{"id":"https://openalex.org/W7128004217","doi":"https://doi.org/10.1145/3748173.3779572","title":"FARE: A <u>F</u> ine-gr <u>a</u> ined Pipelined <u>Re</u> configurable FlashAttention Kernel","display_name":"FARE: A <u>F</u> ine-gr <u>a</u> ined Pipelined <u>Re</u> configurable FlashAttention Kernel","publication_year":2026,"publication_date":"2026-02-05","ids":{"openalex":"https://openalex.org/W7128004217","doi":"https://doi.org/10.1145/3748173.3779572"},"language":null,"primary_location":{"id":"doi:10.1145/3748173.3779572","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3748173.3779572","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 ACM/SIGDA International Symposium on Field Programmable Gate Arrays","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3748173.3779572","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125231996","display_name":"Kaushikkumar S. Rathva","orcid":null},"institutions":[{"id":"https://openalex.org/I59270414","display_name":"Indian Institute of Science Bangalore","ror":"https://ror.org/04dese585","country_code":"IN","type":"education","lineage":["https://openalex.org/I59270414"]}],"countries":["IN"],"is_corresponding":true,"raw_author_name":"Kaushikkumar S. Rathva","raw_affiliation_strings":["Indian Institute of Science (IISc), Bengaluru, Karnataka, India"],"affiliations":[{"raw_affiliation_string":"Indian Institute of Science (IISc), Bengaluru, Karnataka, India","institution_ids":["https://openalex.org/I59270414"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5110228812","display_name":"A. H. M. Badiul Alam","orcid":null},"institutions":[{"id":"https://openalex.org/I145894827","display_name":"Indian Institute of Technology Kharagpur","ror":"https://ror.org/03w5sq511","country_code":"IN","type":"education","lineage":["https://openalex.org/I145894827"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Aakarsh Alam","raw_affiliation_strings":["Indian Institute of Technology (IIT), Kharagpur, Kharagpur, West Bengal, India"],"affiliations":[{"raw_affiliation_string":"Indian Institute of Technology (IIT), Kharagpur, Kharagpur, West Bengal, India","institution_ids":["https://openalex.org/I145894827"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125104358","display_name":"Srini Srinivasan","orcid":null},"institutions":[{"id":"https://openalex.org/I4210137977","display_name":"Advanced Micro Devices (United States)","ror":"https://ror.org/04kd6c783","country_code":"US","type":"company","lineage":["https://openalex.org/I4210137977"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Srini Srinivasan","raw_affiliation_strings":["Advanced Micro Devices, Inc. (AMD), Santa Clara, California, USA"],"affiliations":[{"raw_affiliation_string":"Advanced Micro Devices, Inc. (AMD), Santa Clara, California, USA","institution_ids":["https://openalex.org/I4210137977"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5034794487","display_name":"Sumit K. Mandal","orcid":"https://orcid.org/0000-0002-3041-7486"},"institutions":[{"id":"https://openalex.org/I59270414","display_name":"Indian Institute of Science Bangalore","ror":"https://ror.org/04dese585","country_code":"IN","type":"education","lineage":["https://openalex.org/I59270414"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Sumit K. Mandal","raw_affiliation_strings":["Indian Institute of Science (IISc), Bengaluru, Karnataka, India"],"affiliations":[{"raw_affiliation_string":"Indian Institute of Science (IISc), Bengaluru, Karnataka, India","institution_ids":["https://openalex.org/I59270414"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5125231996"],"corresponding_institution_ids":["https://openalex.org/I59270414"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.34172305,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"179","last_page":"179"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.4607999920845032,"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":0.4607999920845032,"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/T14347","display_name":"Big Data and Digital Economy","score":0.25290000438690186,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.10119999945163727,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/reconfigurability","display_name":"Reconfigurability","score":0.6281999945640564},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.5455999970436096},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.5328999757766724},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.46459999680519104},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.3149000108242035},{"id":"https://openalex.org/keywords/digital-signal-processing","display_name":"Digital signal processing","score":0.3133000135421753}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8093000054359436},{"id":"https://openalex.org/C2780149590","wikidata":"https://www.wikidata.org/wiki/Q7302742","display_name":"Reconfigurability","level":2,"score":0.6281999945640564},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.5455999970436096},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.5328999757766724},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.5163999795913696},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.46459999680519104},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.36890000104904175},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.34869998693466187},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3174000084400177},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3149000108242035},{"id":"https://openalex.org/C84462506","wikidata":"https://www.wikidata.org/wiki/Q173142","display_name":"Digital signal processing","level":2,"score":0.3133000135421753},{"id":"https://openalex.org/C2780165032","wikidata":"https://www.wikidata.org/wiki/Q16869822","display_name":"Energy consumption","level":2,"score":0.2921999990940094},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.2840999960899353},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.27730000019073486},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.26919999718666077},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.2687999904155731},{"id":"https://openalex.org/C142962650","wikidata":"https://www.wikidata.org/wiki/Q240838","display_name":"Reconfigurable computing","level":3,"score":0.2524000108242035}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3748173.3779572","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3748173.3779572","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 ACM/SIGDA International Symposium on Field Programmable Gate Arrays","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3748173.3779572","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3748173.3779572","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 ACM/SIGDA International Symposium on Field Programmable Gate Arrays","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Transformer":[0],"algorithm-based":[1],"large":[2],"machine":[3],"learning":[4],"(ML)":[5],"models":[6,124,198],"are":[7],"widely":[8],"used":[9],"in":[10,46,74,77,81,135,148,175],"various":[11],"application":[12],"domains":[13],"[3].":[14],"We":[15],"observe":[16],"that":[17,226],"the":[18,47,85,115,145,172,194,208,222,231,239],"attention":[19,59,75,94,116,173,207,223],"operation":[20],"of":[21,29,54,61,67,72,92,171,182,221],"a":[22,26,35,65,153,183],"transformer":[23,48,126],"algorithm":[24,49],"consumes":[25],"significant":[27],"amount":[28],"execution":[30],"time":[31],"when":[32],"executed":[33],"on":[34,213],"GPU,":[36],"affecting":[37,84],"real-time":[38],"latency":[39],"for":[40,113,122],"inference.":[41],"While":[42,87],"other":[43],"major":[44],"operations":[45,63,73],"(e.g.,":[50],"feed-forward":[51],"operation)":[52],"consist":[53],"weight":[55],"stationary":[56],"matrix":[57,68],"multiplications,":[58],"consists":[60],"non-linear":[62],"and":[64,200],"sequence":[66,71],"multiplications.":[69],"This":[70],"results":[76],"memory":[78,98],"access":[79,99],"bottlenecks":[80,100],"GPUs,":[82],"directly":[83],"latency.":[86],"there":[88],"exists":[89],"an":[90,93],"implementation":[91],"kernel":[95,158,185,224],"to":[96,167,186,235,244],"minimize":[97],"--":[101,104],"FlashAttention":[102],"[1]":[103],"it":[105],"lacks":[106],"reconfigurability.":[107],"Moreover,":[108],"GPUs":[109],"provide":[110],"limited":[111],"knobs":[112],"customizing":[114],"datapath.":[117],"There":[118],"exist":[119],"application-specific":[120,138],"accelerators":[121],"ML":[123,197],"with":[125,218],"algorithms,":[127],"but":[128],"they":[129],"too":[130],"lack":[131],"reconfigurability":[132],"[2].":[133],"FPGAs,":[134],"contrast,":[136],"enable":[137],"dataflows,":[139],"giving":[140],"us":[141,166],"full":[142],"control":[143],"over":[144,246],"design.":[146],"Therefore,":[147],"this":[149,204],"work,":[150],"we":[151],"propose":[152],"fine-grained":[154,163],"pipelined":[155],"reconfigurable":[156],"flash-attention":[157,184],"targeting":[159],"FPGA":[160],"platforms.":[161],"The":[162],"pipelining":[164],"allows":[165],"execute":[168],"different":[169,180,219],"stages":[170,181],"mechanism":[174],"parallel.":[176],"It":[177],"carefully":[178],"constructs":[179],"reduce":[187],"resource":[188,240],"usage":[189,241],"as":[190,192],"well":[191],"improve":[193],"performance.":[195],"As":[196],"scale":[199],"energy":[201],"demands":[202],"increase,":[203],"makes":[205],"FPGA-based":[206],"sustainable":[209],"choice.":[210],"Experimental":[211],"evaluation":[212],"Xilinx":[214],"Zedboard":[215],"(Zynq":[216],"7000)":[217],"parameters":[220],"shows":[225],"our":[227],"proposed":[228],"design":[229],"improves":[230],"performance":[232],"by":[233,242],"up":[234,243],"2.9\u00d7":[236],"while":[237],"reducing":[238],"16.50\u00d7":[245],"state-of-the-art":[247],"[4].":[248]},"counts_by_year":[],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2026-02-07T00:00:00"}