Amazon Web","year":"2024","unstructured":"Amazon Web Services. Aws trainium: Get high performance for deep learning and generative ai training while lowering costs. https:\/\/aws.amazon.com\/machine-learning\/trainium\/, 2024."},{"key":"e_1_3_2_1_31_1","volume-title":"Megatron-lm: Training multibillion parameter language models using model parallelism. arXiv preprint arXiv:1909.08053","author":"Shoeybi Mohammad","year":"2019","unstructured":"Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, and Bryan Catanzaro. Megatron-lm: Training multibillion parameter language models using model parallelism. arXiv preprint arXiv:1909.08053, 2019."},{"key":"e_1_3_2_1_32_1","volume-title":"Quest: Query-aware sparsity for efficient long-context llm inference. arXiv preprint arXiv:2406.10774","author":"Tang Jiaming","year":"2024","unstructured":"Jiaming Tang, Yilong Zhao, Kan Zhu, Guangxuan Xiao, Baris Kasikci, and Song Han. Quest: Query-aware sparsity for efficient long-context llm inference. arXiv preprint arXiv:2406.10774, 2024."},{"key":"e_1_3_2_1_33_1","doi-asserted-by":"publisher","DOI":"10.1145\/3315508.3329973"},{"key":"e_1_3_2_1_34_1","volume-title":"Advances in Neural Information Processing Systems","author":"Vaswani A","year":"2017","unstructured":"A Vaswani. Attention is all you need. Advances in Neural Information Processing Systems, 2017."},{"key":"e_1_3_2_1_35_1","volume-title":"Efficient streaming language models with attention sinks. arXiv preprint arXiv:2309.17453","author":"Xiao Guangxuan","year":"2023","unstructured":"Guangxuan Xiao, Yuandong Tian, Beidi Chen, Song Han, and Mike Lewis. Efficient streaming language models with attention sinks. arXiv preprint arXiv:2309.17453, 2023."},{"key":"e_1_3_2_1_36_1","doi-asserted-by":"publisher","DOI":"10.1145\/3582016.3582047"},{"key":"e_1_3_2_1_37_1","volume-title":"Native sparse attention: Hardware-aligned and natively trainable sparse attention. arXiv preprint arXiv:2502.11089","author":"Yuan Jingyang","year":"2025","unstructured":"Jingyang Yuan, Huazuo Gao, Damai Dai, Junyu Luo, Liang Zhao, Zhengyan Zhang, Zhenda Xie, YX Wei, Lean Wang, Zhiping Xiao, et al. Native sparse attention: Hardware-aligned and natively trainable sparse attention. arXiv preprint arXiv:2502.11089, 2025."},{"key":"e_1_3_2_1_38_1","unstructured":"Manzil Zaheer Guru Guruganesh Kumar Avinava Dubey Joshua Ainslie Chris Alberti Santiago Ontanon Philip Pham Anirudh Ravula Qifan Wang Li Yang et al. Big bird: Transformers for longer sequences. Advances in neural information processing systems 33:17283\u201317297 2020."},{"key":"e_1_3_2_1_39_1","first-page":"36","article-title":"H2o: Heavy-hitter oracle for efficient generative inference of large language models","author":"Zhang Zhenyu","year":"2024","unstructured":"Zhenyu Zhang, Ying Sheng, Tianyi Zhou, Tianlong Chen, Lianmin Zheng, Ruisi Cai, Zhao Song, Yuandong Tian, Christopher R\u00e9, Clark Barrett, et al. H2o: Heavy-hitter oracle for efficient generative inference of large language models. Advances in Neural Information Processing Systems, 36, 2024.","journal-title":"Advances in Neural Information Processing Systems"}],"event":{"name":"EUROSYS '26: 21st European Conference on Computer Systems","location":"McEwan Hall\/The University of Edinburgh Edinburgh Scotland UK","acronym":"EUROSYS '26","sponsor":["SIGOPS ACM Special Interest Group on Operating Systems"]},"container-title":["Proceedings of the 21st European Conference on Computer Systems"],"original-title":[],"link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3767295.3769364","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,5]],"date-time":"2026-06-05T12:16:32Z","timestamp":1780661792000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3767295.3769364"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,26]]},"references-count":39,"alternative-id":["10.1145\/3767295.3769364","10.1145\/3767295"],"URL":"https:\/\/doi.org\/10.1145\/3767295.3769364","relation":{},"subject":[],"published":{"date-parts":[[2026,4,26]]},"assertion":[{"value":"2026-04-26","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}