{"id":"https://openalex.org/W4200023416","doi":"https://doi.org/10.1145/3470567","title":"Elastic-DF: Scaling Performance of DNN Inference in FPGA Clouds through Automatic Partitioning","display_name":"Elastic-DF: Scaling Performance of DNN Inference in FPGA Clouds through Automatic Partitioning","publication_year":2021,"publication_date":"2021-12-06","ids":{"openalex":"https://openalex.org/W4200023416","doi":"https://doi.org/10.1145/3470567"},"language":"en","primary_location":{"id":"doi:10.1145/3470567","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3470567","pdf_url":null,"source":{"id":"https://openalex.org/S112809824","display_name":"ACM Transactions on Reconfigurable Technology and Systems","issn_l":"1936-7406","issn":["1936-7406","1936-7414"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Reconfigurable Technology and Systems","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://repositorio.uam.es/bitstream/10486/711818/1/9052717.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5059252278","display_name":"Tob\u00edas Alonso","orcid":"https://orcid.org/0000-0001-8632-9146"},"institutions":[{"id":"https://openalex.org/I63634437","display_name":"Universidad Aut\u00f3noma de Madrid","ror":"https://ror.org/01cby8j38","country_code":"ES","type":"education","lineage":["https://openalex.org/I63634437"]}],"countries":["ES"],"is_corresponding":false,"raw_author_name":"Tobias Alonso","raw_affiliation_strings":["Universidad Aut\u00f3noma de Madrid, Madrid, Spain"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universidad Aut\u00f3noma de Madrid, Madrid, Spain","institution_ids":["https://openalex.org/I63634437"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062570405","display_name":"Lucian Petric\u0103","orcid":"https://orcid.org/0000-0002-3480-0570"},"institutions":[{"id":"https://openalex.org/I4210101362","display_name":"Xilinx (Ireland)","ror":"https://ror.org/00pahkj72","country_code":"IE","type":"company","lineage":["https://openalex.org/I32923980","https://openalex.org/I4210101362"]}],"countries":["IE"],"is_corresponding":false,"raw_author_name":"Lucian Petrica","raw_affiliation_strings":["Xilinx Research, Dublin, Ireland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xilinx Research, Dublin, Ireland","institution_ids":["https://openalex.org/I4210101362"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054050822","display_name":"Mario Ruiz","orcid":"https://orcid.org/0000-0002-3433-5883"},"institutions":[{"id":"https://openalex.org/I4210101362","display_name":"Xilinx (Ireland)","ror":"https://ror.org/00pahkj72","country_code":"IE","type":"company","lineage":["https://openalex.org/I32923980","https://openalex.org/I4210101362"]}],"countries":["IE"],"is_corresponding":false,"raw_author_name":"Mario Ruiz","raw_affiliation_strings":["Xilinx University Program, Dublin, Ireland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xilinx University Program, Dublin, Ireland","institution_ids":["https://openalex.org/I4210101362"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5009891908","display_name":"Jakoba Petri-Koenig","orcid":"https://orcid.org/0009-0002-2457-7819"},"institutions":[{"id":"https://openalex.org/I98358874","display_name":"Delft University of Technology","ror":"https://ror.org/02e2c7k09","country_code":"NL","type":"education","lineage":["https://openalex.org/I98358874"]}],"countries":["NL"],"is_corresponding":false,"raw_author_name":"Jakoba Petri-Koenig","raw_affiliation_strings":["Delft University of Technology, Delft, Netherlands"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Delft University of Technology, Delft, Netherlands","institution_ids":["https://openalex.org/I98358874"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001185686","display_name":"Yaman Umuroglu","orcid":null},"institutions":[{"id":"https://openalex.org/I4210101362","display_name":"Xilinx (Ireland)","ror":"https://ror.org/00pahkj72","country_code":"IE","type":"company","lineage":["https://openalex.org/I32923980","https://openalex.org/I4210101362"]}],"countries":["IE"],"is_corresponding":false,"raw_author_name":"Yaman Umuroglu","raw_affiliation_strings":["Xilinx Research, Dublin, Ireland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xilinx Research, Dublin, Ireland","institution_ids":["https://openalex.org/I4210101362"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5009279802","display_name":"Ioannis Stamelos","orcid":"https://orcid.org/0000-0001-9440-3633"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ioannis Stamelos","raw_affiliation_strings":["InAccel, US"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"InAccel, US","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006269716","display_name":"Elias Koromilas","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Elias Koromilas","raw_affiliation_strings":["InAccel, US"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"InAccel, US","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5031107943","display_name":"Michaela Blott","orcid":"https://orcid.org/0000-0002-7833-4057"},"institutions":[{"id":"https://openalex.org/I4210101362","display_name":"Xilinx (Ireland)","ror":"https://ror.org/00pahkj72","country_code":"IE","type":"company","lineage":["https://openalex.org/I32923980","https://openalex.org/I4210101362"]}],"countries":["IE"],"is_corresponding":false,"raw_author_name":"Michaela Blott","raw_affiliation_strings":["Xilinx Research, Dublin, Ireland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xilinx Research, Dublin, Ireland","institution_ids":["https://openalex.org/I4210101362"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5011464762","display_name":"Kees Vissers","orcid":"https://orcid.org/0000-0002-6249-315X"},"institutions":[{"id":"https://openalex.org/I32923980","display_name":"Xilinx (United States)","ror":"https://ror.org/01rb7bk56","country_code":"US","type":"company","lineage":["https://openalex.org/I32923980"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kees Vissers","raw_affiliation_strings":["Xilinx Research, San Jos\u00e9, US"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xilinx Research, San Jos\u00e9, US","institution_ids":["https://openalex.org/I32923980"]}]}],"institutions":[],"countries_distinct_count":4,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":2.8509,"has_fulltext":true,"cited_by_count":44,"citation_normalized_percentile":{"value":0.91342177,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":93,"max":100},"biblio":{"volume":"15","issue":"2","first_page":"1","last_page":"34"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9998999834060669,"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"}},"topics":[{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9998999834060669,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9983000159263611,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9972000122070312,"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/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.8594409227371216},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8432709574699402},{"id":"https://openalex.org/keywords/dataflow","display_name":"Dataflow","score":0.5723205804824829},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5613662600517273},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5366031527519226},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.47590383887290955},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.4304351508617401},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.4240148365497589},{"id":"https://openalex.org/keywords/software-portability","display_name":"Software portability","score":0.4194057285785675},{"id":"https://openalex.org/keywords/compiler","display_name":"Compiler","score":0.41929614543914795},{"id":"https://openalex.org/keywords/ethernet","display_name":"Ethernet","score":0.41132745146751404},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.40991151332855225},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.278594970703125},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.16591790318489075},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.1234917938709259}],"concepts":[{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.8594409227371216},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8432709574699402},{"id":"https://openalex.org/C96324660","wikidata":"https://www.wikidata.org/wiki/Q205446","display_name":"Dataflow","level":2,"score":0.5723205804824829},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5613662600517273},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5366031527519226},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.47590383887290955},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.4304351508617401},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.4240148365497589},{"id":"https://openalex.org/C63000827","wikidata":"https://www.wikidata.org/wiki/Q3080428","display_name":"Software portability","level":2,"score":0.4194057285785675},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.41929614543914795},{"id":"https://openalex.org/C172173386","wikidata":"https://www.wikidata.org/wiki/Q79984","display_name":"Ethernet","level":2,"score":0.41132745146751404},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.40991151332855225},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.278594970703125},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.16591790318489075},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.1234917938709259},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.0},{"id":"https://openalex.org/C555944384","wikidata":"https://www.wikidata.org/wiki/Q249","display_name":"Wireless","level":2,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3470567","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3470567","pdf_url":null,"source":{"id":"https://openalex.org/S112809824","display_name":"ACM Transactions on Reconfigurable Technology and Systems","issn_l":"1936-7406","issn":["1936-7406","1936-7414"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Reconfigurable Technology and Systems","raw_type":"journal-article"},{"id":"pmh:oai:repositorio.uam.es:10486/711818","is_oa":true,"landing_page_url":"http://hdl.handle.net/10486/711818","pdf_url":"https://repositorio.uam.es/bitstream/10486/711818/1/9052717.pdf","source":{"id":"https://openalex.org/S4306400964","display_name":"Biblos-e Archivo (Universidad Aut\u00f3noma de Madrid)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I63634437","host_organization_name":"Universidad Aut\u00f3noma de Madrid","host_organization_lineage":["https://openalex.org/I63634437"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"info:eu-repo/semantics/acceptedVersion"},{"id":"pmh:oai:tudelft.nl:uuid:6a2e4181-2cae-44f4-9ecd-e0ff53136406","is_oa":false,"landing_page_url":"http://resolver.tudelft.nl/uuid:6a2e4181-2cae-44f4-9ecd-e0ff53136406","pdf_url":null,"source":{"id":"https://openalex.org/S4306400906","display_name":"Research Repository (Delft University of Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I98358874","host_organization_name":"Delft University of Technology","host_organization_lineage":["https://openalex.org/I98358874"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"journal article"}],"best_oa_location":{"id":"pmh:oai:repositorio.uam.es:10486/711818","is_oa":true,"landing_page_url":"http://hdl.handle.net/10486/711818","pdf_url":"https://repositorio.uam.es/bitstream/10486/711818/1/9052717.pdf","source":{"id":"https://openalex.org/S4306400964","display_name":"Biblos-e Archivo (Universidad Aut\u00f3noma de Madrid)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I63634437","host_organization_name":"Universidad Aut\u00f3noma de Madrid","host_organization_lineage":["https://openalex.org/I63634437"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"info:eu-repo/semantics/acceptedVersion"},"sustainable_development_goals":[{"score":0.6299999952316284,"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4200023416.pdf","grobid_xml":"https://content.openalex.org/works/W4200023416.grobid-xml"},"referenced_works_count":44,"referenced_works":["https://openalex.org/W1563299586","https://openalex.org/W2028863828","https://openalex.org/W2050456603","https://openalex.org/W2108598243","https://openalex.org/W2124446952","https://openalex.org/W2283189482","https://openalex.org/W2336648157","https://openalex.org/W2386615278","https://openalex.org/W2475840367","https://openalex.org/W2513568085","https://openalex.org/W2527287972","https://openalex.org/W2542189141","https://openalex.org/W2565125333","https://openalex.org/W2604319603","https://openalex.org/W2615256897","https://openalex.org/W2741581557","https://openalex.org/W2766927668","https://openalex.org/W2798956872","https://openalex.org/W2883929540","https://openalex.org/W2887160985","https://openalex.org/W2891946740","https://openalex.org/W2901123440","https://openalex.org/W2944468162","https://openalex.org/W2945306514","https://openalex.org/W2951082499","https://openalex.org/W2953212265","https://openalex.org/W2962953210","https://openalex.org/W2963427045","https://openalex.org/W2973838542","https://openalex.org/W2987319672","https://openalex.org/W3007581096","https://openalex.org/W3008905965","https://openalex.org/W3013748504","https://openalex.org/W3025152414","https://openalex.org/W3034653380","https://openalex.org/W3035177949","https://openalex.org/W3093796669","https://openalex.org/W3098344759","https://openalex.org/W3102328099","https://openalex.org/W3105234421","https://openalex.org/W3119378958","https://openalex.org/W3121413886","https://openalex.org/W3161542527","https://openalex.org/W4238120903"],"related_works":["https://openalex.org/W2998381397","https://openalex.org/W4236419692","https://openalex.org/W3167919718","https://openalex.org/W4239447582","https://openalex.org/W4251718783","https://openalex.org/W2171015181","https://openalex.org/W2128640905","https://openalex.org/W2011460173","https://openalex.org/W2021092055","https://openalex.org/W2352151710"],"abstract_inverted_index":{"Customized":[0],"compute":[1],"acceleration":[2],"in":[3,174],"the":[4,9,29,69,88,92,96,100,118,158,165,195],"datacenter":[5,202],"is":[6],"key":[7],"to":[8,27,77,82,86,133],"wider":[10],"roll-out":[11],"of":[12,33,49,91,191,201],"applications":[13],"based":[14],"on":[15,45,141,146,178],"deep":[16],"neural":[17],"network":[18],"(DNN)":[19],"inference.":[20,203],"In":[21,95],"this":[22],"article,":[23],"we":[24],"investigate":[25],"how":[26],"maximize":[28,87],"performance":[30,90,159,186],"and":[31,61,80,107,149,164,193],"scalability":[32],"field-programmable":[34],"gate":[35],"array":[36],"(FPGA)-based":[37],"pipeline":[38],"dataflow":[39],"DNN":[40,70,78],"inference":[41],"accelerators":[42],"(DFAs)":[43],"automatically":[44],"computing":[46],"infrastructures":[47],"consisting":[48],"multi-die,":[50],"network-connected":[51],"FPGAs.":[52],"We":[53],"present":[54],"Elastic-DF,":[55],"a":[56,137,153],"novel":[57],"resource":[58],"partitioning":[59],"tool":[60],"associated":[62],"FPGA":[63,75,84,126,192],"runtime":[64],"infrastructure":[65],"that":[66],"integrates":[67],"with":[68],"compiler":[71],"FINN.":[72],"Elastic-DF":[73,98,135,151,169,183],"allocates":[74],"resources":[76],"layers":[79,81],"individual":[83],"dies":[85],"total":[89],"multi-FPGA":[93],"system.":[94],"resulting":[97],"mapping,":[99],"accelerator":[101],"may":[102,110],"be":[103,111],"instantiated":[104],"multiple":[105,114],"times,":[106],"each":[108],"instance":[109],"segmented":[112],"across":[113],"FPGAs":[115],"transparently,":[116],"whereby":[117],"segments":[119],"communicate":[120],"peer-to-peer":[121],"through":[122],"100":[123],"Gbps":[124],"Ethernet":[125],"infrastructure,":[127],"without":[128],"host":[129],"involvement.":[130],"When":[131],"applied":[132],"ResNet-50,":[134],"provides":[136],"44%":[138],"latency":[139],"decrease":[140],"Alveo":[142,147,167],"U280.":[143],"For":[144],"MobileNetV1":[145],"U200":[148],"U280,":[150],"enables":[152],"78%":[154],"throughput":[155,197],"increase,":[156],"eliminating":[157],"difference":[160],"between":[161,188],"these":[162],"cards":[163],"larger":[166],"U250.":[168],"also":[170],"increases":[171,185,194],"operating":[172],"frequency":[173],"all":[175],"our":[176],"experiments,":[177],"average":[179],"by":[180],"over":[181],"20%.":[182],"therefore":[184],"portability":[187],"different":[189],"sizes":[190],"critical":[196],"per":[198],"cost":[199],"metric":[200]},"counts_by_year":[{"year":2025,"cited_by_count":15},{"year":2024,"cited_by_count":10},{"year":2023,"cited_by_count":11},{"year":2022,"cited_by_count":6},{"year":2021,"cited_by_count":2}],"updated_date":"2026-07-02T09:51:11.867554","created_date":"2025-10-10T00:00:00"}
