{"id":"https://openalex.org/W4385438314","doi":"https://doi.org/10.1145/3609510.3609815","title":"Improving Throughput-oriented Generative Inference with CPUs","display_name":"Improving Throughput-oriented Generative Inference with CPUs","publication_year":2023,"publication_date":"2023-07-31","ids":{"openalex":"https://openalex.org/W4385438314","doi":"https://doi.org/10.1145/3609510.3609815"},"language":"en","primary_location":{"id":"doi:10.1145/3609510.3609815","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3609510.3609815","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 14th ACM SIGOPS Asia-Pacific Workshop on Systems","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/A5051797786","display_name":"Daon Park","orcid":"https://orcid.org/0000-0003-2312-3049"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":true,"raw_author_name":"Daon Park","raw_affiliation_strings":["Seoul National University, Seoul, Republic of Korea"],"raw_orcid":"https://orcid.org/0000-0003-2312-3049","affiliations":[{"raw_affiliation_string":"Seoul National University, Seoul, Republic of Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101703485","display_name":"Sungbin Jo","orcid":"https://orcid.org/0009-0002-4844-5378"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Sungbin Jo","raw_affiliation_strings":["Seoul National University, Seoul, Republic of Korea"],"raw_orcid":"https://orcid.org/0009-0002-4844-5378","affiliations":[{"raw_affiliation_string":"Seoul National University, Seoul, Republic of Korea","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5044493612","display_name":"Bernhard Egger","orcid":"https://orcid.org/0000-0002-6645-6161"},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Bernhard Egger","raw_affiliation_strings":["Seoul National University, Seoul, Republic of Korea"],"raw_orcid":"https://orcid.org/0000-0002-6645-6161","affiliations":[{"raw_affiliation_string":"Seoul National University, Seoul, Republic of Korea","institution_ids":["https://openalex.org/I139264467"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5051797786"],"corresponding_institution_ids":["https://openalex.org/I139264467"],"apc_list":null,"apc_paid":null,"fwci":0.5112,"has_fulltext":false,"cited_by_count":3,"citation_normalized_percentile":{"value":0.70958606,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":96},"biblio":{"volume":null,"issue":null,"first_page":"37","last_page":"42"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9998000264167786,"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/T10028","display_name":"Topic Modeling","score":0.9998000264167786,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9961000084877014,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9955000281333923,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.834203839302063},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.7040760517120361},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.6964045166969299},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6591801643371582},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.5592759251594543},{"id":"https://openalex.org/keywords/leverage","display_name":"Leverage (statistics)","score":0.5069639682769775},{"id":"https://openalex.org/keywords/dram","display_name":"Dram","score":0.46386733651161194},{"id":"https://openalex.org/keywords/general-purpose-computing-on-graphics-processing-units","display_name":"General-purpose computing on graphics processing units","score":0.4249448776245117},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.2495097517967224},{"id":"https://openalex.org/keywords/graphics","display_name":"Graphics","score":0.16583412885665894},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.1126793920993805},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.10175269842147827}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.834203839302063},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.7040760517120361},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.6964045166969299},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6591801643371582},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.5592759251594543},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.5069639682769775},{"id":"https://openalex.org/C7366592","wikidata":"https://www.wikidata.org/wiki/Q1255620","display_name":"Dram","level":2,"score":0.46386733651161194},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.4249448776245117},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.2495097517967224},{"id":"https://openalex.org/C21442007","wikidata":"https://www.wikidata.org/wiki/Q1027879","display_name":"Graphics","level":2,"score":0.16583412885665894},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.1126793920993805},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.10175269842147827},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.0},{"id":"https://openalex.org/C555944384","wikidata":"https://www.wikidata.org/wiki/Q249","display_name":"Wireless","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3609510.3609815","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3609510.3609815","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 14th ACM SIGOPS Asia-Pacific Workshop on Systems","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9","score":0.41999998688697815}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":4,"referenced_works":["https://openalex.org/W2984100107","https://openalex.org/W3131724164","https://openalex.org/W3165031368","https://openalex.org/W3185341429"],"related_works":["https://openalex.org/W3120961607","https://openalex.org/W4401568740","https://openalex.org/W2098207691","https://openalex.org/W3148568549","https://openalex.org/W1648516568","https://openalex.org/W361036515","https://openalex.org/W2269474412","https://openalex.org/W4211178602","https://openalex.org/W2433923775","https://openalex.org/W2537599394"],"abstract_inverted_index":{"Despite":[0],"recent":[1],"attempts":[2],"to":[3,21,49,77],"reduce":[4],"the":[5,28,62,66,74,83,86,93,97,105,115],"number":[6],"of":[7,9,30,57,61,73,81,100,127],"parameters":[8],"large":[10,20,53],"language":[11],"models":[12,54],"(LLMs),":[13],"their":[14],"parameter":[15],"data":[16,71,119],"is":[17,68],"still":[18],"too":[19],"fit":[22],"into":[23],"a":[24,40,78,108,124],"single":[25,41],"GPU.":[26],"With":[27],"emergence":[29],"throughput-oriented":[31],"tasks,":[32],"high-throughput":[33],"generative":[34],"inference":[35,51],"frameworks":[36,102],"for":[37,138],"LLMs":[38],"on":[39,52,114],"commodity":[42],"GPU":[43,84,118],"leverage":[44],"GPU,":[45],"DRAM,":[46],"and":[47,85,95,111,132],"NVMe":[48],"run":[50],"with":[55,117,135],"terabytes":[56],"data.":[58],"Our":[59,121],"analysis":[60],"technique":[63],"shows":[64,123],"that":[65],"runtime":[67],"dominated":[69],"by":[70,103],"transfers":[72],"weights,":[75],"leading":[76],"low":[79],"utilization":[80],"both":[82],"CPU.":[87],"In":[88],"this":[89],"paper,":[90],"we":[91],"increase":[92],"throughput":[94,131],"decrease":[96],"total":[98,133],"latency":[99],"state-of-the-art":[101],"including":[104],"CPU":[106,116],"as":[107],"compute":[109],"device":[110],"overlapping":[112],"computations":[113],"transfers.":[120],"work":[122],"promising":[125],"improvement":[126],"around":[128],"40%":[129],"in":[130],"latency,":[134],"potential":[136],"room":[137],"further":[139],"improvements.":[140]},"counts_by_year":[{"year":2025,"cited_by_count":2},{"year":2024,"cited_by_count":1}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}