{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,11]],"date-time":"2026-06-11T05:59:16Z","timestamp":1781157556004,"version":"3.54.1"},"reference-count":28,"publisher":"IEEE","license":[{"start":{"date-parts":[[2026,5,13]],"date-time":"2026-05-13T00:00:00Z","timestamp":1778630400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,5,13]],"date-time":"2026-05-13T00:00:00Z","timestamp":1778630400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,5,13]]},"DOI":"10.1109\/fccm68464.2026.00038","type":"proceedings-article","created":{"date-parts":[[2026,6,10]],"date-time":"2026-06-10T19:59:45Z","timestamp":1781121585000},"page":"204-212","source":"Crossref","is-referenced-by-count":0,"title":["LegoMap: Optimization for High-Throughput Transformer Computing on AI Engine-Based FPGAs"],"prefix":"10.1109","author":[{"given":"Hailiang","family":"Hu","sequence":"first","affiliation":[{"name":"Texas A&M University,College Station,TX,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Haodong","family":"Chang","sequence":"additional","affiliation":[{"name":"Texas A&M University,College Station,TX,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Donghao","family":"Fang","sequence":"additional","affiliation":[{"name":"Texas A&M University,College Station,TX,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhenrui","family":"Wang","sequence":"additional","affiliation":[{"name":"Texas A&M University,College Station,TX,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wuxi","family":"Li","sequence":"additional","affiliation":[{"name":"AMD,San Jose,CA,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rongjian","family":"Liang","sequence":"additional","affiliation":[{"name":"NVIDIA,Austin,TX,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bo","family":"Yuan","sequence":"additional","affiliation":[{"name":"Rutgers University,Piscataway,NJ,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jiang","family":"Hu","sequence":"additional","affiliation":[{"name":"Texas A&M University,College Station,TX,USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Attention is all you need","author":"Vaswani","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref2","article-title":"Cross-lingual language model pretraining","author":"Conneau","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.18653\/vl\/N19-142"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2010.11929"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3289602.3293906"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1145\/3543622.3573210"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3686163"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/DAC56929.2023.10247981"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/ICFPT59805.2023.00016"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1145\/3695053.3731088"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2024.3443692"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/3626202.3637569"},{"key":"ref13","article-title":"Compute and energy consumption trends in deep learning inference","author":"Desislavov","year":"2021"},{"key":"ref14","article-title":"On the computational complexity of self-attention","volume-title":"International Conference on Algorithmic Learning Theory","author":"Keles"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/WACV48630.2021.00357"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO50266.2020.00071"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1189"},{"key":"ref18","article-title":"FlashAttention-2: Faster attention with better parallelism and work partitioning","author":"Dao","year":"2023"},{"key":"ref19","article-title":"Online normalizer calculation for softmax","author":"Milakov","year":"2018"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1145\/3575693.3575747"},{"key":"ref21","doi-asserted-by":"crossref","DOI":"10.1145\/3613424.3623792","article-title":"Tileflow: A framework for modeling fusion dataflow via tree-based analysis","volume-title":"International Symposium on Microarchitecture","author":"Zheng"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA56546.2023.10071018"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/iccv48922.2021.00064"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref25","article-title":"Llama: Open and efficient foundation language models","author":"Touvron","year":"2023"},{"key":"ref26","article-title":"Gurobi Optimizer Reference Manual","year":"2024"},{"key":"ref27","article-title":"Charm: Composing heterogeneous accelerators for matrix multiply on versal acap architecture","author":"Zhuang","year":"2023"},{"key":"ref28","doi-asserted-by":"crossref","DOI":"10.1109\/ICFPT59805.2023.00016","article-title":"Maxeva: Maximizing the efficiency of matrix multiplication on versal ai engine","author":"Taka","year":"2023"}],"event":{"name":"2026 IEEE 34th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM)","location":"Atlanta, GA, USA","start":{"date-parts":[[2026,5,13]]},"end":{"date-parts":[[2026,5,16]]}},"container-title":["2026 IEEE 34th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11552597\/11552602\/11552655.pdf?arnumber=11552655","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,11]],"date-time":"2026-06-11T05:22:19Z","timestamp":1781155339000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11552655\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,13]]},"references-count":28,"URL":"https:\/\/doi.org\/10.1109\/fccm68464.2026.00038","relation":{},"subject":[],"published":{"date-parts":[[2026,5,13]]}}}