{"id":"https://openalex.org/W4413822383","doi":"https://doi.org/10.1109/tcad.2025.3604321","title":"APT-LLM: Exploiting Arbitrary-Precision Tensor Core Computing for LLM Acceleration","display_name":"APT-LLM: Exploiting Arbitrary-Precision Tensor Core Computing for LLM Acceleration","publication_year":2025,"publication_date":"2025-08-29","ids":{"openalex":"https://openalex.org/W4413822383","doi":"https://doi.org/10.1109/tcad.2025.3604321"},"language":"en","primary_location":{"id":"doi:10.1109/tcad.2025.3604321","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2025.3604321","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"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 Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-article"},"type":"article","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2508.19087","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5026653305","display_name":"Shaobo Ma","orcid":null},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shaobo Ma","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China","School of Electronic Science and Engineering, Nanjing University, China"],"raw_orcid":"https://orcid.org/0009-0001-6173-3983","affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]},{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066888122","display_name":"Chao Fang","orcid":"https://orcid.org/0000-0003-3430-1189"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chao Fang","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China","School of Electronic Science and Engineering, Nanjing University, China"],"raw_orcid":"https://orcid.org/0000-0003-3430-1189","affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]},{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090170161","display_name":"Haikuo Shao","orcid":"https://orcid.org/0009-0008-6965-3436"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haikuo Shao","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China","School of Electronic Science and Engineering, Nanjing University, China"],"raw_orcid":"https://orcid.org/0009-0008-6965-3436","affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]},{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100696999","display_name":"Zhongfeng Wang","orcid":"https://orcid.org/0000-0002-7227-4786"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhongfeng Wang","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China","School of Electronic Science and Engineering, Nanjing University, China"],"raw_orcid":"https://orcid.org/0000-0002-7227-4786","affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]},{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, China","institution_ids":["https://openalex.org/I881766915"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I881766915"],"apc_list":null,"apc_paid":null,"fwci":0.574,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.63958325,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":97,"max":98},"biblio":{"volume":"45","issue":"4","first_page":"1935","last_page":"1948"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10559","display_name":"Particle Accelerators and Free-Electron Lasers","score":0.9887999892234802,"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/T10559","display_name":"Particle Accelerators and Free-Electron Lasers","score":0.9887999892234802,"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/T10346","display_name":"Magnetic confinement fusion research","score":0.9843000173568726,"subfield":{"id":"https://openalex.org/subfields/3106","display_name":"Nuclear and High Energy Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11367","display_name":"Particle accelerators and beam dynamics","score":0.9805999994277954,"subfield":{"id":"https://openalex.org/subfields/2202","display_name":"Aerospace Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.7117336988449097},{"id":"https://openalex.org/keywords/core","display_name":"Core (optical fiber)","score":0.6358559727668762},{"id":"https://openalex.org/keywords/tensor","display_name":"Tensor (intrinsic definition)","score":0.5502071380615234},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5040234923362732},{"id":"https://openalex.org/keywords/computational-science","display_name":"Computational science","score":0.35710281133651733},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.354034960269928},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.2661741375923157},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.24454128742218018},{"id":"https://openalex.org/keywords/geometry","display_name":"Geometry","score":0.19818946719169617},{"id":"https://openalex.org/keywords/classical-mechanics","display_name":"Classical mechanics","score":0.11461514234542847},{"id":"https://openalex.org/keywords/telecommunications","display_name":"Telecommunications","score":0.06840759515762329}],"concepts":[{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.7117336988449097},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.6358559727668762},{"id":"https://openalex.org/C155281189","wikidata":"https://www.wikidata.org/wiki/Q3518150","display_name":"Tensor (intrinsic definition)","level":2,"score":0.5502071380615234},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5040234923362732},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.35710281133651733},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.354034960269928},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2661741375923157},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.24454128742218018},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.19818946719169617},{"id":"https://openalex.org/C74650414","wikidata":"https://www.wikidata.org/wiki/Q11397","display_name":"Classical mechanics","level":1,"score":0.11461514234542847},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.06840759515762329}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/tcad.2025.3604321","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcad.2025.3604321","pdf_url":null,"source":{"id":"https://openalex.org/S100835903","display_name":"IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems","issn_l":"0278-0070","issn":["0278-0070","1937-4151"],"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 Computer-Aided Design of Integrated Circuits and Systems","raw_type":"journal-article"},{"id":"pmh:oai:lirias2repo.kuleuven.be:20.500.12942/787106","is_oa":false,"landing_page_url":"https://lirias.kuleuven.be/handle/20.500.12942/787106","pdf_url":null,"source":{"id":"https://openalex.org/S7407055369","display_name":"Lirias","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"Ieee Transactions On Computer-Aided Design Of Integrated Circuits And Systems, vol. 45 (4), (1935-1948)","raw_type":"info:eu-repo/semantics/article"},{"id":"pmh:oai:arXiv.org:2508.19087","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2508.19087","pdf_url":"https://arxiv.org/pdf/2508.19087","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2508.19087","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2508.19087","pdf_url":"https://arxiv.org/pdf/2508.19087","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6422954184","display_name":null,"funder_award_id":"2022YFB4400600","funder_id":"https://openalex.org/F4320335777","funder_display_name":"National Key Research and Development Program of China"},{"id":"https://openalex.org/G7961932763","display_name":null,"funder_award_id":"KYCX24_0149","funder_id":"https://openalex.org/F4320327780","funder_display_name":"Key Research and Development Program of Jiangxi Province"}],"funders":[{"id":"https://openalex.org/F4320327780","display_name":"Key Research and Development Program of Jiangxi Province","ror":null},{"id":"https://openalex.org/F4320335777","display_name":"National Key Research and Development Program of China","ror":null}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2051487156","https://openalex.org/W2073681303","https://openalex.org/W4409439182","https://openalex.org/W2565094479","https://openalex.org/W2390829436","https://openalex.org/W1989791859","https://openalex.org/W602859758","https://openalex.org/W1971289376","https://openalex.org/W2379101322","https://openalex.org/W2053286651"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"have":[4],"revolutionized":[5],"AI":[6],"applications,":[7],"yet":[8],"their":[9],"enormous":[10],"computational":[11,24],"demands":[12],"severely":[13],"limit":[14],"deployment":[15],"and":[16,58,90,118,130,162,194,211,226,236],"real-time":[17],"performance.":[18],"Quantization":[19],"methods":[20],"can":[21],"help":[22],"reduce":[23,163],"costs,":[25],"however,":[26],"attaining":[27],"the":[28,48,122,132,177],"extreme":[29],"efficiency":[30],"associated":[31],"with":[32,93],"ultra-low-bit":[33],"quantized":[34],"LLMs":[35],"at":[36,121],"arbitrary":[37,73,114],"precision":[38,74,115,129,195],"presents":[39],"challenges":[40],"on":[41,147,220],"GPUs.":[42],"This":[43,125],"is":[44],"primarily":[45],"due":[46],"to":[47,101,156,203,208,231],"limited":[49],"support":[50],"for":[51,72,88,113,183],"GPU":[52,135],"Tensor":[53,136],"Cores,":[54],"inefficient":[55],"memory":[56,143,155,164],"management,":[57],"inflexible":[59],"kernel":[60,159,171],"optimizations.":[61],"To":[62],"tackle":[63],"these":[64],"challenges,":[65],"we":[66,79,140,168],"propose":[67,141],"a":[68,81,107,142,170,204,212],"comprehensive":[69],"acceleration":[70,219],"scheme":[71],"LLMs,":[75],"namely":[76],"APT-LLM.":[77],"Firstly,":[78],"introduce":[80],"novel":[82],"data":[83,148],"format,":[84],"bipolar-INT,":[85],"which":[86,150],"allows":[87],"efficient":[89],"lossless":[91],"conversion":[92],"signed":[94],"INT,":[95],"while":[96],"also":[97,105],"being":[98],"more":[99],"conducive":[100],"parallel":[102],"computation.":[103],"We":[104],"develop":[106,169],"matrix":[108,185],"multiplication":[109],"(MatMul)":[110],"method":[111,126,173],"allowing":[112],"by":[116],"dismantling":[117],"reassembling":[119],"matrices":[120],"bit":[123],"level.":[124],"provides":[127],"flexible":[128],"optimizes":[131],"utilization":[133],"of":[134,181],"Cores.":[137],"In":[138,197],"addition,":[139],"management":[144],"system":[145],"focused":[146],"recovery,":[149],"strategically":[151],"employs":[152],"fast":[153],"shared":[154],"substantially":[157],"increase":[158],"execution":[160],"speed":[161],"access":[165],"latency.":[166],"Finally,":[167],"mapping":[172],"that":[174],"dynamically":[175],"selects":[176],"optimal":[178,188],"configurable":[179],"hyperparameters":[180],"kernels":[182],"varying":[184],"sizes,":[186],"enabling":[187],"performance":[189],"across":[190],"different":[191],"LLM":[192,198],"architectures":[193],"settings.":[196],"inference,":[199],"APT-LLM":[200,228],"achieves":[201,229],"up":[202,230],"3.99\u00d7":[205],"speedup":[206,214,233,238],"compared":[207],"FP16":[209,235],"baselines":[210],"2.16\u00d7":[213],"over":[215,234,239],"NVIDIA":[216],"CUTLASS":[217,240],"INT4":[218],"RTX":[221,224],"3090.":[222],"On":[223],"4090":[225],"H800,":[227],"2.44\u00d7":[232],"1.65\u00d7":[237],"integer":[241],"baselines.":[242]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-08-11T07:18:39.950985","created_date":"2025-10-10T00:00:00"}
