{"id":"https://openalex.org/W4414197738","doi":"https://doi.org/10.1109/dac63849.2025.11132578","title":"OutlierCIM: Outlier-Aware Digital CIM-Based LLM Accelerator with Hybrid-Strategy Quantization and Unified FP-INT Computation","display_name":"OutlierCIM: Outlier-Aware Digital CIM-Based LLM Accelerator with Hybrid-Strategy Quantization and Unified FP-INT Computation","publication_year":2025,"publication_date":"2025-06-22","ids":{"openalex":"https://openalex.org/W4414197738","doi":"https://doi.org/10.1109/dac63849.2025.11132578"},"language":"en","primary_location":{"id":"doi:10.1109/dac63849.2025.11132578","is_oa":false,"landing_page_url":"https://doi.org/10.1109/dac63849.2025.11132578","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 62nd ACM/IEEE Design Automation Conference (DAC)","raw_type":"proceedings-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/A5062140266","display_name":"Zihan Zou","orcid":"https://orcid.org/0009-0002-1510-1341"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Zihan Zou","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5104310278","display_name":"Shikuang Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shikuang Chen","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101228383","display_name":"Chen Zhang","orcid":"https://orcid.org/0009-0000-1312-422X"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chen Zhang","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120790435","display_name":"Xing Wang","orcid":"https://orcid.org/0000-0003-4711-5676"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xing Wang","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100459795","display_name":"Zhichao Liu","orcid":"https://orcid.org/0000-0001-6805-7095"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhichao Liu","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020272573","display_name":"Haoran Du","orcid":"https://orcid.org/0000-0002-7949-4909"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haoran Du","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000787204","display_name":"Xin Si","orcid":"https://orcid.org/0000-0002-4993-0087"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xin Si","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079498326","display_name":"Hao Cai","orcid":"https://orcid.org/0000-0001-9794-8049"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hao Cai","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100461621","display_name":"Bo Liu","orcid":"https://orcid.org/0000-0002-0894-1054"},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bo Liu","raw_affiliation_strings":["Southeast University,School of Integrated Circuits,Nanjing,China"],"affiliations":[{"raw_affiliation_string":"Southeast University,School of Integrated Circuits,Nanjing,China","institution_ids":["https://openalex.org/I76569877"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":9,"corresponding_author_ids":["https://openalex.org/A5062140266"],"corresponding_institution_ids":["https://openalex.org/I76569877"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.23410377,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"7"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T14276","display_name":"Power Systems and Technologies","score":0.949400007724762,"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/T14276","display_name":"Power Systems and Technologies","score":0.949400007724762,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9332000017166138,"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/quantization","display_name":"Quantization (signal processing)","score":0.7748000025749207},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.7551000118255615},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5795000195503235},{"id":"https://openalex.org/keywords/efficient-energy-use","display_name":"Efficient energy use","score":0.5490000247955322},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.46219998598098755},{"id":"https://openalex.org/keywords/hardware-acceleration","display_name":"Hardware acceleration","score":0.4519999921321869},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.4189999997615814}],"concepts":[{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.7748000025749207},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7641000151634216},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.7551000118255615},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5795000195503235},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.5490000247955322},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.46219998598098755},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.4519999921321869},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.4357999861240387},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.4189999997615814},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.38269999623298645},{"id":"https://openalex.org/C166955791","wikidata":"https://www.wikidata.org/wiki/Q629579","display_name":"Macro","level":2,"score":0.3785000145435333},{"id":"https://openalex.org/C46362747","wikidata":"https://www.wikidata.org/wiki/Q173431","display_name":"CMOS","level":2,"score":0.3693000078201294},{"id":"https://openalex.org/C79337645","wikidata":"https://www.wikidata.org/wiki/Q779824","display_name":"Outlier","level":2,"score":0.3508000075817108},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.35040000081062317},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.32679998874664307},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.31380000710487366},{"id":"https://openalex.org/C134835016","wikidata":"https://www.wikidata.org/wiki/Q690265","display_name":"Lookup table","level":2,"score":0.29750001430511475},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.27720001339912415},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.27309998869895935},{"id":"https://openalex.org/C2780165032","wikidata":"https://www.wikidata.org/wiki/Q16869822","display_name":"Energy consumption","level":2,"score":0.2705000042915344},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.2694000005722046}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/dac63849.2025.11132578","is_oa":false,"landing_page_url":"https://doi.org/10.1109/dac63849.2025.11132578","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 62nd ACM/IEEE Design Automation Conference (DAC)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"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":15,"referenced_works":["https://openalex.org/W2207050309","https://openalex.org/W2252136820","https://openalex.org/W2883920103","https://openalex.org/W3100985894","https://openalex.org/W4214686755","https://openalex.org/W4252402916","https://openalex.org/W4313332258","https://openalex.org/W4360605703","https://openalex.org/W4366341968","https://openalex.org/W4386765206","https://openalex.org/W4390873361","https://openalex.org/W4393407021","https://openalex.org/W4401211807","https://openalex.org/W4404132796","https://openalex.org/W4404133703"],"related_works":[],"abstract_inverted_index":{"Activation":[0],"outliers":[1],"in":[2,55,154],"Large":[3],"Language":[4],"Models":[5],"(LLMs),":[6],"which":[7,101],"exhibit":[8],"large":[9],"magnitudes":[10],"but":[11],"small":[12],"quantities,":[13],"significantly":[14],"affect":[15],"model":[16,47],"performance":[17],"and":[18,39,69,97,111,124,131,138,147,181],"pose":[19],"challenges":[20,45],"for":[21,73],"the":[22,66,74,120,145,187],"acceleration":[23],"of":[24,85,127,149,164],"LLMs.":[25],"To":[26,60],"address":[27],"this":[28,61],"bottleneck,":[29],"researchers":[30],"have":[31],"proposed":[32],"several":[33],"co-design":[34],"frameworks":[35],"with":[36,49,78],"outlier-aware":[37,79,189],"algorithms":[38],"dedicated":[40,140],"hardware.":[41],"However,":[42],"they":[43],"face":[44],"balancing":[46],"accuracy":[48],"hardware":[50,70],"efficiency":[51,179],"when":[52],"accelerating":[53],"LLMs":[54],"a":[56,108,112,133,139,155],"low":[57,121],"bit-width":[58],"manner.":[59],"end,":[62],"we":[63],"propose":[64],"OutlierCIM,":[65],"first":[67],"algorithm":[68],"codesign":[71],"framework":[72],"compute-in-memory":[75],"(CIM)":[76],"accelerator":[77],"quantization":[80,110,134],"algorithm.":[81],"The":[82],"key":[83],"contributions":[84],"OutlierCIM":[86,160,172],"are":[87,102],"1)":[88],"an":[89,162],"outlier-clustered":[90],"tiling":[91],"strategy":[92,137],"that":[93,118,142],"regulates":[94],"memory":[95],"access":[96],"reduces":[98],"inefficient":[99],"workloads":[100],"both":[103],"introduced":[104],"by":[105],"outliers,":[106],"2)":[107],"hybrid-strategy":[109],"reconfigurable":[113],"double-bit":[114],"CIM":[115],"macro":[116],"array":[117],"overcome":[119],"storage":[122],"utilization":[123],"high":[125],"latency":[126],"outlier-based":[128],"LLM":[129],"quantization,":[130],"3)":[132],"factor":[135],"post-processing":[136],"quantizer":[141],"efficiently":[143],"unify":[144],"multiplication":[146],"accumulation":[148],"outlier-caused":[150],"FP-INT":[151],"workloads.":[152],"Implemented":[153],"28":[156],"nm":[157],"CMOS":[158],"technology,":[159],"occupies":[161],"area":[163],"$2.25":[165],"\\mathrm{~mm}^{2}$.":[166],"When":[167],"evaluated":[168],"at":[169],"comprehensive":[170],"benchmarks,":[171],"achieves":[173],"up":[174],"to":[175,186],"$4.54":[176],"\\times$":[177,183],"energy":[178],"improvement":[180],"$3.91":[182],"speedup":[184],"compared":[185],"state-of-the-art":[188],"accelerators.":[190]},"counts_by_year":[],"updated_date":"2026-04-09T08:11:56.329763","created_date":"2025-10-10T00:00:00"}
