[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84089-en":3,"doc-seo-84089-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},84089,1099514067415,"Rowan","https://ap-avatar.wpscdn.com/avatar/100002539d78ffe74a7?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779092875211072502",6,"Technology","UBEP Re-architecting Expert Parallelism Communication Library for Production Superpods","Deployment of Mixture-of-Experts (MoE) models on production high-bandwidth superpods, such as NVIDIA NVL72/576 and Huawei CloudMatrix384, faces bottlenecks beyond raw interconnect speed. Three core blockers arise from BSP-style execution serialization, synchronization overhead that cannot scale with bandwidth, and load imbalance caused by distance-agnostic scheduling of irregular token traffic. This work proposes UBEP (Unified-Bus Expert Parallelism), a production-ready communication library that redesigns MoE All-to-All for modern superpod architectures, reducing All-to-All latency up to 52.4% and MoE TPOT up to 11.1%.","UBEP: Re-architecting Expert Parallelism Communication Library for Production Superpods  \nYipeng Liu 1,∗ , Chang Liu1,∗ , Si Shen1,∗ , Jiaqi Zheng1,†, Mingfan Li2 , Yuyang Yang 1 , Guanhua Li 1 , Yuquan Zhang 1 , Yimeng Xu 1 , Zhongzhe Hu2,†, Zhiyuan Huang2 , Qihang Duan2 , Junsong Wang2 , Wenkai Ling2 , Baochuan Yang2 , Xianzhi Yu2 , Han Bao2 , Yijie Chen2 , Guihai Chen 1  \n1 State Key Laboratory for Novel Software Technology, Nanjing University 2Huawei Technologies Co., Ltd.  \narXiv :2607 .06202v2 [ cs .DC] 8 Jul 2026  \nABSTRACT  \nThe deployment of Mixture-of-Experts (MoE) models on production high-bandwidth superpods, such as NVIDIA’s NVL72/576 and Huawei’s CloudMatrix384, introduces critical challenges beyond raw interconnect bandwidth. While these systems provide unified global address spaces and high-bandwidth fabrics, their full potential for sparse MoE communication is hindered by three fundamental bottlenecks: (1) Strict execution serialization imposed by coarse-grained Bulk Synchronous Parallel (BSP) orchestration of interdependent communication phases; (2) Prohibitive synchronization overhead that fails to scale alongside high interconnect bandwidth; and (3) Severe load imbalance resulting from distanceagnostic scheduling of irregular token traffic. To eliminate these bottlenecks, we introduce UBEP (Unified-Bus Expert Parallelism), a production-ready communication library that rethinks MoE’s All-to-All primitives for modern superpod architectures. Through large-scale experiments, UBEP reduces All-to-All latency by up to 52.4% and MoE inference Time Per Output Token (TPOT) by up to 11. 1% .  \nCCS CONCEPTS  \n• Networks → Data center networks; • Computing methodologies → Parallel computing methodologies;  \nKEYWORDS  \nMixture-of-Experts, Expert parallelism, All-to-All communication, Communication library, Superpod, Data center networks  \nACM Reference Format:  \nYipeng Liu, Chang Liu, Si Shen, Jiaqi Zheng, Mingfan Li, Yuyang Yang, Guanhua Li, Yuquan Zhang, Yimeng Xu, Zhongzhe Hu, Zhiyuan Huang, Qihang Duan, Junsong Wang, Wenkai Ling, Baochuan Yang, Xianzhi Yu, Han Bao, Yijie Chen, and Guihai Chen. 2026. UBEP: Re-architecting Expert Parallelism Communication Library for Production Superpods. In ACM SIGCOMM 2026 Conference (SIGCOMM’26), August 17–21, 2026, Denver, CO,  \n∗ Equal contribution.†Corresponding author.  \nPermission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission [and/or a fee. Request permissions from permissions@acm.org](and/or a fee. Request permissions from permissions@acm.org).  \nSIGCOMM’26, August 17–21, 2026, Denver, CO, USA  \n© 2026 Copyright held by the owner/author(s) . Publication rights licensed to the Association for Computing Machinery.  \nACM ISBN 979-8-4007-2467-1/26/08 .  \n[https://doi.org/10.1145/3789240.3829183](https://doi.org/10.1145/3789240.3829183)  \nUSA. ACM, New York, NY, USA, 18 pages. [https://doi.org/10.1145/3789240](https://doi.org/10.1145/3789240) . 3829183  \n1 INTRODUCTION  \nThe rapid evolution of Large Language Models (LLMs) has established the Mixture-of-Experts (MoE) architecture as the standard for balancing massive parameter scales with inference efficiency [7] . To support these communication-intensive workloads, datacenter infrastructure is shifting from traditional commodity clusters toward specialized superpod architectures, such as NVIDIA’s NVL72/576 [31] and Huawei’s CloudMatrix384 (CM384) [57] . Unlike traditional clusters that rely on scale-out networks such as InfiniBand (IB)/RoCE with high tail-latency and expl","cbCaiezH4P8x6h0j","https://ap.wps.com/l/cbCaiezH4P8x6h0j","pdf",1413433,1,18,"English","en",105,"# Abstract\n# Introduction\n## MoE and superpod context\n## Existing EPCL approaches and limitations\n## Bottlenecks in re-architecting EPCL","[{\"question\":\"Why do MoE communication challenges on superpods go beyond interconnect bandwidth?\",\"answer\":\"Even with unified high-bandwidth fabrics, sparse and irregular MoE traffic is limited by BSP-style serialization, synchronization overhead, and load imbalance from distance-agnostic scheduling.\"},{\"question\":\"What are the three fundamental bottlenecks identified for expert parallelism communication libraries?\",\"answer\":\"The paper identifies BSP serialization, non-scaling synchronization overhead, and severe load imbalance due to scheduling that ignores token distance effects.\"},{\"question\":\"How does UBEP improve production MoE communication on superpods?\",\"answer\":\"UBEP re-architects MoE All-to-All primitives for modern superpod designs, achieving up to 52.4% lower All-to-All latency and up to 11.1% reduction in MoE Time Per Output Token.\"}]",1784192656,45,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"ubep-re-architecting-expert-parallelism-communication-library-for-production-superpods","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/technology/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/ubep-re-architecting-expert-parallelism-communication-library-for-production-superpods/84089/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why do MoE communication challenges on superpods go beyond interconnect bandwidth?","Question",{"text":75,"@type":76},"Even with unified high-bandwidth fabrics, sparse and irregular MoE traffic is limited by BSP-style serialization, synchronization overhead, and load imbalance from distance-agnostic scheduling.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What are the three fundamental bottlenecks identified for expert parallelism communication libraries?",{"text":80,"@type":76},"The paper identifies BSP serialization, non-scaling synchronization overhead, and severe load imbalance due to scheduling that ignores token distance effects.",{"name":82,"@type":73,"acceptedAnswer":83},"How does UBEP improve production MoE communication on superpods?",{"text":84,"@type":76},"UBEP re-architects MoE All-to-All primitives for modern superpod designs, achieving up to 52.4% lower All-to-All latency and up to 11.1% reduction in MoE Time Per Output Token.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,113,118,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":111,"slug":112},50,"technology",{"id":114,"doc_module":4,"doc_module_name":45,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":45,"category_name":120,"show_sort_weight":121,"slug":122},8,"Research & Report",30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]