{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,1]],"date-time":"2026-04-01T18:50:19Z","timestamp":1775069419171,"version":"3.50.1"},"reference-count":74,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"3","license":[{"start":{"date-parts":[[2024,6,1]],"date-time":"2024-06-01T00:00:00Z","timestamp":1717200000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2024,6,1]],"date-time":"2024-06-01T00:00:00Z","timestamp":1717200000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2024,6,1]],"date-time":"2024-06-01T00:00:00Z","timestamp":1717200000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Science Foundation of China","doi-asserted-by":"publisher","award":["62372426"],"award-info":[{"award-number":["62372426"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Science Foundation of China","doi-asserted-by":"publisher","award":["62102392"],"award-info":[{"award-number":["62102392"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Open Research of Projects of Zhejiang Laboratory","award":["2022QA0AB04"],"award-info":[{"award-number":["2022QA0AB04"]}]},{"DOI":"10.13039\/501100004608","name":"National Science Foundation of Jiangsu Province","doi-asserted-by":"publisher","award":["BK20210121"],"award-info":[{"award-number":["BK20210121"]}],"id":[{"id":"10.13039\/501100004608","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Hefei Municipal Natural Science Foundation","award":["2022013"],"award-info":[{"award-number":["2022013"]}]},{"DOI":"10.13039\/501100002367","name":"Youth Innovation Promotion Association of Chinese Academy of Sciences","doi-asserted-by":"publisher","award":["2023481"],"award-info":[{"award-number":["2023481"]}],"id":[{"id":"10.13039\/501100002367","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100012226","name":"Fundamental Research Funds for the Central Universities","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100012226","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE\/ACM Trans. Networking"],"published-print":{"date-parts":[[2024,6]]},"DOI":"10.1109\/tnet.2023.3339524","type":"journal-article","created":{"date-parts":[[2023,12,14]],"date-time":"2023-12-14T20:06:06Z","timestamp":1702584366000},"page":"2174-2188","source":"Crossref","is-referenced-by-count":8,"title":["XAgg: Accelerating Heterogeneous Distributed Training Through XDP-Based Gradient Aggregation"],"prefix":"10.1109","volume":"32","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-3805-4480","authenticated-orcid":false,"given":"Qianyu","family":"Zhang","sequence":"first","affiliation":[{"name":"School of Computer Science and Technology, University of Science and Technology of China, Hefei, Anhui, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1311-8908","authenticated-orcid":false,"given":"Gongming","family":"Zhao","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, University of Science and Technology of China, Hefei, Anhui, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3831-4577","authenticated-orcid":false,"given":"Hongli","family":"Xu","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, University of Science and Technology of China, Hefei, Anhui, China"}]},{"given":"Peng","family":"Yang","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, University of Science and Technology of China, Hefei, Anhui, China"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1706.03762"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1145\/3285029"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.97"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/2640087.2644155"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1145\/3152434.3152461"},{"key":"ref7","volume-title":"NVIDIA Ampere Architecture in-Depth","year":"2020"},{"key":"ref8","volume-title":"Cloud TPU: Train and Run Machine Learning Models Faster Than Ever Before","year":"2022"},{"key":"ref9","volume-title":"High Performance Training and Inference","year":"2022"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1145\/3267809.3267840"},{"key":"ref11","volume-title":"NVIDIA Data Center Deep Learning Product Performance","year":"2022"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/3452296.3472904"},{"key":"ref13","first-page":"785","article-title":"Scaling distributed machine learning with in-network aggregation","volume-title":"Proc. 18th USENIX Symp. Networked Syst. Design Implement.","author":"Sapio"},{"key":"ref14","article-title":"Deep gradient compression: Reducing the communication bandwidth for distributed training","author":"Lin","year":"2017","journal-title":"arXiv:1712.01887"},{"issue":"1","key":"ref15","first-page":"6869","article-title":"Quantized neural networks: Training neural networks with low precision weights and activations","volume":"18","author":"Hubara","year":"2017","journal-title":"J. Mach. Learn. Res."},{"key":"ref16","first-page":"1508","article-title":"TernGrad: Ternary gradients to reduce communication in distributed deep learning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"30","author":"Wen"},{"key":"ref17","first-page":"418","article-title":"TicTac: Accelerating distributed deep learning with communication scheduling","volume-title":"Proc. Mach. Learn. Syst.","volume":"1","author":"Hashemi"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359642"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM41043.2020.9155282"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.48550\/arxiv.1811.06965"},{"key":"ref21","first-page":"463","article-title":"A unified architecture for accelerating distributed DNN training in heterogeneous GPU\/CPU clusters","volume-title":"Proc. 14th USENIX Conf. Operating Syst. Design Implement.","author":"Jiang"},{"key":"ref22","first-page":"741","article-title":"ATP: In-network aggregation for multi-tenant learning","volume-title":"Proc. NSDI","author":"Lao"},{"key":"ref23","volume-title":"Intel Tofino","year":"2022"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2019.2897898"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1145\/3326285.3329074"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1145\/3452296.3472888"},{"key":"ref27","first-page":"43","article-title":"StackMap: Low-latency networking with the OS stack and dedicated NICs","volume-title":"Proc. USENIX Annu. Tech. Conf.","author":"Yasukata"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1145\/3050220.3050236"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1145\/3281411.3281443"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1145\/3371038"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1145\/633025.633045"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/INFCOM.2010.5462038"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1145\/3419394.3423637"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/NetGames.2014.7008960"},{"key":"ref35","volume-title":"Data Plane Development Kit","year":"2022"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/ANCS.2015.7110118"},{"key":"ref37","first-page":"8026","article-title":"Pytorch: An imperative style, high-performance deep learning library","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"32","author":"Paszke"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/JSYST.2019.2918446"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1145\/1374376.1374485"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1145\/1806689.1806769"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/HOTI.2013.23"},{"key":"ref42","first-page":"373","article-title":"Andromeda: Performance, isolation, and velocity at scale in cloud network virtualization","volume-title":"Proc. 15th USENIX Symp. Networked Syst. design Implement.","author":"Dalton"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM42981.2021.9488756"},{"key":"ref44","volume-title":"Iftop: Display Bandwidth Usage on an Interface","year":"2022"},{"key":"ref45","volume-title":"NVIDIA Collective Communications Library (NCCL)","year":"2023"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.308"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-015-0816-y"},{"key":"ref48","article-title":"Very deep convolutional networks for large-scale image recognition","author":"Simonyan","year":"2014","journal-title":"arXiv:1409.1556"},{"key":"ref49","volume-title":"The Cifar-10 Dataset","author":"Krizhevsky","year":"2022"},{"key":"ref50","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","author":"Devlin","year":"2018","journal-title":"arXiv:1810.04805"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/CISP-BMEI48845.2019.8965823"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1109\/TKDE.2020.2981314"},{"key":"ref53","article-title":"CLUENER2020: Fine-grained named entity recognition dataset and benchmark for Chinese","author":"Xu","year":"2020","journal-title":"arXiv:2001.04351"},{"key":"ref54","volume-title":"BPF Portability and CO-RE","author":"Nakryiko","year":"2020"},{"key":"ref55","article-title":"Communication-efficient distributed deep learning: A comprehensive survey","author":"Tang","year":"2020","journal-title":"arXiv:2003.06307"},{"key":"ref56","article-title":"Megatron-LM: Training multi-billion parameter language models using model parallelism","author":"Shoeybi","year":"2019","journal-title":"arXiv:1909.08053"},{"key":"ref57","article-title":"Measuring the effects of data parallelism on neural network training","author":"Shallue","year":"2018","journal-title":"arXiv:1811.03600"},{"key":"ref58","article-title":"Horovod: Fast and easy distributed deep learning in TensorFlow","author":"Sergeev","year":"2018","journal-title":"arXiv:1802.05799"},{"key":"ref59","article-title":"Bandwidth reduction using importance weighted pruning on ring AllReduce","author":"Cheng","year":"2019","journal-title":"arXiv:1901.01544"},{"key":"ref60","article-title":"Highly scalable deep learning training system with mixed-precision: Training ImageNet in four minutes","author":"Jia","year":"2018","journal-title":"arXiv:1807.11205"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1145\/3229543.3229544"},{"key":"ref62","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2021.3062721"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1145\/2185448.2185474"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.1145\/2674005.2674996"},{"key":"ref65","first-page":"279","article-title":"Accelerating distributed reinforcement learning with in-switch computing","volume-title":"Proc. ACM\/IEEE 46th Annu. Int. Symp. Comput. Archit. (ISCA)","author":"Li"},{"key":"ref66","first-page":"101","article-title":"netmap: A novel framework for fast packet I\/O","volume-title":"Proc. 21st USENIX Secur. Symp.","author":"Rizzo"},{"key":"ref67","volume-title":"BPF and XDP Reference Guide","year":"2022"},{"key":"ref68","volume-title":"Cloud Native Runtime Security","year":"2022"},{"key":"ref69","volume-title":"Netdata: Monitoring and Troubleshooting Transformed","year":"2022"},{"key":"ref70","first-page":"487","article-title":"BMC: Accelerating memcached using safe in-kernel caching and pre-stack processing","volume-title":"Proc. 18th USENIX Symp. Networked Syst. Design Implement.","author":"Ghigoff"},{"key":"ref71","first-page":"375","article-title":"XPR: In-kernel storage functions with eBPF","volume-title":"Proc. 16th USENIX Symp. Operating Syst. Design Implement.","author":"Zhong"},{"key":"ref72","article-title":"MXNet: A flexible and efficient machine learning library for heterogeneous distributed systems","author":"Chen","year":"2015","journal-title":"arXiv:1512.01274"},{"key":"ref73","first-page":"1223","article-title":"Large scale distributed deep networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"25","author":"Dean"},{"key":"ref74","doi-asserted-by":"publisher","DOI":"10.5555\/3026877.3026899"}],"container-title":["IEEE\/ACM Transactions on Networking"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/90\/10559910\/10360433.pdf?arnumber=10360433","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,9,3]],"date-time":"2024-09-03T17:44:16Z","timestamp":1725385456000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10360433\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,6]]},"references-count":74,"journal-issue":{"issue":"3"},"URL":"https:\/\/doi.org\/10.1109\/tnet.2023.3339524","relation":{},"ISSN":["1063-6692","1558-2566"],"issn-type":[{"value":"1063-6692","type":"print"},{"value":"1558-2566","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,6]]}}}