{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,5]],"date-time":"2026-06-05T06:01:17Z","timestamp":1780639277106,"version":"3.54.1"},"reference-count":35,"publisher":"IEEE","license":[{"start":{"date-parts":[[2026,4,20]],"date-time":"2026-04-20T00:00:00Z","timestamp":1776643200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,4,20]],"date-time":"2026-04-20T00:00:00Z","timestamp":1776643200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,4,20]]},"DOI":"10.23919\/date69613.2026.11539662","type":"proceedings-article","created":{"date-parts":[[2026,6,4]],"date-time":"2026-06-04T19:53:10Z","timestamp":1780602790000},"page":"1-7","source":"Crossref","is-referenced-by-count":0,"title":["From Buffers to Registers: Unlocking Fine-Grained FlashAttention with Hybrid-Bonded 3D NPU Co-Design"],"prefix":"10.23919","author":[{"given":"Jinxin","family":"Yu","sequence":"first","affiliation":[{"name":"SKLP, Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yudong","family":"Pan","sequence":"additional","affiliation":[{"name":"SKLP, Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mengdi","family":"Wang","sequence":"additional","affiliation":[{"name":"SKLP, Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Huawei","family":"Li","sequence":"additional","affiliation":[{"name":"SKLP, Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yinhe","family":"Han","sequence":"additional","affiliation":[{"name":"SKLP, Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaowei","family":"Li","sequence":"additional","affiliation":[{"name":"SKLP, Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ying","family":"Wang","sequence":"additional","affiliation":[{"name":"SKLP, Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Language models are unsupervised multitask learners","author":"Radford","year":"2019"},{"key":"ref2","article-title":"Language models are few-shot learners","author":"Brown","year":"2020"},{"key":"ref3","article-title":"Llama: Open and efficient foundation language models","author":"Touvron","year":"2023"},{"key":"ref4","article-title":"Deepseek-v3 technical report","author":"Liu","year":"2024"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2016.2616357"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1145\/2749469.2750389"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA51647.2021.00071"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1145\/3297858.3304014"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1189"},{"key":"ref10","article-title":"Flashattention-2: Faster attention with better parallelism and work partitioning","author":"Dao","year":"2023"},{"key":"ref11","article-title":"Flashattention-3: Fast and accurate attention with asynchrony and low-precision","author":"Shah","year":"2024"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC.2014.6757323"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1109\/EDTM.2019.8731186"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1115\/1.4064750"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/DAC56929.2023.10247678"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO61859.2024.00107"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.3390\/app13148301"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/IMW.2017.7939084"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC42614.2022.9731565"},{"key":"ref20","article-title":"Semiconductor and ic package thermal metrics","year":"2025"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijheatmasstransfer.2024.126212"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/ISCAS48785.2022.9937238"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/LCA.2023.3333759"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD45719.2019.8942149"},{"key":"ref25","article-title":"Design compiler rtl synthesis","year":"2025"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/DAC56929.2023.10247815"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1109\/JXCDC.2021.3092436"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1145\/3575693.3575747"},{"key":"ref29","first-page":"1271","article-title":"Tileflow: A framework for modeling fusion dataflow via tree-based analysis","volume-title":"Proceedings of the 56th Annual IEEE\/ACM International Symposium on Microarchitecture","author":"Zheng"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1109\/ISQED51717.2021.9424349"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/SiPS.2018.8598454"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/ISPASS.2019.00042"},{"key":"ref33","article-title":"Opt: Open pre-trained transformer language models","author":"Zhang","year":"2022"},{"key":"ref34","article-title":"Qwen technical report","author":"Bai","year":"2023"},{"key":"ref35","article-title":"Longrope2: Near-lossless llm context window scaling","author":"Shang","year":"2025"}],"event":{"name":"2026 Design, Automation & Test in Europe Conference (DATE)","location":"Verona, Italy","start":{"date-parts":[[2026,4,20]]},"end":{"date-parts":[[2026,4,22]]}},"container-title":["2026 Design, Automation & Test in Europe Conference (DATE)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11539023\/11539024\/11539662.pdf?arnumber=11539662","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,5]],"date-time":"2026-06-05T05:11:14Z","timestamp":1780636274000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11539662\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,20]]},"references-count":35,"URL":"https:\/\/doi.org\/10.23919\/date69613.2026.11539662","relation":{},"subject":[],"published":{"date-parts":[[2026,4,20]]}}}