{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,8,2]],"date-time":"2025-08-02T14:45:35Z","timestamp":1754145935448,"version":"3.41.2"},"reference-count":26,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2025,1,1]],"date-time":"2025-01-01T00:00:00Z","timestamp":1735689600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"funder":[{"name":"National Research Foundation of Kore"},{"name":"Korea government [Ministry of Science and Information and Communication Technology (MSIT)]","award":["RS-2024-00405495"],"award-info":[{"award-number":["RS-2024-00405495"]}]},{"name":"Plug&Play"},{"name":"Technology Innovation Program","award":["RS-2024-00420541","2410000802"],"award-info":[{"award-number":["RS-2024-00420541","2410000802"]}]},{"name":"Ministry of Trade, Industry and Energy of Korea"},{"name":"Institute of Information & communications Technology Planning & Evaluatio"},{"name":"Korea government (MSIT) (Development of Processing-in-Memory","award":["2022-0-00050"],"award-info":[{"award-number":["2022-0-00050"]}]},{"name":"Samsung Electronics Company, Ltd., Hwaseong, Korea"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Access"],"published-print":{"date-parts":[[2025]]},"DOI":"10.1109\/access.2025.3587387","type":"journal-article","created":{"date-parts":[[2025,7,9]],"date-time":"2025-07-09T23:20:23Z","timestamp":1752103223000},"page":"120591-120602","source":"Crossref","is-referenced-by-count":0,"title":["AsymGroup: Asymmetric Grouping and Communication Optimization for 2D Tensor Parallelism in LLM Inference"],"prefix":"10.1109","volume":"13","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9115-7868","authenticated-orcid":false,"given":"Ki","family":"Tae Kim","sequence":"first","affiliation":[{"name":"Department of Electrical and Electronic Engineering, Yonsei University, Seoul, South Korea"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-4160-0758","authenticated-orcid":false,"given":"Seok-Ju","family":"Im","sequence":"additional","affiliation":[{"name":"Department of Electrical and Electronic Engineering, Yonsei University, Seoul, South Korea"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2013-8763","authenticated-orcid":false,"given":"Eui-Young","family":"Chung","sequence":"additional","affiliation":[{"name":"Department of Electrical and Electronic Engineering, Yonsei University, Seoul, South Korea"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","article-title":"GPT-4 technical report","volume-title":"arXiv:2303.08774","author":"OpenAI","year":"2023"},{"key":"ref2","article-title":"LLaMA: Open and efficient foundation language models","author":"Touvron","year":"2023","journal-title":"arXiv:2302.13971"},{"key":"ref3","article-title":"Using DeepSpeed and megatron to train megatron-turing NLG 530B, a large-scale generative language model","author":"Smith","year":"2022","journal-title":"arXiv:2201.11990"},{"key":"ref4","article-title":"PanGu-\u2211: Towards trillion parameter language model with sparse heterogeneous computing","author":"Ren","year":"2023","journal-title":"arXiv:2303.10845"},{"key":"ref5","article-title":"M6\u201310T: A sharing-delinking paradigm for efficient multi-trillion parameter pretraining","author":"Lin","year":"2021","journal-title":"arXiv:2110.03888"},{"key":"ref6","first-page":"6543","article-title":"Terapipe: Token-level pipeline parallelism for training large-scale language models","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Li"},{"key":"ref7","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"},{"key":"ref8","article-title":"Megatron-LM: Training multi-billion parameter language models using model parallelism","author":"Shoeybi","year":"2019","journal-title":"arXiv:1909.08053"},{"key":"ref9","article-title":"Demystifying AI platform design for distributed inference of next-generation LLM models","author":"Bambhaniya","year":"2024","journal-title":"arXiv:2406.01698"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1145\/3437801.3441620"},{"key":"ref11","first-page":"526","article-title":"On optimizing the communication of model parallelism","volume-title":"Proc. Mach. Learn. Syst.","author":"Zhuang"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS54959.2023.00031"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2024.3397800"},{"key":"ref14","article-title":"Colossal-AI: A unified deep learning system for large-scale parallel training","author":"Li","year":"2021","journal-title":"arXiv:2110.14883"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18072.2020.9218538"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO61859.2024.00068"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS54959.2023.00023"},{"key":"ref18","first-page":"593","article-title":"TACCL: Guiding collective algorithm synthesis using communication sketches","volume-title":"Proc. 20th USENIX Symp. Networked Syst. Design Implement. (NSDI 23)","author":"Shah"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1002\/cpe.1206"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-540-39924-7_38"},{"key":"ref21","article-title":"Blink: Fast and generic collectives for distributed ML","author":"Wang","year":"2019","journal-title":"arXiv:1910.04940"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1145\/3392717.3392771"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1145\/3620666.3651379"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.1997.569606"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/CLUSTER49012.2020.00013"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1145\/3627703.3629580"}],"container-title":["IEEE Access"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/6287639\/10820123\/11075777.pdf?arnumber=11075777","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,7,17]],"date-time":"2025-07-17T18:01:16Z","timestamp":1752775276000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11075777\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025]]},"references-count":26,"URL":"https:\/\/doi.org\/10.1109\/access.2025.3587387","relation":{},"ISSN":["2169-3536"],"issn-type":[{"type":"electronic","value":"2169-3536"}],"subject":[],"published":{"date-parts":[[2025]]}}}