{"id":"https://openalex.org/W7133570121","doi":"https://doi.org/10.1109/hpca68181.2026.11408460","title":"Towards Compute-Aware In-Switch Computing for LLMs Tensor-Parallelism on Multi-GPU Systems","display_name":"Towards Compute-Aware In-Switch Computing for LLMs Tensor-Parallelism on Multi-GPU Systems","publication_year":2026,"publication_date":"2026-01-31","ids":{"openalex":"https://openalex.org/W7133570121","doi":"https://doi.org/10.1109/hpca68181.2026.11408460"},"language":null,"primary_location":{"id":"doi:10.1109/hpca68181.2026.11408460","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408460","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128101731","display_name":"Chen Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Chen Zhang","raw_affiliation_strings":["Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085826827","display_name":"Qiang Zhang","orcid":"https://orcid.org/0000-0001-6662-1668"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qijun Zhang","raw_affiliation_strings":["Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128051850","display_name":"Zhuoshan Zhou","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]},{"id":"https://openalex.org/I4210160618","display_name":"Huawei Technologies (United Kingdom)","ror":"https://ror.org/056gzgs71","country_code":"GB","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210160618"]}],"countries":["CN","GB"],"is_corresponding":false,"raw_author_name":"Zhuoshan Zhou","raw_affiliation_strings":["Huawei Technologies Co. Ltd","Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Huawei Technologies Co. Ltd","institution_ids":["https://openalex.org/I4210160618"]},{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080651207","display_name":"Yijia Diao","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yijia Diao","raw_affiliation_strings":["Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128085928","display_name":"Haibo Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210160618","display_name":"Huawei Technologies (United Kingdom)","ror":"https://ror.org/056gzgs71","country_code":"GB","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210160618"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Haibo Wang","raw_affiliation_strings":["Huawei Technologies Co. Ltd"],"affiliations":[{"raw_affiliation_string":"Huawei Technologies Co. Ltd","institution_ids":["https://openalex.org/I4210160618"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128078385","display_name":"Zhe Zhou","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]},{"id":"https://openalex.org/I4210160618","display_name":"Huawei Technologies (United Kingdom)","ror":"https://ror.org/056gzgs71","country_code":"GB","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210160618"]}],"countries":["CN","GB"],"is_corresponding":false,"raw_author_name":"Zhe Zhou","raw_affiliation_strings":["Huawei Technologies Co. Ltd","Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Huawei Technologies Co. Ltd","institution_ids":["https://openalex.org/I4210160618"]},{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049071763","display_name":"Zhipeng Tu","orcid":null},"institutions":[{"id":"https://openalex.org/I4210160618","display_name":"Huawei Technologies (United Kingdom)","ror":"https://ror.org/056gzgs71","country_code":"GB","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210160618"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Zhipeng Tu","raw_affiliation_strings":["Huawei Technologies Co. Ltd"],"affiliations":[{"raw_affiliation_string":"Huawei Technologies Co. Ltd","institution_ids":["https://openalex.org/I4210160618"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128122866","display_name":"Zhiyao Li","orcid":null},"institutions":[{"id":"https://openalex.org/I4210160618","display_name":"Huawei Technologies (United Kingdom)","ror":"https://ror.org/056gzgs71","country_code":"GB","type":"company","lineage":["https://openalex.org/I2250955327","https://openalex.org/I4210160618"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Zhiyao Li","raw_affiliation_strings":["Huawei Technologies Co. Ltd"],"affiliations":[{"raw_affiliation_string":"Huawei Technologies Co. Ltd","institution_ids":["https://openalex.org/I4210160618"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089719475","display_name":"Guangyu Sun","orcid":"https://orcid.org/0000-0002-7315-6589"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guangyu Sun","raw_affiliation_strings":["Peking University"],"affiliations":[{"raw_affiliation_string":"Peking University","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067161373","display_name":"Zhuoran Song","orcid":"https://orcid.org/0000-0002-6494-4786"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhuoran Song","raw_affiliation_strings":["Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020545376","display_name":"Zhigang Ji","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhigang Ji","raw_affiliation_strings":["Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003939279","display_name":"Jingwen Leng","orcid":"https://orcid.org/0000-0002-5660-5493"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jingwen Leng","raw_affiliation_strings":["Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5128045567","display_name":"Minyi Guo","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Minyi Guo","raw_affiliation_strings":["Shanghai Jiao Tong University"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":13,"corresponding_author_ids":["https://openalex.org/A5128101731"],"corresponding_institution_ids":["https://openalex.org/I183067930"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.41373341,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"15"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.626800000667572,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.626800000667572,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10101","display_name":"Cloud Computing and Resource Management","score":0.04050000011920929,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12292","display_name":"Graph Theory and Algorithms","score":0.027000000700354576,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.31949999928474426},{"id":"https://openalex.org/keywords/government","display_name":"Government (linguistics)","score":0.31290000677108765},{"id":"https://openalex.org/keywords/work","display_name":"Work (physics)","score":0.2639999985694885},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.26080000400543213}],"concepts":[{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.43140000104904175},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.40049999952316284},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.3375000059604645},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.31949999928474426},{"id":"https://openalex.org/C2778137410","wikidata":"https://www.wikidata.org/wiki/Q2732820","display_name":"Government (linguistics)","level":2,"score":0.31290000677108765},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.30239999294281006},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2639999985694885},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.26080000400543213},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.258899986743927},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.250900000333786}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/hpca68181.2026.11408460","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hpca68181.2026.11408460","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE International Symposium on High Performance Computer Architecture (HPCA)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1702756715","display_name":null,"funder_award_id":"2022YFB4501400","funder_id":"https://openalex.org/F4320335777","funder_display_name":"National Key Research and Development Program of China"}],"funders":[{"id":"https://openalex.org/F4320335777","display_name":"National Key Research and Development Program of China","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":37,"referenced_works":["https://openalex.org/W2118231264","https://openalex.org/W2604553456","https://openalex.org/W2612387305","https://openalex.org/W2761338514","https://openalex.org/W2761710529","https://openalex.org/W2949161920","https://openalex.org/W2991040477","https://openalex.org/W3043522163","https://openalex.org/W3081168214","https://openalex.org/W3091097978","https://openalex.org/W3094502228","https://openalex.org/W3100877701","https://openalex.org/W3102510044","https://openalex.org/W3129831491","https://openalex.org/W3138516171","https://openalex.org/W3188065709","https://openalex.org/W3190445506","https://openalex.org/W3193985311","https://openalex.org/W3204998121","https://openalex.org/W3206778761","https://openalex.org/W3206832494","https://openalex.org/W4242634435","https://openalex.org/W4245911027","https://openalex.org/W4290991009","https://openalex.org/W4312060029","https://openalex.org/W4318540560","https://openalex.org/W4327911839","https://openalex.org/W4381327268","https://openalex.org/W4381327474","https://openalex.org/W4385245566","https://openalex.org/W4385571616","https://openalex.org/W4394998995","https://openalex.org/W4395117348","https://openalex.org/W4399283082","https://openalex.org/W4405756071","https://openalex.org/W4405756102","https://openalex.org/W4414311214"],"related_works":[],"abstract_inverted_index":{"Tensor":[0],"parallelism":[1],"(TP)":[2],"in":[3,63,69,177],"large-scale":[4],"LLM":[5,138],"inference":[6],"and":[7,44,59,66,103,157],"training":[8,150],"introduces":[9,37],"frequent":[10],"collective":[11,26],"operations":[12,27],"that":[13,85,141],"dominate":[14],"inter-GPU":[15],"communication.":[16],"While":[17],"in-switch":[18,109],"computing,":[19],"exemplified":[20],"by":[21,28],"NVLink":[22],"SHARP":[23],"(NVLS),":[24],"accelerates":[25],"reducing":[29],"redundant":[30],"data":[31],"transfer,":[32],"its":[33,41,175],"communication-centric":[34],"design":[35],"philosophy":[36],"the":[38,45,57,74,79,119,153,164],"mismatch":[39,55],"between":[40],"communication":[42,60,87],"mode":[43],"memory":[46,91],"semantic":[47],"requirement":[48],"of":[49,96],"LLM's":[50],"computation":[51],"kernel.":[52],"Such":[53],"a":[54,132],"isolates":[56],"compute":[58],"phases,":[61],"resulting":[62],"underutilized":[64],"resources":[65],"limited":[67],"overlap":[68,167],"multi-GPU":[70,181],"systems.":[71,182],"To":[72],"address":[73],"limitation,":[75],"we":[76],"propose":[77],"CAIS,":[78],"first":[80],"ComputeAware":[81],"In-Switch":[82],"computing":[83],"framework":[84],"aligns":[86],"modes":[88],"with":[89],"computation's":[90],"semantics":[92],"requirement.":[93],"CAIS":[94,142],"consists":[95],"three":[97],"integral":[98],"techniques:":[99],"(1)":[100],"compute-aware":[101,108],"ISA":[102],"microarchitecture":[104],"extension":[105],"to":[106,117,130],"enable":[107],"computing.":[110],"(2)":[111],"merging-aware":[112],"TB":[113],"(Thread":[114],"Block)":[115],"coordination":[116],"improve":[118],"temporal":[120],"alignment":[121],"for":[122],"efficient":[123],"request":[124],"merging.":[125],"(3)":[126],"graph-level":[127],"dataflow":[128],"optimizer":[129],"achieve":[131],"tight":[133],"cross-kernel":[134],"overlap.":[135],"Evaluations":[136],"on":[137,180],"workloads":[139],"show":[140],"achieves":[143],"<tex":[144,158],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[145,159],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$1.38":[146],"\\times$</tex>":[147,161],"average":[148],"end-to-end":[149],"speedup":[151],"over":[152,162],"SOTA":[154,165],"NVLS-enabled":[155],"solution,":[156],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$1.61":[160],"T3,":[163],"compute-communicate":[166],"solutions":[168],"but":[169],"do":[170],"not":[171],"leverage":[172],"NVLS,":[173],"demonstrating":[174],"effectiveness":[176],"accelerating":[178],"TP":[179]},"counts_by_year":[],"updated_date":"2026-04-09T08:11:56.329763","created_date":"2026-03-05T00:00:00"}
