{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,6,19]],"date-time":"2025-06-19T05:05:48Z","timestamp":1750309548975,"version":"3.41.0"},"reference-count":12,"publisher":"Association for Computing Machinery (ACM)","issue":"3","license":[{"start":{"date-parts":[[2024,10,22]],"date-time":"2024-10-22T00:00:00Z","timestamp":1729555200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["GetMobile: Mobile Comp. and Comm."],"published-print":{"date-parts":[[2024,10,22]]},"abstract":"<jats:p>Deploying deep learning (DL) on mobile devices has become increasingly prevalent. DL software libraries are crucial for efficient on-device inference, alongside algorithms and hardware. However, there has been limited understanding on the performance of modern DL libraries. We fill this gap by benchmarking 6 popular DL libraries and 15 diverse models across 10 mobile devices, which reveal an unsatisfactory landscape of mobile DL: their performance is highly disparate and fragmented across different models and hardware, and the impacts often surpass algorithm or hardware optimizations, such as model quantization and GPU\/NPU-based computing. Finally, we provide practical implications for stakeholders in the DL library ecosystem, and envision a more ambitious picture of future mobile AI landscape in the LLM era.<\/jats:p>","DOI":"10.1145\/3701701.3701703","type":"journal-article","created":{"date-parts":[[2024,10,22]],"date-time":"2024-10-22T22:26:35Z","timestamp":1729635995000},"page":"5-8","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Benchmarking Mobile Deep Learning Software"],"prefix":"10.1145","volume":"28","author":[{"given":"Qiyang","family":"Zhang","sequence":"first","affiliation":[{"name":"Peking University, Beijing, China"}]},{"given":"Mengwei","family":"Xu","sequence":"additional","affiliation":[{"name":"Beijing University of Posts and Telecommunications (BUPT), Beijing, China"}]}],"member":"320","published-online":{"date-parts":[[2024,10,22]]},"reference":[{"key":"e_1_2_1_1_1","doi-asserted-by":"crossref","unstructured":"E. Charles Leiserson et al. 2020. There's plenty of room at the top: What will drive computer performance after Moore's law? Science.","DOI":"10.1126\/science.aam9744"},{"key":"e_1_2_1_2_1","doi-asserted-by":"publisher","DOI":"10.1109\/TSE.2018.2876439"},{"key":"e_1_2_1_3_1","doi-asserted-by":"publisher","DOI":"10.1145\/3485447.3512148"},{"key":"e_1_2_1_4_1","doi-asserted-by":"publisher","DOI":"10.1145\/2568058.2568064"},{"key":"e_1_2_1_5_1","unstructured":"Yuhang Li et al. 2020. Efficient bitwidth search for practical mixed precision neural network. arXiv preprint arXiv:2003.07577."},{"key":"e_1_2_1_6_1","unstructured":"Qualcomm Hexagon. 2021. https:\/\/en.wikipedia. org\/wiki\/Qualcomm_Hexagon."},{"key":"e_1_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1145\/3437801.3441588"},{"key":"e_1_2_1_8_1","volume-title":"Proceedings of 1st Information Technology Data Science.","author":"Tornai Robert","year":"2021","unstructured":"Robert Tornai, and P\u00e9ter F\u00fcrjes-Benke. 2021. Compute shader in image processing development. Proceedings of 1st Information Technology Data Science."},{"key":"e_1_2_1_9_1","unstructured":"XiaoMi. 2017. Mace Deep Learning Framework. https:\/\/github.com\/XiaoMi\/"},{"key":"e_1_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1145\/3581791.3596842"},{"key":"e_1_2_1_11_1","doi-asserted-by":"publisher","DOI":"10.1145\/3636534.3649361"},{"key":"e_1_2_1_12_1","unstructured":"Google AI Edge SDK for Gemini Nano. 2024. https:\/\/developer.android.com\/ai\/aicore"}],"container-title":["GetMobile: Mobile Computing and Communications"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3701701.3701703","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3701701.3701703","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,19]],"date-time":"2025-06-19T01:18:44Z","timestamp":1750295924000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3701701.3701703"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,10,22]]},"references-count":12,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2024,10,22]]}},"alternative-id":["10.1145\/3701701.3701703"],"URL":"https:\/\/doi.org\/10.1145\/3701701.3701703","relation":{},"ISSN":["2375-0529","2375-0537"],"issn-type":[{"type":"print","value":"2375-0529"},{"type":"electronic","value":"2375-0537"}],"subject":[],"published":{"date-parts":[[2024,10,22]]},"assertion":[{"value":"2024-10-22","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}