{"id":"https://openalex.org/W4414856580","doi":"https://doi.org/10.1109/tcad.2025.3616078","title":"Terafly: A Multinode FPGA-Based Accelerator Design for Efficient Cooperative Inference in LLMs","display_name":"Terafly: A Multinode FPGA-Based Accelerator Design for Efficient Cooperative Inference in LLMs","publication_year":2025,"publication_date":"2025-10-06","ids":{"openalex":"https://openalex.org/W4414856580","doi":"https://doi.org/10.1109/tcad.2025.3616078"},"language":"en","primary_location":{"id":"doi:10.1109/tcad.2025.3616078","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2025.3616078","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-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":null,"display_name":"Jianing Zheng","orcid":"https://orcid.org/0009-0009-5037-4848"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jianing Zheng","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0009-0009-5037-4848","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]},{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100389329","display_name":"Gang Chen","orcid":"https://orcid.org/0000-0003-4234-1359"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Gang Chen","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0003-4234-1359","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]},{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044840341","display_name":"Libo Huang","orcid":"https://orcid.org/0000-0001-7878-3998"},"institutions":[{"id":"https://openalex.org/I170215575","display_name":"National University of Defense Technology","ror":"https://ror.org/05d2yfz11","country_code":"CN","type":"education","lineage":["https://openalex.org/I170215575"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Libo Huang","raw_affiliation_strings":["College of Computer Science and Technology, National University of Defense Technology, Changsha, China"],"raw_orcid":"https://orcid.org/0000-0001-7878-3998","affiliations":[{"raw_affiliation_string":"College of Computer Science and Technology, National University of Defense Technology, Changsha, China","institution_ids":["https://openalex.org/I170215575"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103146448","display_name":"Xin Lou","orcid":"https://orcid.org/0000-0001-7196-9861"},"institutions":[{"id":"https://openalex.org/I30809798","display_name":"ShanghaiTech University","ror":"https://ror.org/030bhh786","country_code":"CN","type":"education","lineage":["https://openalex.org/I30809798"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xin Lou","raw_affiliation_strings":["School of Information Science and Technology, ShanghaiTech University, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0001-7196-9861","affiliations":[{"raw_affiliation_string":"School of Information Science and Technology, ShanghaiTech University, Shanghai, China","institution_ids":["https://openalex.org/I30809798"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5108050904","display_name":"Wei\u2010Shi Zheng","orcid":"https://orcid.org/0000-0001-8327-0003"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wei-Shi Zheng","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China"],"raw_orcid":"https://orcid.org/0000-0001-8327-0003","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]},{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.1081,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.83450805,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":98},"biblio":{"volume":"45","issue":"5","first_page":"2446","last_page":"2459"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11181","display_name":"Advanced Data Storage Technologies","score":0.9593999981880188,"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"}},"topics":[{"id":"https://openalex.org/T11181","display_name":"Advanced Data Storage Technologies","score":0.9593999981880188,"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/software-deployment","display_name":"Software deployment","score":0.6887000203132629},{"id":"https://openalex.org/keywords/flexibility","display_name":"Flexibility (engineering)","score":0.6340000033378601},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.557699978351593},{"id":"https://openalex.org/keywords/personalization","display_name":"Personalization","score":0.5253000259399414},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5065000057220459},{"id":"https://openalex.org/keywords/efficient-energy-use","display_name":"Efficient energy use","score":0.4505000114440918},{"id":"https://openalex.org/keywords/hardware-acceleration","display_name":"Hardware acceleration","score":0.390500009059906},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.36399999260902405}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8057000041007996},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.6887000203132629},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.6340000033378601},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.557699978351593},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.5360000133514404},{"id":"https://openalex.org/C183003079","wikidata":"https://www.wikidata.org/wiki/Q1000371","display_name":"Personalization","level":2,"score":0.5253000259399414},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5065000057220459},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.4505000114440918},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.41679999232292175},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.390500009059906},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.36399999260902405},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.3255000114440918},{"id":"https://openalex.org/C115051666","wikidata":"https://www.wikidata.org/wiki/Q6522493","display_name":"Ranging","level":2,"score":0.3244999945163727},{"id":"https://openalex.org/C2780395129","wikidata":"https://www.wikidata.org/wiki/Q1128971","display_name":"Rapid prototyping","level":2,"score":0.320499986410141},{"id":"https://openalex.org/C113843644","wikidata":"https://www.wikidata.org/wiki/Q901882","display_name":"Interface (matter)","level":4,"score":0.2939000129699707},{"id":"https://openalex.org/C186370098","wikidata":"https://www.wikidata.org/wiki/Q442787","display_name":"Energy (signal processing)","level":2,"score":0.28700000047683716},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.2849000096321106},{"id":"https://openalex.org/C81293917","wikidata":"https://www.wikidata.org/wiki/Q4189534","display_name":"System deployment","level":3,"score":0.28060001134872437},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.27900001406669617},{"id":"https://openalex.org/C2776221188","wikidata":"https://www.wikidata.org/wiki/Q21072556","display_name":"Design space exploration","level":2,"score":0.2782000005245209},{"id":"https://openalex.org/C98025372","wikidata":"https://www.wikidata.org/wiki/Q477538","display_name":"Systems architecture","level":3,"score":0.257099986076355},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.2567000091075897},{"id":"https://openalex.org/C58013763","wikidata":"https://www.wikidata.org/wiki/Q5754574","display_name":"High-level synthesis","level":3,"score":0.25619998574256897},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.25519999861717224},{"id":"https://openalex.org/C157922185","wikidata":"https://www.wikidata.org/wiki/Q173198","display_name":"Logic synthesis","level":3,"score":0.2502000033855438}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tcad.2025.3616078","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2025.3616078","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G4459438753","display_name":null,"funder_award_id":"62272475","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G8062148801","display_name":null,"funder_award_id":"SL2024A04J0183","funder_id":"https://openalex.org/F4320320671","funder_display_name":"National Research Foundation"},{"id":"https://openalex.org/G8949795292","display_name":null,"funder_award_id":"92470202","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320320671","display_name":"National Research Foundation","ror":"https://ror.org/05s0g1g46"},{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In":[0,69],"this":[1,37,89],"paper,":[2],"we":[3,39,71,175],"propose":[4,72],"Terafly,":[5],"a":[6,41,73,122,129,170,182],"multi-node":[7],"accelerator":[8,21,110,137,168],"design":[9,83],"tailored":[10],"for":[11,84],"efficient":[12],"Large":[13],"Language":[14],"Model":[15],"(LLM)":[16],"deployment":[17,86],"and":[18,30,102,112,146,181],"inference.":[19,34],"Conventional":[20],"architectures":[22,52],"struggle":[23],"to":[24,61,76,115,154,189],"effectively":[25],"handle":[26],"both":[27,143],"the":[28,47,54,63,78,165,190],"prefill":[29],"decode":[31],"stages":[32],"during":[33],"To":[35],"address":[36],"limitation,":[38],"introduce":[40],"hybrid":[42],"spatial-temporal":[43],"architecture":[44],"that":[45],"combines":[46],"high-throughput":[48],"advantages":[49],"of":[50,56,67,80],"spatial":[51],"with":[53],"flexibility":[55],"temporal":[57],"architectures,":[58],"enabling":[59,125],"it":[60],"accommodate":[62],"diverse":[64],"inference":[65,127],"patterns":[66],"LLMs.":[68],"addition,":[70],"generation":[74],"framework":[75,106],"streamline":[77],"customization":[79],"our":[81,135],"LLM-friendly":[82],"various":[85,157],"scenarios.":[87],"Within":[88],"framework,":[90],"users":[91],"can":[92,138],"specify":[93],"their":[94],"requirements":[95],"such":[96],"as":[97],"model":[98,130],"type,":[99],"target":[100],"platform,":[101],"performance":[103,158],"goals.":[104],"The":[105],"then":[107],"generates":[108],"multiple":[109],"nodes":[111],"maps":[113],"them":[114],"distinct":[116],"Super":[117],"Logic":[118],"Regions":[119],"(SLRs)":[120],"within":[121],"single":[123,171],"FPGA,":[124],"cooperative":[126],"under":[128,156],"parallelism":[131],"scheme.":[132],"Through":[133],"experiments,":[134],"generated":[136,166],"be":[139],"easily":[140],"deployed":[141],"on":[142,169],"Alveo":[144,172],"U250":[145],"U50lv":[147,173],"cards,":[148],"serving":[149],"models":[150],"ranging":[151],"from":[152],"OPT-350M":[153],"OPT-1.3B":[155,163],"settings.":[159],"Notably,":[160],"when":[161],"running":[162],"using":[164],"dual-node":[167],"card,":[174],"achieve":[176],"an":[177],"average":[178],"1.1x":[179],"speed-up":[180],"3.4x":[183],"improvement":[184],"in":[185],"energy":[186],"efficiency":[187],"compared":[188],"Nvidia":[191],"A100":[192],"GPU.":[193]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}
