{"id":"https://openalex.org/W4409282445","doi":"https://doi.org/10.1145/3676536.3676796","title":"Fast and Efficient 2-bit LLM Inference on GPU: 2/4/16-bit in a Weight Matrix with Asynchronous Dequantization","display_name":"Fast and Efficient 2-bit LLM Inference on GPU: 2/4/16-bit in a Weight Matrix with Asynchronous Dequantization","publication_year":2024,"publication_date":"2024-10-27","ids":{"openalex":"https://openalex.org/W4409282445","doi":"https://doi.org/10.1145/3676536.3676796"},"language":"en","primary_location":{"id":"doi:10.1145/3676536.3676796","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3676536.3676796","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676796","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 43rd IEEE/ACM International Conference on Computer-Aided Design","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676796","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5068894182","display_name":"Jinhao Li","orcid":"https://orcid.org/0009-0009-4286-6359"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Jinhao Li","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111878216","display_name":"Jianrong Xu","orcid":"https://orcid.org/0009-0000-7000-6537"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiaming Xu","raw_affiliation_strings":["Shanghai Jiao Tong University & Infinigence-AI, Shanghai, China"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University & Infinigence-AI, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005872704","display_name":"Shiyao Li","orcid":"https://orcid.org/0009-0007-4755-6881"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shiyao Li","raw_affiliation_strings":["Tsinghua University & Infinigence-AI, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Tsinghua University & Infinigence-AI, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109113870","display_name":"Shan Huang","orcid":"https://orcid.org/0009-0005-2012-8540"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shan Huang","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058051415","display_name":"J Liu","orcid":"https://orcid.org/0009-0003-8280-9072"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jun Liu","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102585583","display_name":"Yaoxiu Lian","orcid":"https://orcid.org/0009-0007-7858-5132"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yaoxiu Lian","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5015946486","display_name":"Guohao Dai","orcid":"https://orcid.org/0000-0003-0849-3252"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guohao Dai","raw_affiliation_strings":["Shanghai Jiao Tong University & Infinigence-AI, Shanghai, China"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University & Infinigence-AI, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":7,"corresponding_author_ids":["https://openalex.org/A5068894182"],"corresponding_institution_ids":["https://openalex.org/I183067930"],"apc_list":null,"apc_paid":null,"fwci":1.3627,"has_fulltext":true,"cited_by_count":4,"citation_normalized_percentile":{"value":0.85285899,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"9"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11269","display_name":"Algorithms and Data Compression","score":0.9983000159263611,"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"}},"topics":[{"id":"https://openalex.org/T11269","display_name":"Algorithms and Data Compression","score":0.9983000159263611,"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/T10601","display_name":"Handwritten Text Recognition Techniques","score":0.9889000058174133,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.986299991607666,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/bit","display_name":"Bit (key)","score":0.7328831553459167},{"id":"https://openalex.org/keywords/asynchronous-communication","display_name":"Asynchronous communication","score":0.6682196259498596},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6332138180732727},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.5250532627105713},{"id":"https://openalex.org/keywords/matrix","display_name":"Matrix (chemical analysis)","score":0.41742315888404846},{"id":"https://openalex.org/keywords/computational-science","display_name":"Computational science","score":0.3496771454811096},{"id":"https://openalex.org/keywords/materials-science","display_name":"Materials science","score":0.12044388055801392},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.10608509182929993}],"concepts":[{"id":"https://openalex.org/C117011727","wikidata":"https://www.wikidata.org/wiki/Q1278488","display_name":"Bit (key)","level":2,"score":0.7328831553459167},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.6682196259498596},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6332138180732727},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5250532627105713},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.41742315888404846},{"id":"https://openalex.org/C459310","wikidata":"https://www.wikidata.org/wiki/Q117801","display_name":"Computational science","level":1,"score":0.3496771454811096},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.12044388055801392},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.10608509182929993},{"id":"https://openalex.org/C159985019","wikidata":"https://www.wikidata.org/wiki/Q181790","display_name":"Composite material","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3676536.3676796","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3676536.3676796","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676796","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 43rd IEEE/ACM International Conference on Computer-Aided Design","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3676536.3676796","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3676536.3676796","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3676536.3676796","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 43rd IEEE/ACM International Conference on Computer-Aided Design","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2087396116","display_name":null,"funder_award_id":"China","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G3317480652","display_name":null,"funder_award_id":"Science","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G5994120800","display_name":null,"funder_award_id":"Natural","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6803541034","display_name":null,"funder_award_id":"U21B2031","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G8262591375","display_name":null,"funder_award_id":"62104128","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":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4409282445.pdf","grobid_xml":"https://content.openalex.org/works/W4409282445.grobid-xml"},"referenced_works_count":9,"referenced_works":["https://openalex.org/W2154347068","https://openalex.org/W2998183051","https://openalex.org/W3034765865","https://openalex.org/W3100985894","https://openalex.org/W4292474994","https://openalex.org/W4323848232","https://openalex.org/W4382246105","https://openalex.org/W4390043316","https://openalex.org/W4406650295"],"related_works":["https://openalex.org/W4327546585","https://openalex.org/W2411923897","https://openalex.org/W2116677773","https://openalex.org/W4394546135","https://openalex.org/W4285347720","https://openalex.org/W4200259850","https://openalex.org/W2333831899","https://openalex.org/W2484894494","https://openalex.org/W2367385042","https://openalex.org/W2155261584"],"abstract_inverted_index":{"Large":[0,107],"language":[1],"models":[2],"(LLMs)":[3],"have":[4],"demonstrated":[5],"impressive":[6],"abilities":[7],"in":[8,66,98,133,137],"various":[9],"domains":[10],"while":[11,74],"the":[12,44,57,90,122,126,155,159],"inference":[13,26],"cost":[14,27],"is":[15,163],"expensive.":[16],"Many":[17],"previous":[18],"studies":[19],"exploit":[20],"quantization":[21,38,87],"methods":[22,46,49,83,146],"to":[23,55,151,176],"reduce":[24],"LLM":[25],"by":[28,72,110,125],"reducing":[29],"latency":[30],"and":[31,88,104,158],"memory":[32],"consumption.":[33],"Applying":[34],"2-bit":[35],"single-precision":[36],"weight":[37,67,95],"brings":[39],">3%":[40],"accuracy":[41,100,118],"loss,":[42],"so":[43],"state-of-the-art":[45],"use":[47],"mixed-precision":[48,86],"for":[50],"LLMs":[51],"(e.g.":[52,102,129],"Llama2-7b,":[53],"etc.)":[54],"improve":[56],"accuracy.":[58],"However,":[59],"challenges":[60],"still":[61],"exist:":[62],"(1)":[63],"Uneven":[64],"distribution":[65],"matrix.":[68],"Weights":[69],"are":[70,169],"quantized":[71,156],"groups,":[73],"some":[75],"groups":[76,168],"contain":[77],"weights":[78],"with":[79],"large":[80],"range.":[81],"Previous":[82],"apply":[84],"inter-weight":[85],"neglect":[89],"range":[91],"difference":[92],"inside":[93],"each":[94],"matrix,":[96],"resulting":[97,132],">2.7%":[99],"loss":[101],"LLM-MQ":[103],"APTQ).":[105],"(2)":[106],"speed":[108,123,135],"degradation":[109,136],"adding":[111],"sparse":[112,115],"outliers.":[113],"Reserving":[114],"outliers":[116,131],"improves":[117],"but":[119],"slows":[120],"down":[121],"affected":[124],"outlier":[127],"ratio":[128],"1.5%":[130],">30%":[134],"SpQR).":[138],"(3)":[139],"Time-consuming":[140],"dequantization":[141,149,161,173],"operations":[142,174],"on":[143,154],"GPUs.":[144],"Mainstream":[145],"require":[147],"a":[148],"operation":[150,162],"perform":[152],"computation":[153],"weights,":[157],"2-order":[160],"applied":[164],"because":[165],"scales":[166],"of":[167],"also":[170],"quantized.":[171],"These":[172],"lead":[175],">50%":[177],"execution":[178],"time.":[179]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":2}],"updated_date":"2026-04-10T15:06:20.359241","created_date":"2025-10-10T00:00:00"}