{"id":"https://openalex.org/W4381327585","doi":"https://doi.org/10.1109/tpds.2023.3287883","title":"GraphAGILE: An FPGA-Based Overlay Accelerator for Low-Latency GNN Inference","display_name":"GraphAGILE: An FPGA-Based Overlay Accelerator for Low-Latency GNN Inference","publication_year":2023,"publication_date":"2023-06-20","ids":{"openalex":"https://openalex.org/W4381327585","doi":"https://doi.org/10.1109/tpds.2023.3287883"},"language":"en","primary_location":{"id":"doi:10.1109/tpds.2023.3287883","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2023.3287883","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"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 Parallel and Distributed 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":"https://openalex.org/A5101916609","display_name":"Bingyi Zhang","orcid":"https://orcid.org/0000-0002-8115-0814"},"institutions":[{"id":"https://openalex.org/I1174212","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60","country_code":"US","type":"education","lineage":["https://openalex.org/I1174212"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Bingyi Zhang","raw_affiliation_strings":["Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, USA"],"raw_orcid":"https://orcid.org/0000-0002-8115-0814","affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, USA","institution_ids":["https://openalex.org/I1174212"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014103790","display_name":"Hanqing Zeng","orcid":"https://orcid.org/0000-0002-2578-2147"},"institutions":[{"id":"https://openalex.org/I4210114444","display_name":"Meta (United States)","ror":"https://ror.org/01zbnvs85","country_code":"US","type":"company","lineage":["https://openalex.org/I4210114444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hanqing Zeng","raw_affiliation_strings":["Meta Platforms, Inc., Menlo Park, CA, USA"],"raw_orcid":"https://orcid.org/0000-0002-2578-2147","affiliations":[{"raw_affiliation_string":"Meta Platforms, Inc., Menlo Park, CA, USA","institution_ids":["https://openalex.org/I4210114444"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5033166029","display_name":"Viktor K. Prasanna","orcid":"https://orcid.org/0000-0002-1609-8589"},"institutions":[{"id":"https://openalex.org/I1174212","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60","country_code":"US","type":"education","lineage":["https://openalex.org/I1174212"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Viktor K. Prasanna","raw_affiliation_strings":["Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, USA"],"raw_orcid":"https://orcid.org/0000-0002-1609-8589","affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, USA","institution_ids":["https://openalex.org/I1174212"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":4.4519,"has_fulltext":false,"cited_by_count":34,"citation_normalized_percentile":{"value":0.95666771,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":97,"max":100},"biblio":{"volume":"34","issue":"9","first_page":"2580","last_page":"2597"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11273","display_name":"Advanced Graph Neural Networks","score":0.998199999332428,"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":0.998199999332428,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9937999844551086,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9919000267982483,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.894293487071991},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.5889253616333008},{"id":"https://openalex.org/keywords/matrix-multiplication","display_name":"Matrix multiplication","score":0.575727105140686},{"id":"https://openalex.org/keywords/loop-unrolling","display_name":"Loop unrolling","score":0.5414881706237793},{"id":"https://openalex.org/keywords/compiler","display_name":"Compiler","score":0.534807562828064},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.4844663143157959},{"id":"https://openalex.org/keywords/control-reconfiguration","display_name":"Control reconfiguration","score":0.4639776349067688},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4607014060020447},{"id":"https://openalex.org/keywords/compile-time","display_name":"Compile time","score":0.4132012724876404},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.27066075801849365},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.22625726461410522},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.14028075337409973}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.894293487071991},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5889253616333008},{"id":"https://openalex.org/C17349429","wikidata":"https://www.wikidata.org/wiki/Q1049914","display_name":"Matrix multiplication","level":3,"score":0.575727105140686},{"id":"https://openalex.org/C76970557","wikidata":"https://www.wikidata.org/wiki/Q1869750","display_name":"Loop unrolling","level":3,"score":0.5414881706237793},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.534807562828064},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.4844663143157959},{"id":"https://openalex.org/C119701452","wikidata":"https://www.wikidata.org/wiki/Q5165881","display_name":"Control reconfiguration","level":2,"score":0.4639776349067688},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4607014060020447},{"id":"https://openalex.org/C200833197","wikidata":"https://www.wikidata.org/wiki/Q333707","display_name":"Compile time","level":3,"score":0.4132012724876404},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.27066075801849365},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.22625726461410522},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.14028075337409973},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C84114770","wikidata":"https://www.wikidata.org/wiki/Q46344","display_name":"Quantum","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tpds.2023.3287883","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpds.2023.3287883","pdf_url":null,"source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"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 Parallel and Distributed Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2891451553","display_name":null,"funder_award_id":"CCF-1919289","funder_id":"https://openalex.org/F4320335353","funder_display_name":"National Science Foundation of Sri Lanka"},{"id":"https://openalex.org/G7050272079","display_name":null,"funder_award_id":"OAC-2209563","funder_id":"https://openalex.org/F4320335353","funder_display_name":"National Science Foundation of Sri Lanka"}],"funders":[{"id":"https://openalex.org/F4320335353","display_name":"National Science Foundation of Sri Lanka","ror":"https://ror.org/010xaa060"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":47,"referenced_works":["https://openalex.org/W1836465849","https://openalex.org/W2034861439","https://openalex.org/W2916106175","https://openalex.org/W2926442184","https://openalex.org/W2961295589","https://openalex.org/W2962711740","https://openalex.org/W2963114857","https://openalex.org/W2963511748","https://openalex.org/W2964015378","https://openalex.org/W2973152414","https://openalex.org/W2979455536","https://openalex.org/W3000310304","https://openalex.org/W3017228913","https://openalex.org/W3046757167","https://openalex.org/W3080555959","https://openalex.org/W3091862797","https://openalex.org/W3098230582","https://openalex.org/W3100078588","https://openalex.org/W3105753905","https://openalex.org/W3106125969","https://openalex.org/W3111579839","https://openalex.org/W3123909522","https://openalex.org/W3133347161","https://openalex.org/W3172512547","https://openalex.org/W3206743063","https://openalex.org/W4211165432","https://openalex.org/W4287645102","https://openalex.org/W4288419263","https://openalex.org/W4294558607","https://openalex.org/W4295312788","https://openalex.org/W4297733535","https://openalex.org/W4367147545","https://openalex.org/W6638667902","https://openalex.org/W6726873649","https://openalex.org/W6738964360","https://openalex.org/W6754929296","https://openalex.org/W6760001035","https://openalex.org/W6760045743","https://openalex.org/W6760755035","https://openalex.org/W6765543928","https://openalex.org/W6766978945","https://openalex.org/W6776488958","https://openalex.org/W6781932242","https://openalex.org/W6783741718","https://openalex.org/W6785059380","https://openalex.org/W6789720248","https://openalex.org/W6808054070"],"related_works":["https://openalex.org/W2371266106","https://openalex.org/W2136609022","https://openalex.org/W2117382851","https://openalex.org/W2382449560","https://openalex.org/W2767298477","https://openalex.org/W2110432562","https://openalex.org/W2025344","https://openalex.org/W2117864619","https://openalex.org/W2090337950","https://openalex.org/W1495975288"],"abstract_inverted_index":{"This":[0],"article":[1],"presents":[2],"GraphAGILE,":[3],"a":[4,72,110,129,178,224],"domain-specific":[5],"FPGA-based":[6],"overlay":[7],"accelerator":[8],"for":[9,133,203],"graph":[10,115,155,186],"neural":[11],"network":[12],"(GNN)":[13],"inference.":[14],"GraphAGILE":[15,49,222,231,255],"consists":[16],"of":[17,87,109,121,131,193,266],"(1)a":[18],"novel":[19,73],"unified":[20],"architecture":[21,44,68],"designwith":[22],"aninstruction":[23],"set,":[24],"and":[25,46,63,98,113,123,127,205,215,244,268,279],"(2)a":[26],"compilerbuilt":[27],"upon":[28],"the":[29,40,47,67,107,114,119,138,153,158,184,189,264,273,289],"instruction":[30,42],"set":[31,43],"that":[32,81,150,165,182,198,254],"can":[33,82,232],"quickly":[34],"generate":[35],"optimized":[36],"code.":[37],"Due":[38],"to":[39,142,156,169,187,209],"proposed":[41],"(ISA)":[45],"compiler,":[48],"does":[50],"not":[51],"require":[52],"any":[53],"FPGA":[54,226,291],"reconfiguration":[55],"when":[56],"performing":[57],"inference":[58,134,144],"on":[59,223,276],"various":[60,84],"GNN":[61,111,236,246],"models":[62,247],"input":[64,185],"graphs.":[65],"For":[66],"design,":[69],"we":[70],"propose":[71],"hardware":[74,269,284],"module":[75],"named":[76],"Adaptive":[77],"Computation":[78],"Kernel":[79],"(ACK),":[80],"execute":[83,233],"computation":[85,147,154,160,211],"kernels":[86],"GNNs,":[88],"including":[89,238,263],"general":[90],"matrix":[91,95,101],"multiplication":[92,96,102],"(GEMM),":[93],"sparse-dense":[94],"(SpDMM),":[97],"sampled":[99],"dense-dense":[100],"(SDDMM).":[103],"The":[104],"compiler":[105,140],"takes":[106],"specifications":[108],"model":[112],"meta":[116],"data":[117,171,175,213],"(e.g.,":[118],"number":[120],"vertices":[122],"edges)":[124],"as":[125],"input,":[126],"generates":[128],"sequence":[130],"instructions":[132],"execution.":[135],"We":[136,220],"develop":[137],"following":[139],"optimizations":[141],"reduce":[143,157,170],"latency:":[145],"1)":[146],"order":[148],"optimization":[149],"automatically":[151,199],"reorders":[152],"total":[159],"complexity,":[161],"2)":[162],"layer":[163],"fusion":[164],"merges":[166],"adjacent":[167],"layers":[168],"communication":[172,214],"volume,":[173],"3)":[174],"partitioning":[176],"with":[177,212,272,288],"partition-centric":[179],"execution":[180,201,285],"scheme":[181],"partitions":[183],"fit":[188],"available":[190],"on-chip":[191],"memory":[192],"FPGA,":[194],"4)":[195],"kernel":[196],"mapping":[197],"selects":[200],"mode":[202],"ACK,":[204],"performs":[206],"task":[207],"scheduling":[208],"overlap":[210],"achieves":[216,256,280],"dynamic":[217],"load":[218],"balance.":[219],"implement":[221],"state-of-the-art":[225,274,290],"platform,":[227],"Xilinx":[228],"Alveo":[229],"U250.":[230],"widely":[234],"used":[235],"models,":[237],"GCN,":[239],"GAT,":[240],"GIN,":[241],"GraphSAGE,":[242],"SGC":[243],"other":[245],"supported":[248],"by":[249],"GraphGym.":[250],"Experimental":[251],"results":[252],"show":[253],"up":[257,281],"to$47.1\\times$($3.9\\times$)":[258],"reduction":[259],"in":[260,283],"end-to-end":[261],"latency,":[262],"latency":[265,286],"compilation":[267],"execution,":[270],"compared":[271,287],"implementations":[275],"CPU":[277],"(GPU),":[278],"to$2.9\\times$reduction":[282],"accelerators.":[292]},"counts_by_year":[{"year":2026,"cited_by_count":7},{"year":2025,"cited_by_count":15},{"year":2024,"cited_by_count":8},{"year":2023,"cited_by_count":4}],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2025-10-10T00:00:00"}
