{"id":"https://openalex.org/W4389159500","doi":"https://doi.org/10.1109/iccad57390.2023.10323944","title":"Automatic Kernel Generation for Large Language Models on Deep Learning Accelerators","display_name":"Automatic Kernel Generation for Large Language Models on Deep Learning Accelerators","publication_year":2023,"publication_date":"2023-10-28","ids":{"openalex":"https://openalex.org/W4389159500","doi":"https://doi.org/10.1109/iccad57390.2023.10323944"},"language":"en","primary_location":{"id":"doi:10.1109/iccad57390.2023.10323944","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccad57390.2023.10323944","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/ACM International Conference on Computer Aided Design (ICCAD)","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/A5101534623","display_name":"Fuyu Wang","orcid":"https://orcid.org/0000-0003-3165-873X"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Fuyu Wang","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-Sen University,Guangzhou,China","School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China"],"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-Sen University,Guangzhou,China","institution_ids":["https://openalex.org/I157773358"]},{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5002840650","display_name":"Minghua Shen","orcid":"https://orcid.org/0000-0003-4747-8020"},"institutions":[{"id":"https://openalex.org/I157773358","display_name":"Sun Yat-sen University","ror":"https://ror.org/0064kty71","country_code":"CN","type":"education","lineage":["https://openalex.org/I157773358"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Minghua Shen","raw_affiliation_strings":["School of Computer Science and Engineering, Sun Yat-Sen University,Guangzhou,China","School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China"],"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-Sen University,Guangzhou,China","institution_ids":["https://openalex.org/I157773358"]},{"raw_affiliation_string":"School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou, China","institution_ids":["https://openalex.org/I157773358"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5101534623"],"corresponding_institution_ids":["https://openalex.org/I157773358"],"apc_list":null,"apc_paid":null,"fwci":0.7006,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.76529148,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":97},"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/T10028","display_name":"Topic Modeling","score":0.9987999796867371,"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":0.9987999796867371,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9894000291824341,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.989300012588501,"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/speedup","display_name":"Speedup","score":0.8657510280609131},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8202028274536133},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.6493039131164551},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.6396545171737671},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.5778003931045532},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5305016040802002},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.48212385177612305},{"id":"https://openalex.org/keywords/solver","display_name":"Solver","score":0.47605177760124207},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.4305270314216614},{"id":"https://openalex.org/keywords/design-space-exploration","display_name":"Design space exploration","score":0.42948243021965027},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.285752534866333},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.11447611451148987}],"concepts":[{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.8657510280609131},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8202028274536133},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.6493039131164551},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6396545171737671},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.5778003931045532},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5305016040802002},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.48212385177612305},{"id":"https://openalex.org/C2778770139","wikidata":"https://www.wikidata.org/wiki/Q1966904","display_name":"Solver","level":2,"score":0.47605177760124207},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.4305270314216614},{"id":"https://openalex.org/C2776221188","wikidata":"https://www.wikidata.org/wiki/Q21072556","display_name":"Design space exploration","level":2,"score":0.42948243021965027},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.285752534866333},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.11447611451148987},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.0},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/iccad57390.2023.10323944","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccad57390.2023.10323944","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/ACM International Conference on Computer Aided Design (ICCAD)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.7400000095367432}],"awards":[{"id":"https://openalex.org/G155340880","display_name":null,"funder_award_id":"202201011388","funder_id":"https://openalex.org/F4320334898","funder_display_name":"Guangzhou Science and Technology Program key projects"},{"id":"https://openalex.org/G4037675443","display_name":null,"funder_award_id":"2021A1515011836","funder_id":"https://openalex.org/F4320321921","funder_display_name":"Natural Science Foundation of Guangdong Province"},{"id":"https://openalex.org/G4536363530","display_name":null,"funder_award_id":"2022ZD0115304","funder_id":"https://openalex.org/F4320335777","funder_display_name":"National Key Research and Development Program of China"},{"id":"https://openalex.org/G5292652112","display_name":null,"funder_award_id":"62072479","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"},{"id":"https://openalex.org/F4320321921","display_name":"Natural Science Foundation of Guangdong Province","ror":null},{"id":"https://openalex.org/F4320334898","display_name":"Guangzhou Science and Technology Program key projects","ror":"https://ror.org/00tjzgn92"},{"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":47,"referenced_works":["https://openalex.org/W1558370006","https://openalex.org/W2155027007","https://openalex.org/W2606722458","https://openalex.org/W2781726626","https://openalex.org/W2804032941","https://openalex.org/W2804500013","https://openalex.org/W2896457183","https://openalex.org/W2908064123","https://openalex.org/W2940862705","https://openalex.org/W2963341956","https://openalex.org/W2964043796","https://openalex.org/W2980104813","https://openalex.org/W2980200167","https://openalex.org/W3012249773","https://openalex.org/W3017024317","https://openalex.org/W3096395190","https://openalex.org/W3102790199","https://openalex.org/W3135807226","https://openalex.org/W3153174751","https://openalex.org/W3169291081","https://openalex.org/W3177136099","https://openalex.org/W3177452048","https://openalex.org/W3192336523","https://openalex.org/W3203303453","https://openalex.org/W3205610680","https://openalex.org/W3210432446","https://openalex.org/W4221160486","https://openalex.org/W4224308101","https://openalex.org/W4226151089","https://openalex.org/W4281768277","https://openalex.org/W4292779060","https://openalex.org/W4318541553","https://openalex.org/W4327930463","https://openalex.org/W4385245566","https://openalex.org/W6683204974","https://openalex.org/W6692846177","https://openalex.org/W6739901393","https://openalex.org/W6747473740","https://openalex.org/W6751349269","https://openalex.org/W6752057402","https://openalex.org/W6757184387","https://openalex.org/W6778883912","https://openalex.org/W6779728309","https://openalex.org/W6796289742","https://openalex.org/W6797349370","https://openalex.org/W6810137820","https://openalex.org/W6810745647"],"related_works":["https://openalex.org/W2168940796","https://openalex.org/W4282568311","https://openalex.org/W2951473296","https://openalex.org/W2883928845","https://openalex.org/W4288420200","https://openalex.org/W3145095675","https://openalex.org/W2024329643","https://openalex.org/W4365793791","https://openalex.org/W3134072570","https://openalex.org/W2171195786"],"abstract_inverted_index":{"Large":[0],"language":[1,103],"model":[2],"(LLM)":[3],"is":[4,73,79,112],"a":[5,120,128,141,147],"promising":[6],"trend":[7],"to":[8,34,66,118,132,189],"sustain":[9],"accuracy":[10],"growth":[11],"with":[12,122,150,170,178],"billions":[13],"of":[14,40,136],"parameters":[15],"for":[16,45,101,156],"various":[17],"application":[18],"domains.":[19],"Deep":[20],"learning":[21,107,115],"(DL)":[22],"accelerators":[23,44],"that":[24],"are":[25,31,51],"typically":[26],"designed":[27],"in":[28],"spatial":[29],"architecture":[30],"potential":[32],"platforms":[33],"handle":[35],"the":[36,75,113,190],"substantial":[37],"computational":[38],"demands":[39],"LLMs.":[41],"Exploiting":[42],"DL":[43],"LLMs":[46],"require":[47],"high-performance":[48],"kernels,":[49,84],"which":[50],"manually":[52],"optimized":[53],"or":[54,69],"automatically":[55],"generated.":[56],"While":[57],"prior":[58],"automatic":[59,97],"works":[60],"reduce":[61],"development":[62],"costs,":[63],"they":[64],"fail":[65],"find":[67],"optimal":[68],"near-optimal":[70],"kernels.":[71],"This":[72],"because":[74],"kernel":[76,98,121],"design":[77],"space":[78,131],"large":[80,102],"containing":[81,87],"many":[82,88],"invalid":[83],"and":[85,152],"non-convex":[86],"local":[89],"minima.":[90],"In":[91],"this":[92],"paper,":[93],"we":[94,139],"propose":[95],"an":[96],"generation":[99],"framework":[100,162,183],"models":[104],"on":[105,167,175],"deep":[106],"accelerators.":[108],"The":[109],"key":[110],"idea":[111],"reinforcement":[114],"(RL)":[116],"formulation":[117],"generate":[119],"multi-step":[123],"decision-making.":[124],"We":[125],"first":[126],"develop":[127],"high-quality":[129],"action":[130],"satisfy":[133],"architectural":[134],"constraints":[135],"accelerator.":[137],"Then,":[138],"provide":[140],"practical":[142],"RL":[143],"implementation":[144],"by":[145],"devising":[146],"policy":[148],"network":[149],"Transformer":[151],"variance":[153],"reduction":[154],"techniques":[155],"gradients.":[157],"Experimental":[158],"results":[159],"show":[160],"our":[161,182],"achieves":[163,184],"average":[164],"3.5\u00d7":[165],"speedup":[166,174],"TensorCore":[168],"compared":[169,177,188],"exploration-based":[171],"Ansor;":[172],"2.6\u00d7":[173],"Simba":[176],"solver-based":[179],"CoSA.":[180],"Also,":[181],"better":[185],"energy":[186],"efficiency":[187],"state-of-the-art":[191],"works.":[192]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":1}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}