{"id":"https://openalex.org/W4320067926","doi":"https://doi.org/10.1145/3545008.3545052","title":"TCB: Accelerating Transformer Inference Services with Request Concatenation","display_name":"TCB: Accelerating Transformer Inference Services with Request Concatenation","publication_year":2022,"publication_date":"2022-08-29","ids":{"openalex":"https://openalex.org/W4320067926","doi":"https://doi.org/10.1145/3545008.3545052"},"language":"en","primary_location":{"id":"doi:10.1145/3545008.3545052","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3545008.3545052","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 51st International Conference on Parallel Processing","raw_type":"proceedings-article"},"type":"conference-paper","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/A5027046756","display_name":"Boqian Fu","orcid":null},"institutions":[{"id":"https://openalex.org/I141591182","display_name":"University of Aizu","ror":"https://ror.org/02pg0e883","country_code":"JP","type":"education","lineage":["https://openalex.org/I141591182"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Boqian Fu","raw_affiliation_strings":["The University of Aizu, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Aizu, Japan","institution_ids":["https://openalex.org/I141591182"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088461411","display_name":"Fahao Chen","orcid":"https://orcid.org/0000-0002-4345-1296"},"institutions":[{"id":"https://openalex.org/I141591182","display_name":"University of Aizu","ror":"https://ror.org/02pg0e883","country_code":"JP","type":"education","lineage":["https://openalex.org/I141591182"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Fahao Chen","raw_affiliation_strings":["The University of Aizu, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Aizu, Japan","institution_ids":["https://openalex.org/I141591182"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100432795","display_name":"Peng Li","orcid":"https://orcid.org/0000-0003-4981-0496"},"institutions":[{"id":"https://openalex.org/I141591182","display_name":"University of Aizu","ror":"https://ror.org/02pg0e883","country_code":"JP","type":"education","lineage":["https://openalex.org/I141591182"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Peng Li","raw_affiliation_strings":["The University of Aizu, Japan"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Aizu, Japan","institution_ids":["https://openalex.org/I141591182"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5077093242","display_name":"Deze Zeng","orcid":"https://orcid.org/0000-0003-3276-1202"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deze Zeng","raw_affiliation_strings":["China University of Geoscience, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"China University of Geoscience, China","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":13,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"11"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11273","display_name":"Advanced Graph Neural Networks","score":0.9970999956130981,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.9970999956130981,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9921000003814697,"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"}},{"id":"https://openalex.org/T10273","display_name":"IoT and Edge/Fog Computing","score":0.9902999997138977,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/computer-science","display_name":"Computer science","score":0.7539935111999512},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6473208665847778},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5277048945426941},{"id":"https://openalex.org/keywords/scheduling","display_name":"Scheduling (production processes)","score":0.5065230131149292},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.4786145091056824},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.26895982027053833},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.15244504809379578},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.1347447633743286},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.07787752151489258},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.07768303155899048}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7539935111999512},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6473208665847778},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5277048945426941},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.5065230131149292},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.4786145091056824},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.26895982027053833},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.15244504809379578},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1347447633743286},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.07787752151489258},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.07768303155899048},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3545008.3545052","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3545008.3545052","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 51st International Conference on Parallel Processing","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":27,"referenced_works":["https://openalex.org/W95608104","https://openalex.org/W2251939518","https://openalex.org/W2734941459","https://openalex.org/W2771472444","https://openalex.org/W2970971581","https://openalex.org/W3005664618","https://openalex.org/W3016939927","https://openalex.org/W3035016936","https://openalex.org/W3038827379","https://openalex.org/W3047371394","https://openalex.org/W3097132740","https://openalex.org/W3097411828","https://openalex.org/W3098625345","https://openalex.org/W3112101075","https://openalex.org/W3130716829","https://openalex.org/W3158444059","https://openalex.org/W3165292244","https://openalex.org/W3189877953","https://openalex.org/W3204998121","https://openalex.org/W3210432446","https://openalex.org/W3211336323","https://openalex.org/W3211525823","https://openalex.org/W4212986322","https://openalex.org/W4249510566","https://openalex.org/W4287748555","https://openalex.org/W6713134421","https://openalex.org/W6778883912"],"related_works":["https://openalex.org/W1596201972","https://openalex.org/W4206178588","https://openalex.org/W1485627940","https://openalex.org/W2152433827","https://openalex.org/W1986253068","https://openalex.org/W2385146268","https://openalex.org/W2160425906","https://openalex.org/W1882733036","https://openalex.org/W2546696010","https://openalex.org/W1992741870"],"abstract_inverted_index":{"Transformer":[0,115],"has":[1,46,67,176],"dominated":[2],"the":[3,40,88,108],"field":[4],"of":[5,10,42,53,165],"natural":[6],"language":[7],"processing":[8],"because":[9],"its":[11],"strong":[12],"capability":[13],"in":[14],"learning":[15],"from":[16,87],"sequential":[17],"input":[18],"data.":[19],"In":[20,59],"recent":[21],"years,":[22],"various":[23],"computing":[24],"and":[25,96,101,175],"networking":[26],"optimizations":[27],"have":[28,103],"been":[29,47],"proposed":[30,68],"for":[31],"improving":[32],"transformer":[33,37,54,82],"training":[34],"efficiency.":[35],"However,":[36],"inference,":[38],"as":[39,123,125],"core":[41],"many":[43],"AI":[44],"services,":[45],"seldom":[48],"studied.":[49],"A":[50],"key":[51],"challenge":[52],"inference":[55,83,116],"is":[56],"variable-length":[57],"input.":[58],"order":[60],"to":[61],"align":[62],"these":[63],"input,":[64],"existing":[65,81],"work":[66],"batching":[69,95],"schemes":[70],"by":[71,136,157],"padding":[72],"zeros,":[73],"which":[74],"unfortunately":[75],"introduces":[76],"significant":[77],"computational":[78,134],"redundancy.":[79],"Moreover,":[80,151],"studies":[84],"are":[85,99],"separated":[86],"whole":[89],"serving":[90],"system,":[91],"where":[92],"both":[93],"request":[94,97,161,173],"scheduling":[98,130,162,168],"critical":[100],"they":[102],"complex":[104],"interaction.":[105],"To":[106],"fill":[107],"research":[109],"gap,":[110],"we":[111,152],"propose":[112],"TCB,":[113],"a":[114,119,126,154],"system":[117],"with":[118,147],"novel":[120],"ConcatBatching":[121,132],"scheme":[122],"well":[124],"jointly":[127],"designed":[128],"online":[129,160],"algorithm.":[131],"minimizes":[133],"redundancy":[135],"concatenating":[137],"multiple":[138],"requests,":[139],"so":[140],"that":[141,183],"batch":[142],"rows":[143],"can":[144,185],"be":[145],"aligned":[146],"reduced":[148],"padded":[149],"zeros.":[150],"conduct":[153],"systemic":[155],"study":[156],"designing":[158],"an":[159],"algorithm":[163,169],"aware":[164],"ConcatBatching.":[166],"This":[167],"needs":[170],"no":[171],"future":[172],"information":[174],"provable":[177],"theoretical":[178],"guarantee.":[179],"Experimental":[180],"results":[181],"show":[182],"TCB":[184],"significantly":[186],"outperform":[187],"state-of-the-art.":[188]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":3},{"year":2023,"cited_by_count":7}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}