{"id":"https://openalex.org/W4398774961","doi":"https://doi.org/10.1109/lca.2024.3397747","title":"Exploiting Intel Advanced Matrix Extensions (AMX) for Large Language Model Inference","display_name":"Exploiting Intel Advanced Matrix Extensions (AMX) for Large Language Model Inference","publication_year":2024,"publication_date":"2024-01-01","ids":{"openalex":"https://openalex.org/W4398774961","doi":"https://doi.org/10.1109/lca.2024.3397747"},"language":"en","primary_location":{"id":"doi:10.1109/lca.2024.3397747","is_oa":true,"landing_page_url":"https://doi.org/10.1109/lca.2024.3397747","pdf_url":"https://ieeexplore.ieee.org/ielx7/10208/4357966/10538369.pdf","source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Computer Architecture Letters","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://ieeexplore.ieee.org/ielx7/10208/4357966/10538369.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5050020235","display_name":"Hyungyo Kim","orcid":"https://orcid.org/0000-0002-1157-0127"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hyungyo Kim","raw_affiliation_strings":["University of Illinois at Urbana-Champaign, Urbana, IL, USA"],"raw_orcid":"https://orcid.org/0000-0002-1157-0127","affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana-Champaign, Urbana, IL, USA","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5106525939","display_name":"Gaohan Ye","orcid":"https://orcid.org/0009-0005-7211-9750"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gaohan Ye","raw_affiliation_strings":["University of Illinois at Urbana-Champaign, Urbana, IL, USA"],"raw_orcid":"https://orcid.org/0009-0005-7211-9750","affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana-Champaign, Urbana, IL, USA","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030337371","display_name":"Nachuan Wang","orcid":"https://orcid.org/0009-0000-1431-0067"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Nachuan Wang","raw_affiliation_strings":["University of Illinois at Urbana-Champaign, Urbana, IL, USA"],"raw_orcid":"https://orcid.org/0009-0000-1431-0067","affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana-Champaign, Urbana, IL, USA","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070172290","display_name":"Amir Yazdanbakhsh","orcid":"https://orcid.org/0000-0001-8199-7671"},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Amir Yazdanbakhsh","raw_affiliation_strings":["Google Deepmind, Mountain View, CA, USA"],"raw_orcid":"https://orcid.org/0000-0001-8199-7671","affiliations":[{"raw_affiliation_string":"Google Deepmind, Mountain View, CA, USA","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5037648751","display_name":"Nam Sung Kim","orcid":"https://orcid.org/0000-0002-0442-5634"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Nam Sung Kim","raw_affiliation_strings":["University of Illinois at Urbana-Champaign, Urbana, IL, USA"],"raw_orcid":"https://orcid.org/0000-0002-0442-5634","affiliations":[{"raw_affiliation_string":"University of Illinois at Urbana-Champaign, Urbana, IL, USA","institution_ids":["https://openalex.org/I157725225"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":5.5569,"has_fulltext":false,"cited_by_count":20,"citation_normalized_percentile":{"value":0.96350885,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":96,"max":100},"biblio":{"volume":"23","issue":"1","first_page":"117","last_page":"120"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.782800018787384,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.782800018787384,"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/computer-science","display_name":"Computer science","score":0.7802668809890747},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.6418144702911377},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.571075975894928},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.46169599890708923},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.3775075078010559},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.33031508326530457},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.19623255729675293}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7802668809890747},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.6418144702911377},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.571075975894928},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.46169599890708923},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.3775075078010559},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.33031508326530457},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.19623255729675293}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/lca.2024.3397747","is_oa":true,"landing_page_url":"https://doi.org/10.1109/lca.2024.3397747","pdf_url":"https://ieeexplore.ieee.org/ielx7/10208/4357966/10538369.pdf","source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Computer Architecture Letters","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1109/lca.2024.3397747","is_oa":true,"landing_page_url":"https://doi.org/10.1109/lca.2024.3397747","pdf_url":"https://ieeexplore.ieee.org/ielx7/10208/4357966/10538369.pdf","source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Computer Architecture Letters","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":false},"content_urls":{"pdf":"https://content.openalex.org/works/W4398774961.pdf"},"referenced_works_count":7,"referenced_works":["https://openalex.org/W3001279689","https://openalex.org/W3006131567","https://openalex.org/W4321636575","https://openalex.org/W4385245566","https://openalex.org/W6772383348","https://openalex.org/W6774125022","https://openalex.org/W6850927664"],"related_works":["https://openalex.org/W2055243143","https://openalex.org/W2418291489","https://openalex.org/W2068121105","https://openalex.org/W3096519538","https://openalex.org/W2744747300","https://openalex.org/W4321636575","https://openalex.org/W4241166160","https://openalex.org/W2384826897","https://openalex.org/W1986418932","https://openalex.org/W1997466117"],"abstract_inverted_index":{"The":[0],"ever-increasing":[1],"number":[2,47],"of":[3,36,48,61,109,137,158,197],"parameters":[4,37,72],"in":[5,116,247],"Large":[6],"Language":[7],"Models":[8],"(LLMs)":[9],"demands":[10],"many":[11],"expensive":[12],"GPUs":[13],"for":[14,52,235],"both":[15,110,239],"inference":[16,211,237],"and":[17,94,113,168,178,228,241],"training.":[18],"This":[19],"is":[20,204],"because":[21],"even":[22],"such":[23,27,120],"a":[24,34,81,103,121,159],"high-end":[25],"GPU":[26,96],"as":[28],"NVIDIA":[29],"A100":[30],"can":[31,78],"store":[32,66,244],"only":[33,68],"subset":[35],"due":[38],"to":[39,65,162],"its":[40],"limited":[41],"memory":[42,59,83,175,189],"capacity.":[43,84],"To":[44,118],"reduce":[45],"the":[46,57,62,70,92,95,98,107,131,138,156,166,169,182,185,195,201,209,245],"required":[49],"GPUs,":[50],"especially":[51],"inference,":[53],"we":[54,123,147],"may":[55],"exploit":[56],"large":[58,188],"capacity":[60,176,190],"(host)":[63],"CPU":[64,93,167,183,248],"not":[67],"all":[69],"model":[71,151,246],"but":[73,191],"also":[74,79],"intermediate":[75],"outputs":[76],"which":[77],"require":[80],"substantial":[82],"However,":[85],"this":[86,222],"necessitates":[87],"frequent":[88],"data":[89,198],"transfers":[90],"between":[91],"over":[97],"slow":[99],"PCIe":[100,202],"interface,":[101],"creating":[102],"bottleneck":[104],"that":[105,129,154,216],"hinders":[106],"accomplishment":[108],"low":[111,192],"latency":[112,227],"high":[114],"throughput":[115,231],"inference.":[117],"address":[119],"challenge,":[122],"first":[124],"propose":[125,148],"CPU-GPU":[126,217,240],"cooperative":[127,218],"computing":[128,234,243],"exploits":[130],"Advanced":[132],"Matrix":[133],"Extensions":[134],"(AMX)":[135],"capability":[136],"latest":[139],"Intel":[140],"CPU,":[141],"codenamed":[142],"Sapphire":[143],"Rapids":[144],"(SPR).":[145],"Second,":[146],"an":[149],"adaptive":[150],"partitioning":[152],"policy":[153],"determines":[155],"layers":[157,186],"given":[160],"LLM":[161,210],"be":[163],"run":[164],"on":[165,173,221],"GPU,":[170],"respectively,":[171],"based":[172,220],"their":[174],"requirement":[177],"arithmetic":[179,193],"intensity.":[180],"As":[181],"executes":[184],"with":[187],"intensity,":[194],"amount":[196],"transferred":[199],"through":[200],"interface":[203],"significantly":[205],"reduced,":[206],"thereby":[207],"improving":[208],"performance.":[212],"Our":[213],"evaluation":[214],"demonstrates":[215],"computing,":[219],"policy,":[223],"delivers":[224],"12.1\u00d7":[225],"lower":[226],"5.4\u00d7":[229],"higher":[230],"than":[232],"GPU-only":[233,242],"OPT-30B":[236],"when":[238],"memory.":[249]},"counts_by_year":[{"year":2026,"cited_by_count":8},{"year":2025,"cited_by_count":9},{"year":2024,"cited_by_count":3}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2024-05-25T00:00:00"}
