{"id":"https://openalex.org/W3213445774","doi":"https://doi.org/10.1145/3469661","title":"When Massive GPU Parallelism Ain\u2019t Enough: A Novel Hardware Architecture of 2D-LSTM Neural Network","display_name":"When Massive GPU Parallelism Ain\u2019t Enough: A Novel Hardware Architecture of 2D-LSTM Neural Network","publication_year":2021,"publication_date":"2021-11-09","ids":{"openalex":"https://openalex.org/W3213445774","doi":"https://doi.org/10.1145/3469661","mag":"3213445774"},"language":"en","primary_location":{"id":"doi:10.1145/3469661","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3469661","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":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5026761218","display_name":"Vladimir Rybalkin","orcid":"https://orcid.org/0000-0002-0926-6062"},"institutions":[{"id":"https://openalex.org/I153267046","display_name":"University of Kaiserslautern","ror":"https://ror.org/04zrf7b53","country_code":"DE","type":"education","lineage":["https://openalex.org/I153267046"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Vladimir Rybalkin","raw_affiliation_strings":["Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany","institution_ids":["https://openalex.org/I153267046"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044188420","display_name":"Jonas Ney","orcid":null},"institutions":[{"id":"https://openalex.org/I153267046","display_name":"University of Kaiserslautern","ror":"https://ror.org/04zrf7b53","country_code":"DE","type":"education","lineage":["https://openalex.org/I153267046"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Jonas Ney","raw_affiliation_strings":["Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany","institution_ids":["https://openalex.org/I153267046"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065374950","display_name":"Menbere Kina Tekleyohannes","orcid":null},"institutions":[{"id":"https://openalex.org/I153267046","display_name":"University of Kaiserslautern","ror":"https://ror.org/04zrf7b53","country_code":"DE","type":"education","lineage":["https://openalex.org/I153267046"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Menbere Kina Tekleyohannes","raw_affiliation_strings":["Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany","institution_ids":["https://openalex.org/I153267046"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5059285190","display_name":"Norbert Wehn","orcid":"https://orcid.org/0000-0002-9010-086X"},"institutions":[{"id":"https://openalex.org/I153267046","display_name":"University of Kaiserslautern","ror":"https://ror.org/04zrf7b53","country_code":"DE","type":"education","lineage":["https://openalex.org/I153267046"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Norbert Wehn","raw_affiliation_strings":["Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microelectronic Systems Design Research Group, University of Kaiserslautern, Kaiserslautern, Germany","institution_ids":["https://openalex.org/I153267046"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I153267046"],"apc_list":null,"apc_paid":null,"fwci":0.5822,"has_fulltext":false,"cited_by_count":9,"citation_normalized_percentile":{"value":0.6937887,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":97},"biblio":{"volume":"15","issue":"1","first_page":"1","last_page":"35"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9995999932289124,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9995999932289124,"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/T10601","display_name":"Handwritten Text Recognition Techniques","score":0.9991000294685364,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9965999722480774,"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/computer-science","display_name":"Computer science","score":0.9048680067062378},{"id":"https://openalex.org/keywords/mnist-database","display_name":"MNIST database","score":0.7682969570159912},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.6923882961273193},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.5790273547172546},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.5137602090835571},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.5118705630302429},{"id":"https://openalex.org/keywords/segmentation","display_name":"Segmentation","score":0.4827558398246765},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4827079772949219},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.4737566113471985},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.45427095890045166},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.4473417103290558},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.4441235065460205},{"id":"https://openalex.org/keywords/hardware-acceleration","display_name":"Hardware acceleration","score":0.4319363236427307},{"id":"https://openalex.org/keywords/data-parallelism","display_name":"Data parallelism","score":0.425159752368927},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.38163426518440247},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.35354483127593994},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.3455597758293152},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.27508771419525146},{"id":"https://openalex.org/keywords/parallelism","display_name":"Parallelism (grammar)","score":0.2314874529838562}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.9048680067062378},{"id":"https://openalex.org/C190502265","wikidata":"https://www.wikidata.org/wiki/Q17069496","display_name":"MNIST database","level":3,"score":0.7682969570159912},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.6923882961273193},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.5790273547172546},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.5137602090835571},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.5118705630302429},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.4827558398246765},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4827079772949219},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4737566113471985},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.45427095890045166},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.4473417103290558},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.4441235065460205},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.4319363236427307},{"id":"https://openalex.org/C61483411","wikidata":"https://www.wikidata.org/wiki/Q3124522","display_name":"Data parallelism","level":3,"score":0.425159752368927},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.38163426518440247},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.35354483127593994},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3455597758293152},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.27508771419525146},{"id":"https://openalex.org/C2781172179","wikidata":"https://www.wikidata.org/wiki/Q853109","display_name":"Parallelism (grammar)","level":2,"score":0.2314874529838562},{"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":1,"locations":[{"id":"doi:10.1145/3469661","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3469661","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"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/8","score":0.4699999988079071,"display_name":"Decent work and economic growth"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":73,"referenced_works":["https://openalex.org/W1444003893","https://openalex.org/W1771459135","https://openalex.org/W1895577753","https://openalex.org/W1901129140","https://openalex.org/W1903029394","https://openalex.org/W1909234690","https://openalex.org/W2016829658","https://openalex.org/W2112796928","https://openalex.org/W2141273399","https://openalex.org/W2171943915","https://openalex.org/W2177436562","https://openalex.org/W2267126114","https://openalex.org/W2300242332","https://openalex.org/W2317505761","https://openalex.org/W2327501763","https://openalex.org/W2340897893","https://openalex.org/W2469490737","https://openalex.org/W2529096783","https://openalex.org/W2559655401","https://openalex.org/W2560023338","https://openalex.org/W2565125333","https://openalex.org/W2565639579","https://openalex.org/W2573601984","https://openalex.org/W2585720638","https://openalex.org/W2588448445","https://openalex.org/W2589086007","https://openalex.org/W2604272474","https://openalex.org/W2604738573","https://openalex.org/W2727238169","https://openalex.org/W2752782242","https://openalex.org/W2762910930","https://openalex.org/W2765235648","https://openalex.org/W2765994921","https://openalex.org/W2786345741","https://openalex.org/W2786974559","https://openalex.org/W2787480186","https://openalex.org/W2802022891","https://openalex.org/W2803263632","https://openalex.org/W2827334269","https://openalex.org/W2894619617","https://openalex.org/W2906403229","https://openalex.org/W2908483391","https://openalex.org/W2908839997","https://openalex.org/W2915106038","https://openalex.org/W2916222163","https://openalex.org/W2926066262","https://openalex.org/W2950941578","https://openalex.org/W2952026396","https://openalex.org/W2962820060","https://openalex.org/W2962896736","https://openalex.org/W2963526839","https://openalex.org/W2964033223","https://openalex.org/W2964217532","https://openalex.org/W2964333985","https://openalex.org/W2966450079","https://openalex.org/W2969825080","https://openalex.org/W2972295109","https://openalex.org/W2988282777","https://openalex.org/W2988314300","https://openalex.org/W2996689340","https://openalex.org/W3008515469","https://openalex.org/W3013596877","https://openalex.org/W3014641072","https://openalex.org/W3040903763","https://openalex.org/W3090325485","https://openalex.org/W3102058420","https://openalex.org/W3102169921","https://openalex.org/W3105988348","https://openalex.org/W3117114782","https://openalex.org/W3119048801","https://openalex.org/W4243521730","https://openalex.org/W4246411305","https://openalex.org/W4300479382"],"related_works":["https://openalex.org/W4386603768","https://openalex.org/W2950475743","https://openalex.org/W2518118925","https://openalex.org/W4285104150","https://openalex.org/W3159273459","https://openalex.org/W4319952061","https://openalex.org/W4280636456","https://openalex.org/W4388913998","https://openalex.org/W4310584535","https://openalex.org/W4295935044"],"abstract_inverted_index":{"Multidimensional":[0],"Long":[1],"Short-Term":[2],"Memory":[3],"(MD-LSTM)":[4],"neural":[5,56,213],"network":[6],"is":[7,65,80,241],"an":[8,81],"extension":[9],"of":[10,61,71,96,211,250],"one-dimensional":[11,251],"LSTM":[12],"for":[13,69,84,117,137,155,216],"data":[14],"with":[15,175,188,253],"more":[16],"than":[17,208],"one":[18],"dimension.":[19],"MD-LSTM":[20],"achieves":[21],"state-of-the-art":[22],"results":[23],"in":[24],"various":[25],"applications,":[26],"including":[27],"handwritten":[28,156,222],"text":[29],"recognition,":[30],"medical":[31],"imaging,":[32],"and":[33,51,130,150,183,205,232,259],"many":[34],"more.":[35],"However,":[36],"its":[37],"implementation":[38],"suffers":[39],"from":[40,145],"the":[41,62,93,101,106,113,146,221],"inherently":[42],"sequential":[43],"execution":[44],"that":[45,75,87,168],"tremendously":[46],"slows":[47],"down":[48],"both":[49],"training":[50],"inference":[52,70],"compared":[53],"to":[54,66,124,177,181,190,195,255],"other":[55],"networks.":[57],"The":[58],"main":[59],"goal":[60],"current":[63],"research":[64],"provide":[67,100,229],"acceleration":[68],"MD-LSTM.":[72,118],"We":[73,119,132],"advocate":[74],"Field-Programmable":[76],"Gate":[77],"Array":[78],"(FPGA)":[79],"alternative":[82],"platform":[83],"deep":[85],"learning":[86],"can":[88,233],"offer":[89],"a":[90,121,126,134,151,237,242],"solution":[91,238],"when":[92,239],"massive":[94],"parallelism":[95],"GPUs":[97],"does":[98],"not":[99],"necessary":[102],"performance":[103],"required":[104],"by":[105,179,193],"application.":[107],"In":[108],"this":[109],"article,":[110],"we":[111,164],"present":[112],"first":[114],"hardware":[115],"architecture":[116],"conduct":[120],"systematic":[122],"exploration":[123],"analyze":[125],"tradeoff":[127],"between":[128],"precision":[129],"accuracy.":[131],"use":[133],"challenging":[135],"dataset":[136,154],"semantic":[138,217],"segmentation,":[139],"namely":[140],"historical":[141],"document":[142],"image":[143],"binarization":[144],"DIBCO":[147],"2017":[148],"contest":[149],"well-known":[152],"MNIST":[153],"digit":[157,223],"recognition.":[158],"Based":[159],"on":[160],"our":[161,226],"new":[162],"architecture,":[163],"implement":[165],"FPGA-based":[166,209],"accelerators":[167,199],"outperform":[169,246],"Nvidia":[170,184],"Geforce":[171],"RTX":[172],"2080":[173],"Ti":[174],"respect":[176,189,254],"throughput":[178],"up":[180,194],"9.9":[182],"Jetson":[185],"AGX":[186],"Xavier":[187],"energy":[191,203,257],"efficiency":[192,207],"48":[196],".":[197],"Our":[198],"achieve":[200],"higher":[201,230],"throughput,":[202,256],"efficiency,":[204,258],"resource":[206,260],"implementations":[210,228,249],"convolutional":[212],"networks":[214],"(CNNs)":[215],"segmentation":[218],"tasks.":[219],"For":[220],"recognition":[224],"task,":[225],"FPGA":[227,248],"accuracy":[231,240],"be":[234],"considered":[235],"as":[236],"priority.":[243],"Furthermore,":[244],"they":[245],"earlier":[247],"LSTMs":[252],"efficiency.":[261]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":2},{"year":2021,"cited_by_count":1}],"updated_date":"2026-06-26T08:34:08.712188","created_date":"2025-10-10T00:00:00"}