{"id":"https://openalex.org/W7128704250","doi":"https://doi.org/10.1109/tvlsi.2026.3659893","title":"Area- and Utilization-Efficient LLM Accelerator With Fused Speculative Decoding for Edge-Side Inference","display_name":"Area- and Utilization-Efficient LLM Accelerator With Fused Speculative Decoding for Edge-Side Inference","publication_year":2026,"publication_date":"2026-02-12","ids":{"openalex":"https://openalex.org/W7128704250","doi":"https://doi.org/10.1109/tvlsi.2026.3659893"},"language":null,"primary_location":{"id":"doi:10.1109/tvlsi.2026.3659893","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tvlsi.2026.3659893","pdf_url":null,"source":{"id":"https://openalex.org/S37538908","display_name":"IEEE Transactions on Very Large Scale Integration (VLSI) Systems","issn_l":"1063-8210","issn":["1063-8210","1557-9999"],"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 Very Large Scale Integration (VLSI) 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/A5022033867","display_name":"Kaiqi Chen","orcid":"https://orcid.org/0009-0006-6439-3255"},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kaiqi Chen","raw_affiliation_strings":["State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0006-6439-3255","affiliations":[{"raw_affiliation_string":"State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125762373","display_name":"Zikang Zhou","orcid":null},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zikang Zhou","raw_affiliation_strings":["State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0006-2530-3886","affiliations":[{"raw_affiliation_string":"State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125707891","display_name":"Yaqi Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yaqi Chen","raw_affiliation_strings":["State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5121165610","display_name":"Jun Hyun Han","orcid":null},"institutions":[{"id":"https://openalex.org/I24943067","display_name":"Fudan University","ror":"https://ror.org/013q1eq08","country_code":"CN","type":"education","lineage":["https://openalex.org/I24943067"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jun Han","raw_affiliation_strings":["State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0002-5245-0754","affiliations":[{"raw_affiliation_string":"State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China","institution_ids":["https://openalex.org/I24943067"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I24943067"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.1422814,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"34","issue":"4","first_page":"1393","last_page":"1397"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.23019999265670776,"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":0.23019999265670776,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.08449999988079071,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials 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":0.07490000128746033,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.8409000039100647},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.6439999938011169},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.5946999788284302},{"id":"https://openalex.org/keywords/hardware-acceleration","display_name":"Hardware acceleration","score":0.5622000098228455},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.4361000061035156},{"id":"https://openalex.org/keywords/autoregressive-model","display_name":"Autoregressive model","score":0.4032999873161316}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.8409000039100647},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7764000296592712},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.6439999938011169},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.5946999788284302},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.5622000098228455},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.4361000061035156},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.4032999873161316},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.40220001339912415},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.39890000224113464},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.38659998774528503},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.37279999256134033},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.34869998693466187},{"id":"https://openalex.org/C2778456923","wikidata":"https://www.wikidata.org/wiki/Q5337692","display_name":"Edge computing","level":3,"score":0.3431999981403351},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.32749998569488525},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.32260000705718994},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.30320000648498535},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.2948000133037567},{"id":"https://openalex.org/C165005293","wikidata":"https://www.wikidata.org/wiki/Q1074500","display_name":"Chip","level":2,"score":0.2784999907016754},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.2628999948501587},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2563999891281128},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.25529998540878296},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.25119999051094055}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tvlsi.2026.3659893","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tvlsi.2026.3659893","pdf_url":null,"source":{"id":"https://openalex.org/S37538908","display_name":"IEEE Transactions on Very Large Scale Integration (VLSI) Systems","issn_l":"1063-8210","issn":["1063-8210","1557-9999"],"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 Very Large Scale Integration (VLSI) Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2519052758","display_name":"\u81ea\u611f\u77e5\u81ea\u7ec4\u7ec7\u5f02\u6784\u4f17\u6838\u667a\u80fd\u82af\u7247\u7684\u4e92\u8fde\u4e0e\u5b58\u50a8\u6280\u672f\u7814\u7a76","funder_award_id":"61934002","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G4369517068","display_name":null,"funder_award_id":"62234008","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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0,90],"inference":[1,17,26,34,75,113,170,209],"of":[2,21,48,100],"large":[3],"language":[4],"models":[5],"(LLMs)":[6],"on":[7,76,183],"edge":[8,77,208],"devices":[9,78],"has":[10],"always":[11],"been":[12],"a":[13],"challenge,":[14],"the":[15,44,49,82,95,134,195,203],"autoregressive":[16,146],"and":[18,51,62,97,104,112,127,166,199,225],"enormous":[19],"number":[20],"parameters":[22],"result":[23],"in":[24],"long":[25],"latency.":[27],"Although":[28],"speculative":[29,69,102],"decoding":[30,70,103],"is":[31],"proposed":[32,91],"for":[33,39,207],"acceleration,":[35],"it":[36,55],"poses":[37],"challenges":[38],"hardware":[40,63,129],"accelerator":[41,130,206],"design,":[42],"as":[43],"differing":[45],"computational":[46],"characteristics":[47],"drafting":[50,96],"verification":[52,98],"stages":[53],"make":[54],"difficult":[56],"to":[57,72,84,153,173],"optimize":[58,73],"both":[59],"chip":[60],"area":[61,197],"utilization.":[64],"This":[65],"brief":[66],"introduces":[67],"fused":[68,135],"(FSD)":[71],"LLM":[74,205],"by":[79,138,151,158,171,210],"unifying":[80],"all":[81],"operations":[83,136],"general":[85],"matrix":[86],"multiplications":[87],"during":[88],"inference.":[89],"FSD-Infer":[92,148],"algorithm":[93],"fuses":[94],"phases":[99],"conventional":[101],"enables":[105],"weight":[106],"sharing,":[107],"reducing":[108],"off-chip":[109],"memory":[110],"access,":[111],"latency":[114],"without":[115],"requiring":[116],"retraining":[117],"or":[118],"fine-tuning.":[119],"For":[120],"hardware,":[121],"we":[122],"introduce":[123],"FSD-Acc,":[124],"an":[125],"area-":[126],"utilization-efficient":[128],"that":[131,143],"efficiently":[132],"executes":[133],"enabled":[137],"FSD-Infer.":[139],"Experimental":[140],"results":[141],"show":[142],"compared":[144],"with":[145],"inference,":[147],"reduces":[149],"EMA":[150],"up":[152,168,172],"26.11%,":[154],"enhances":[155],"arithmetic":[156],"intensity":[157],"<inline-formula":[159,174,211,218,226],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[160,175,212,219,227],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[161,176,213,220,228],"<tex-math":[162,177,214,221,229],"notation=\"LaTeX\">$10.26\\times":[163],"$</tex-math>":[164,179,216,223,231],"</inline-formula>,":[165,232],"speeds":[167],"GPU":[169],"notation=\"LaTeX\">$1.38\\times":[178],"</inline-formula>.":[180],"When":[181],"deployed":[182],"Xilinx":[184],"ZCU102":[185],"FPGA":[186],"board":[187],"at":[188],"200":[189],"MHz":[190],"working":[191],"frequency,":[192],"FSD-Acc":[193],"achieves":[194],"best":[196],"efficiency":[198],"energy":[200],"efficiency,":[201],"outperforming":[202],"state-of-the-art":[204],"notation=\"LaTeX\">$2.53\\times":[215],"</inline-formula>(TPS/kDSP),":[217],"notation=\"LaTeX\">$5.90\\times":[222],"</inline-formula>(TPS/kLUT)":[224],"notation=\"LaTeX\">$2.02\\times":[230],"respectively.":[233]},"counts_by_year":[],"updated_date":"2026-05-12T06:07:45.972803","created_date":"2026-02-13T00:00:00"}