{"id":"https://openalex.org/W4410582417","doi":"https://doi.org/10.23919/date64628.2025.10993078","title":"LoopLynx: A Scalable Dataflow Architecture for Efficient LLM Inference","display_name":"LoopLynx: A Scalable Dataflow Architecture for Efficient LLM Inference","publication_year":2025,"publication_date":"2025-03-31","ids":{"openalex":"https://openalex.org/W4410582417","doi":"https://doi.org/10.23919/date64628.2025.10993078"},"language":"en","primary_location":{"id":"doi:10.23919/date64628.2025.10993078","is_oa":false,"landing_page_url":"https://doi.org/10.23919/date64628.2025.10993078","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 Design, Automation & Test in Europe Conference (DATE)","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/A5110040473","display_name":"Jianing Zheng","orcid":null},"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":true,"raw_author_name":"Jianing Zheng","raw_affiliation_strings":["Sun Yat-sen University"],"affiliations":[{"raw_affiliation_string":"Sun Yat-sen University","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"last","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":["Sun Yat-sen University"],"affiliations":[{"raw_affiliation_string":"Sun Yat-sen University","institution_ids":["https://openalex.org/I157773358"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5110040473"],"corresponding_institution_ids":["https://openalex.org/I157773358"],"apc_list":null,"apc_paid":null,"fwci":3.1134,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.91687183,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":95,"max":96},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"7"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11181","display_name":"Advanced Data Storage Technologies","score":0.9545000195503235,"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":0.9545000195503235,"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/dataflow","display_name":"Dataflow","score":0.8790222406387329},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7932497262954712},{"id":"https://openalex.org/keywords/dataflow-architecture","display_name":"Dataflow architecture","score":0.7128728628158569},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.7051281332969666},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.6892749667167664},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.5991876721382141},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.5617635846138},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3625774681568146},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.3296715319156647},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.2732875347137451},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.08589580655097961},{"id":"https://openalex.org/keywords/history","display_name":"History","score":0.05225434899330139}],"concepts":[{"id":"https://openalex.org/C96324660","wikidata":"https://www.wikidata.org/wiki/Q205446","display_name":"Dataflow","level":2,"score":0.8790222406387329},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7932497262954712},{"id":"https://openalex.org/C176727019","wikidata":"https://www.wikidata.org/wiki/Q1172415","display_name":"Dataflow architecture","level":3,"score":0.7128728628158569},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7051281332969666},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6892749667167664},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.5991876721382141},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.5617635846138},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3625774681568146},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.3296715319156647},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2732875347137451},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.08589580655097961},{"id":"https://openalex.org/C95457728","wikidata":"https://www.wikidata.org/wiki/Q309","display_name":"History","level":0,"score":0.05225434899330139},{"id":"https://openalex.org/C166957645","wikidata":"https://www.wikidata.org/wiki/Q23498","display_name":"Archaeology","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.23919/date64628.2025.10993078","is_oa":false,"landing_page_url":"https://doi.org/10.23919/date64628.2025.10993078","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 Design, Automation & Test in Europe Conference (DATE)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"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/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":15,"referenced_works":["https://openalex.org/W3047848469","https://openalex.org/W3115388607","https://openalex.org/W3130240120","https://openalex.org/W3162542754","https://openalex.org/W3176468986","https://openalex.org/W3184454880","https://openalex.org/W3206837665","https://openalex.org/W4308083513","https://openalex.org/W4321637273","https://openalex.org/W4393578753","https://openalex.org/W4393949386","https://openalex.org/W6767997687","https://openalex.org/W6846659131","https://openalex.org/W6847478871","https://openalex.org/W6850625674"],"related_works":["https://openalex.org/W2564598376","https://openalex.org/W1484403103","https://openalex.org/W2584408851","https://openalex.org/W2115158825","https://openalex.org/W2101960124","https://openalex.org/W4377693460","https://openalex.org/W2033683327","https://openalex.org/W2783505431","https://openalex.org/W2521947294","https://openalex.org/W4236419692"],"abstract_inverted_index":{"In":[0,132],"this":[1],"paper,":[2],"we":[3,75],"propose":[4],"LoopLynx,":[5],"a":[6,19,28,57,72,77,141,145],"scalable":[7],"dataflow":[8,40],"architecture":[9,80],"for":[10,112],"efficient":[11],"LLM":[12,114],"inference":[13,149],"that":[14,81,89,121],"optimizes":[15],"FPGA":[16,62],"usage":[17],"through":[18],"hybrid":[20,29],"spatial-temporal":[21],"design.":[22],"The":[23],"design":[24],"of":[25,71,117,155],"LoopLynx":[26,99,122],"incorporates":[27],"temporal-spatial":[30],"architecture,":[31,49],"where":[32],"computationally":[33],"intensive":[34],"operators":[35],"are":[36,93],"implemented":[37],"as":[38],"large":[39],"kernels.":[41],"This":[42],"achieves":[43],"high":[44],"throughput":[45],"similar":[46],"to":[47,66,104,107,127,135],"spatial":[48],"and":[50,52,83],"organizing":[51],"reusing":[53],"these":[54],"kernels":[55],"in":[56,148],"temporal":[58],"way":[59],"together":[60],"enhances":[61],"peak":[63],"performance.":[64],"Furthermore,":[65],"overcome":[67],"the":[68,90,156],"resource":[69],"limitations":[70],"single":[73,129],"device,":[74],"provide":[76],"multi-FPGA":[78],"distributed":[79,91],"overlaps":[82],"hides":[84],"all":[85],"data":[86],"transfers":[87],"so":[88],"accelerators":[92],"fully":[94],"utilized.":[95],"By":[96],"doing":[97],"so,":[98],"can":[100,123],"be":[101],"effectively":[102],"scaled":[103],"multiple":[105],"devices":[106],"further":[108],"explore":[109],"model":[110,119],"parallelism":[111],"large-scale":[113],"inference.":[115],"Evaluation":[116],"GPT-2":[118],"demonstrates":[120],"achieve":[124],"comparable":[125],"performance":[126],"state-of-the-art":[128],"FPGA-based":[130],"accelerations.":[131],"addition,":[133],"compared":[134],"Nvidia":[136],"A100,":[137],"our":[138],"accelerator":[139],"with":[140],"dual-FPGA":[142],"configuration":[143],"delivers":[144],"2.52x":[146],"speed-up":[147],"latency":[150],"while":[151],"consuming":[152],"only":[153],"48.1%":[154],"energy.":[157]},"counts_by_year":[{"year":2025,"cited_by_count":2}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}