{"id":"https://openalex.org/W7160262490","doi":"https://doi.org/10.1109/tcsii.2026.3690531","title":"PredLM: A Sparse Large Language Model Decoding Accelerator With Hardware-Efficient Zero-Activation Prediction","display_name":"PredLM: A Sparse Large Language Model Decoding Accelerator With Hardware-Efficient Zero-Activation Prediction","publication_year":2026,"publication_date":"2026-05-05","ids":{"openalex":"https://openalex.org/W7160262490","doi":"https://doi.org/10.1109/tcsii.2026.3690531"},"language":null,"primary_location":{"id":"doi:10.1109/tcsii.2026.3690531","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsii.2026.3690531","pdf_url":null,"source":{"id":"https://openalex.org/S93916849","display_name":"IEEE Transactions on Circuits & Systems II Express Briefs","issn_l":"1549-7747","issn":["1549-7747","1558-3791"],"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 Circuits and Systems II: Express Briefs","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/A5042758402","display_name":"Seunghyun Moon","orcid":"https://orcid.org/0000-0003-0027-2666"},"institutions":[{"id":"https://openalex.org/I24062138","display_name":"Konkuk University","ror":"https://ror.org/025h1m602","country_code":"KR","type":"education","lineage":["https://openalex.org/I24062138"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Seunghyun Moon","raw_affiliation_strings":["Konkuk University, Seoul, Republic of Korea"],"raw_orcid":"https://orcid.org/0000-0003-0027-2666","affiliations":[{"raw_affiliation_string":"Konkuk University, Seoul, Republic of Korea","institution_ids":["https://openalex.org/I24062138"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135399301","display_name":"Seowon Ji","orcid":"https://orcid.org/0000-0001-8700-8440"},"institutions":[{"id":"https://openalex.org/I24062138","display_name":"Konkuk University","ror":"https://ror.org/025h1m602","country_code":"KR","type":"education","lineage":["https://openalex.org/I24062138"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Seowon Ji","raw_affiliation_strings":["Konkuk University, Seoul, Republic of Korea"],"raw_orcid":"https://orcid.org/0000-0001-8700-8440","affiliations":[{"raw_affiliation_string":"Konkuk University, Seoul, Republic of Korea","institution_ids":["https://openalex.org/I24062138"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5083191379","display_name":"Sangwoo Hong","orcid":"https://orcid.org/0000-0002-0270-2781"},"institutions":[{"id":"https://openalex.org/I24062138","display_name":"Konkuk University","ror":"https://ror.org/025h1m602","country_code":"KR","type":"education","lineage":["https://openalex.org/I24062138"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Sangwoo Hong","raw_affiliation_strings":["Konkuk University, Seoul, Republic of Korea"],"raw_orcid":"https://orcid.org/0000-0002-0270-2781","affiliations":[{"raw_affiliation_string":"Konkuk University, Seoul, Republic of Korea","institution_ids":["https://openalex.org/I24062138"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5035447590","display_name":"Eunji Kwon","orcid":"https://orcid.org/0000-0001-5313-8976"},"institutions":[{"id":"https://openalex.org/I110273157","display_name":"Kookmin University","ror":"https://ror.org/0049erg63","country_code":"KR","type":"education","lineage":["https://openalex.org/I110273157"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Eunji Kwon","raw_affiliation_strings":["Kookmin University, Seoul, Republic of Korea"],"raw_orcid":"https://orcid.org/0000-0001-5313-8976","affiliations":[{"raw_affiliation_string":"Kookmin University, Seoul, Republic of Korea","institution_ids":["https://openalex.org/I110273157"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.53334372,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"73","issue":"6","first_page":"653","last_page":"657"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.1754000037908554,"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":0.1754000037908554,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.17239999771118164,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.15809999406337738,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.4936000108718872},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.301800012588501},{"id":"https://openalex.org/keywords/sparse-matrix","display_name":"Sparse matrix","score":0.2989000082015991},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.29789999127388},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.29109999537467957},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.2808000147342682}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6413000226020813},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.4936000108718872},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4677000045776367},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.40220001339912415},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.301800012588501},{"id":"https://openalex.org/C56372850","wikidata":"https://www.wikidata.org/wiki/Q1050404","display_name":"Sparse matrix","level":3,"score":0.2989000082015991},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.29789999127388},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.29109999537467957},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2808000147342682},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.2759999930858612},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2759999930858612},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.27129998803138733},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2694000005722046},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.25130000710487366}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tcsii.2026.3690531","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsii.2026.3690531","pdf_url":null,"source":{"id":"https://openalex.org/S93916849","display_name":"IEEE Transactions on Circuits & Systems II Express Briefs","issn_l":"1549-7747","issn":["1549-7747","1558-3791"],"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 Circuits and Systems II: Express Briefs","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G8506061720","display_name":null,"funder_award_id":"RS-2026-25480598","funder_id":"https://openalex.org/F4320322120","funder_display_name":"National Research Foundation of Korea"}],"funders":[{"id":"https://openalex.org/F4320322120","display_name":"National Research Foundation of Korea","ror":"https://ror.org/013aysd81"},{"id":"https://openalex.org/F4320322202","display_name":"IC Design Education Center","ror":"https://ror.org/005v57z85"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"This":[0],"brief":[1],"presents":[2],"PredLM,":[3],"a":[4,26,60,88,108,122],"hardware-algorithm":[5],"co-designed":[6],"accelerator":[7],"that":[8,36],"aggressively":[9],"reduces":[10],"external":[11],"memory":[12],"access":[13],"(EMA)":[14],"to":[15,71,94,144,154],"achieve":[16],"real-time":[17],"large":[18],"language":[19],"model":[20],"(LLM)":[21],"decoding.":[22],"First,":[23],"we":[24,58],"introduce":[25],"simple":[27],"but":[28],"effective":[29],"Key-Value":[30],"(KV)":[31],"cache":[32,41],"management":[33],"policy":[34],"so":[35],"the":[37,46,51,54,69,96,115],"size":[38],"of":[39,45,98,124,133],"KV":[40,55],"is":[42],"fixed":[43],"regardless":[44],"context":[47],"length,":[48],"significantly":[49],"reducing":[50],"EMA":[52],"for":[53,65],"cache.":[56],"Second,":[57],"propose":[59],"dynamic":[61],"zero-activation":[62],"prediction":[63],"scheme":[64],"feed-forward":[66],"networks,":[67],"allowing":[68],"hardware":[70],"dynamically":[72],"skip":[73],"unnecessary":[74],"multiply-accumulate":[75],"operations":[76],"and":[77,102,112,117,126,129,135,148],"weight":[78],"fetches.":[79],"To":[80],"seamlessly":[81],"support":[82,95],"these":[83],"predictive":[84],"algorithms,":[85],"PredLM":[86,120],"features":[87],"homogeneous":[89],"processing":[90],"element":[91],"array":[92],"able":[93],"execution":[97],"both":[99],"structural":[100],"sparse":[101],"dense":[103],"vector-matrix":[104],"multiplications.":[105],"Implemented":[106],"in":[107],"28-nm":[109],"CMOS":[110],"technology":[111],"evaluated":[113],"with":[114,138],"OPT-1.3B":[116],"Llama2-7B":[118],"models,":[119],"achieves":[121],"time-per-output-token":[123],"1.76":[125],"9.41":[127],"ms/token":[128],"an":[130],"energy":[131,151],"efficiency":[132,152],"19.77":[134],"105.09":[136],"mJ/token":[137],"negligible":[139],"perplexity":[140],"degradation,":[141],"delivering":[142],"up":[143],"18.1\u00d7":[145],"higher":[146],"throughput":[147],"4.92\u00d7":[149],"better":[150],"compared":[153],"state-of-the-art":[155],"LLM":[156],"accelerators.":[157]},"counts_by_year":[],"updated_date":"2026-05-30T06:14:24.967023","created_date":"2026-05-06T00:00:00"}