{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,4,5]],"date-time":"2025-04-05T04:09:30Z","timestamp":1743826170884,"version":"3.40.3"},"reference-count":25,"publisher":"IEEE","license":[{"start":{"date-parts":[[2024,9,23]],"date-time":"2024-09-23T00:00:00Z","timestamp":1727049600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2024,9,23]],"date-time":"2024-09-23T00:00:00Z","timestamp":1727049600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/100000001","name":"NSF","doi-asserted-by":"publisher","award":["OAC-2402542"],"award-info":[{"award-number":["OAC-2402542"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2024,9,23]]},"DOI":"10.1109\/hpec62836.2024.10938425","type":"proceedings-article","created":{"date-parts":[[2025,4,3]],"date-time":"2025-04-03T19:07:19Z","timestamp":1743707239000},"page":"1-7","source":"Crossref","is-referenced-by-count":0,"title":["High Performance Im2win and Direct Convolutions Using Three Tensor Layouts on SIMD Architectures"],"prefix":"10.1109","author":[{"given":"Xiang","family":"Fu","sequence":"first","affiliation":[{"name":"Nanchang Hangkong University,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xinpeng","family":"Zhang","sequence":"additional","affiliation":[{"name":"Nanchang Hangkong University,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jixiang","family":"Ma","sequence":"additional","affiliation":[{"name":"Nanchang Hangkong University,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Peng","family":"Zhao","sequence":"additional","affiliation":[{"name":"Microsoft,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Shuai","family":"Lu","sequence":"additional","affiliation":[{"name":"Nanchang Hangkong University,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xu T.","family":"Liu","sequence":"additional","affiliation":[{"name":"University of Washington,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/SMC.2018.00519"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01264-9_8"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2017.2761740"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAE59995.2024.10569297"},{"key":"ref5","article-title":"Prediction of car purchase amount based on genetic algorithm optimised bp neural network regression algorithm","volume-title":"Preprints","author":"Chen","year":"2024"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2018.00069"},{"key":"ref7","article-title":"High performance convolutional neural networks for document processing","volume-title":"10th Int\u2019l Workshop on Frontiers in Handwriting Recog.","author":"Chellapilla","year":"2006"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2021.3135690"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1145\/77626.79170"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/HPEC55821.2022.9926408"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-39698-4_40"},{"key":"ref12","first-page":"8024","article-title":"Pytorch: An imperative style, high-performance deep learning library","volume-title":"Proceedings of the 33rd International Conference on Neural Information Processing Systems (NeurIPS)","author":"Paszke","year":"2019"},{"key":"ref13","first-page":"265","article-title":"Tensorflow: A system for large-scale machine learning","volume-title":"Proceedings of the 12th USENIX Conference on Operating Systems Design and Implementation, ser. OSDI\u201916","author":"Abadi","year":"2016"},{"key":"ref14","article-title":"cuDNN: Efficient primitives for deep learning","volume":"abs\/1410.0759","author":"Chetlur","year":"2014","journal-title":"CoRR"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2016.53"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1145\/3487553.3524856"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1145\/3625004"},{"key":"ref18","first-page":"5776","article-title":"High performance zero-memory overhead direct convolutions","volume-title":"International Conference on Machine Learning","author":"Zhang","year":"2018"},{"key":"ref19","first-page":"815","article-title":"MEC: memory-efficient convolution for deep neural network","volume-title":"International Conference on Machine Learning (ICML)","author":"Cho","year":"2017"},{"issue":"4","key":"ref20","doi-asserted-by":"crossref","first-page":"65","DOI":"10.1145\/1498765.1498785","article-title":"Roofline: an insightful visual performance model for multicore architectures","volume":"52","author":"Williams","year":"2009","journal-title":"Commun. ACM"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2018.00069"},{"key":"ref22","first-page":"357","article-title":"Iteration space tiling for memory hierarchies","volume-title":"Proceedings of the Third SIAM Conference on Parallel Processing for Scientific Computing","author":"Wolfe","year":"1987"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1145\/76263.76337"},{"volume-title":"PyTorch","year":"2024","key":"ref24"},{"volume-title":"MKL","year":"2024","key":"ref25"}],"event":{"name":"2024 IEEE High Performance Extreme Computing Conference (HPEC)","start":{"date-parts":[[2024,9,23]]},"location":"Wakefield, MA, USA","end":{"date-parts":[[2024,9,27]]}},"container-title":["2024 IEEE High Performance Extreme Computing Conference (HPEC)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/10938401\/10938415\/10938425.pdf?arnumber=10938425","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,4,4]],"date-time":"2025-04-04T11:30:31Z","timestamp":1743766231000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10938425\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,9,23]]},"references-count":25,"URL":"https:\/\/doi.org\/10.1109\/hpec62836.2024.10938425","relation":{},"subject":[],"published":{"date-parts":[[2024,9,23]]}}}