{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,4]],"date-time":"2026-06-04T11:58:18Z","timestamp":1780574298347,"version":"3.54.1"},"reference-count":67,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"8","license":[{"start":{"date-parts":[[2024,8,1]],"date-time":"2024-08-01T00:00:00Z","timestamp":1722470400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2024,8,1]],"date-time":"2024-08-01T00:00:00Z","timestamp":1722470400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2024,8,1]],"date-time":"2024-08-01T00:00:00Z","timestamp":1722470400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"name":"National Key R&D Program of China","award":["2021YFB0301200"],"award-info":[{"award-number":["2021YFB0301200"]}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62025208"],"award-info":[{"award-number":["62025208"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62302512"],"award-info":[{"award-number":["62302512"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Parallel Distrib. Syst."],"published-print":{"date-parts":[[2024,8]]},"DOI":"10.1109\/tpds.2024.3406420","type":"journal-article","created":{"date-parts":[[2024,5,28]],"date-time":"2024-05-28T18:15:35Z","timestamp":1716920135000},"page":"1415-1428","source":"Crossref","is-referenced-by-count":14,"title":["A Multidimensional Communication Scheduling Method for Hybrid Parallel DNN Training"],"prefix":"10.1109","volume":"35","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-7419-1511","authenticated-orcid":false,"given":"Shengwei","family":"Li","sequence":"first","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6378-7002","authenticated-orcid":false,"given":"Kai","family":"Lu","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3458-4732","authenticated-orcid":false,"given":"Zhiquan","family":"Lai","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0354-9467","authenticated-orcid":false,"given":"Weijie","family":"Liu","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0669-6892","authenticated-orcid":false,"given":"Keshi","family":"Ge","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9743-2034","authenticated-orcid":false,"given":"Dongsheng","family":"Li","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1706.03762"},{"issue":"8","key":"ref2","first-page":"9","article-title":"Language models are unsupervised multitask learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI Blog"},{"key":"ref3","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","author":"Devlin","year":"2018"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV48922.2021.00986"},{"key":"ref5","article-title":"An image is worth 16x16 words: Transformers for image recognition at scale","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Dosovitskiy","year":"2021"},{"key":"ref6","article-title":"Machine learning model sizes and the parameter gap","author":"Villalobos","year":"2022"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2020.3046870"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.14778\/3415478.3415530"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.14778\/3503585.3503590"},{"key":"ref10","article-title":"GPipe: Efficient training of giant neural networks using pipeline parallelism","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"32","author":"Huang","year":"2019"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359646"},{"key":"ref12","article-title":"Megatron-LM: Training multi-billion parameter language models using model parallelism","author":"Shoeybi","year":"2019"},{"key":"ref13","first-page":"1","article-title":"Efficient large-scale language model training on GPU clusters using megatron-LM","volume-title":"Proc. Int. Conf. High Perform. Comput. Netw. Storage Anal.","author":"Narayanan","year":"2021"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1145\/3437801.3441593"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/CLUSTER51413.2022.00042"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476145"},{"key":"ref17","first-page":"132","article-title":"Priority-based parameter propagation for distributed DNN training","volume-title":"Proc. Mach. Learn. Syst.","volume":"1","author":"Jayarajan","year":"2019"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359642"},{"key":"ref19","article-title":"Decoupling the all-reduce primitive for accelerating distributed deep learning","author":"Zhang","year":"2023"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM48880.2022.9796752"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM41043.2020.9155446"},{"key":"ref22","first-page":"463","article-title":"A unified architecture for accelerating distributed DNN training in heterogeneous GPU\/CPU clusters","volume-title":"Proc. 14th USENIX Symp. Operating Syst. Des. Implementation","author":"Jiang","year":"2020"},{"key":"ref23","first-page":"809","article-title":"Better together: Jointly optimizing ML collective scheduling and execution planning using SYNDICATE","volume-title":"Proc. 20th USENIX Symp. Networked Syst. Des. Implementation","author":"Mahajan","year":"2023"},{"key":"ref24","first-page":"56","article-title":"Logical\/physical topology-aware collective communication in deep learning training","volume-title":"Proc. IEEE Int. Symp. High- Perform. Comput. Archit.","author":"Sanghoon","year":"2023"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1145\/3470496.3527382"},{"key":"ref26","first-page":"241","article-title":"BlueConnect: Decomposing All-Reduce for Deep Learning on Heterogeneous Network Hierarchy","volume-title":"Proc. Mach. Learn. Syst.","volume":"1","author":"Cho","year":"2019"},{"issue":"140","key":"ref27","first-page":"1","article-title":"Exploring the limits of transfer learning with a unified text-to-text transformer","volume":"21","author":"Raffel","year":"2020","journal-title":"J. Mach. Learn. Res."},{"key":"ref28","article-title":"Using deepspeed and megatron to train megatron-turing NLG 530B, a large-scale generative language model","author":"Smith","year":"2022"},{"key":"ref29","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Brown","year":"2020"},{"key":"ref30","article-title":"Horovod: Fast and easy distributed deep learning in tensorflow","author":"Sergeev","year":"2018"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1016\/j.jpdc.2008.09.002"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1145\/3472456.3472497"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/CLUSTER51413.2022.00043"},{"key":"ref34","first-page":"7937","article-title":"Memory-efficient pipeline-parallel DNN training","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Narayanan","year":"2021"},{"key":"ref35","first-page":"307","article-title":"HetPipe: Enabling large DNN training on (whimpy) heterogeneous GPU clusters through integration of pipelined model parallelism and data parallelism","volume-title":"Proc. USENIX Annu. Tech. Conf","author":"Park","year":"2020"},{"key":"ref36","first-page":"381","article-title":"Fine-tuning giant neural networks on commodity hardware with automatic pipeline model parallelism","volume-title":"Proc. USENIX Annu. Tech. Conf.","author":"Eliad","year":"2021"},{"key":"ref37","first-page":"479","article-title":"Pipelined backpropagation at scale: Training large models without batches","volume-title":"Proc. Mach. Learn. Syst.","volume":"3","author":"Kosson","year":"2021"},{"key":"ref38","first-page":"181","article-title":"Poseidon: An efficient communication architecture for distributed deep learning on GPU clusters","volume-title":"Proc. USENIX Annu. Tech. Conf.","author":"Zhang","year":"2017"},{"key":"ref39","first-page":"8026","article-title":"PyTorch: An imperative style, high-performance deep learning library","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Paszke","year":"2019"},{"key":"ref40","article-title":"TicTac: Accelerating distributed deep learning with communication scheduling","author":"Hashemi","year":"2018"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2023.3247883"},{"key":"ref42","article-title":"An introduction to cuda-aware MPI","year":"2013"},{"key":"ref43","article-title":"Nvidia collective communications library (NCCL)","year":"2019"},{"issue":"4","key":"ref44","doi-asserted-by":"crossref","first-page":"63","DOI":"10.1145\/1402946.1402967","article-title":"A scalable","volume":"38","author":"Al-Fares","year":"2008","journal-title":"ACM SIGCOMM Comput. Commun. Rev."},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1145\/3492321.3519584"},{"key":"ref46","first-page":"1","article-title":"Hanayo: Harnessing wave-like pipeline parallelism for enhanced large model training efficiency","volume-title":"Proc. Int. Conf. High Perform. Comput. Netw. Storage Anal.","author":"Liu","year":"2023"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2015.11"},{"key":"ref48","article-title":"Skip-thought vectors","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"28","author":"Kiros","year":"2015"},{"key":"ref49","article-title":"Openwebtext corpus","author":"Gokaslan","year":"2019"},{"key":"ref50","article-title":"Exploring the limits of language modeling","author":"Jozefowicz","year":"2016"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.21437\/interspeech.2014-564"},{"key":"ref52","article-title":"Google\u2019s neural machine translation system: Bridging the gap between human and machine translation","author":"Wu","year":"2016"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/W16-2301"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.3115\/v1\/W14-3302"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D16-1264"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1145\/3302424.3303957"},{"key":"ref57","first-page":"172","article-title":"Blink: Fast and generic collectives for distributed ML","volume-title":"Proc. Mach. Learn. Syst.","volume":"2","author":"Wang","year":"2020"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1109\/JSAC.2023.3242733"},{"key":"ref59","article-title":"ZeRO: Extremely efficient collective communication for giant model training","author":"Wang","year":"2023"},{"key":"ref60","first-page":"785","article-title":"Scaling distributed machine learning with in-network aggregation","volume-title":"Proc. 18th USENIX Symp. Netw. Syst. Des. Implementation","author":"Sapio","year":"2021"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1145\/3552326.3587436"},{"key":"ref62","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1145\/3394486.3406703"},{"key":"ref64","first-page":"551","article-title":"ZeRO-Offload: Democratizing billion-scale model training","volume-title":"Proc. USENIX Annu. Tech. Conf. USENIX Assoc.","author":"Ren","year":"2021"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476205"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2022.3219819"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1145\/3605573.3605613"}],"container-title":["IEEE Transactions on Parallel and Distributed Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/71\/10557478\/10540313.pdf?arnumber=10540313","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,9,10]],"date-time":"2024-09-10T04:26:34Z","timestamp":1725942394000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10540313\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,8]]},"references-count":67,"journal-issue":{"issue":"8"},"URL":"https:\/\/doi.org\/10.1109\/tpds.2024.3406420","relation":{},"ISSN":["1045-9219","1558-2183","2161-9883"],"issn-type":[{"value":"1045-9219","type":"print"},{"value":"1558-2183","type":"electronic"},{"value":"2161-9883","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,8]]}}}