{"id":"https://openalex.org/W7116430641","doi":"https://doi.org/10.1145/3750720.3757289","title":"Optimizing Large Language Model Deployment with TVM: Overcoming Dynamic Shape and Scheduling Challenges","display_name":"Optimizing Large Language Model Deployment with TVM: Overcoming Dynamic Shape and Scheduling Challenges","publication_year":2025,"publication_date":"2025-09-08","ids":{"openalex":"https://openalex.org/W7116430641","doi":"https://doi.org/10.1145/3750720.3757289"},"language":null,"primary_location":{"id":"doi:10.1145/3750720.3757289","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3750720.3757289","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Workshop Proceedings of the 54th International Conference on Parallel Processing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3750720.3757289","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5120923187","display_name":"Kang Wei Kuo","orcid":null},"institutions":[{"id":"https://openalex.org/I148099254","display_name":"National Chung Cheng University","ror":"https://ror.org/0028v3876","country_code":"TW","type":"education","lineage":["https://openalex.org/I148099254"]}],"countries":["TW"],"is_corresponding":true,"raw_author_name":"Kang Wei Kuo","raw_affiliation_strings":["Computer Science and Information Engineering, National Chung Cheng University, ChiaYi, Taiwan"],"raw_orcid":"https://orcid.org/0009-0005-5697-1396","affiliations":[{"raw_affiliation_string":"Computer Science and Information Engineering, National Chung Cheng University, ChiaYi, Taiwan","institution_ids":["https://openalex.org/I148099254"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5049069830","display_name":"P. S. Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I148099254","display_name":"National Chung Cheng University","ror":"https://ror.org/0028v3876","country_code":"TW","type":"education","lineage":["https://openalex.org/I148099254"]}],"countries":["TW"],"is_corresponding":false,"raw_author_name":"Peng-Sheng Chen","raw_affiliation_strings":["Computer Science and Information Engineering, National Chung Cheng University, ChiaYi, Taiwan"],"raw_orcid":"https://orcid.org/0000-0003-3875-4150","affiliations":[{"raw_affiliation_string":"Computer Science and Information Engineering, National Chung Cheng University, ChiaYi, Taiwan","institution_ids":["https://openalex.org/I148099254"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5120923187"],"corresponding_institution_ids":["https://openalex.org/I148099254"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.64760925,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"80","last_page":"89"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.7523000240325928,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.7523000240325928,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T11948","display_name":"Machine Learning in Materials Science","score":0.039500001817941666,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10126","display_name":"Logic, programming, and type systems","score":0.031599998474121094,"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/scheduling","display_name":"Scheduling (production processes)","score":0.6406999826431274},{"id":"https://openalex.org/keywords/compiler","display_name":"Compiler","score":0.5756000280380249},{"id":"https://openalex.org/keywords/software-deployment","display_name":"Software deployment","score":0.5634999871253967},{"id":"https://openalex.org/keywords/compatibility","display_name":"Compatibility (geochemistry)","score":0.5604000091552734},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.3898000121116638},{"id":"https://openalex.org/keywords/shape-analysis","display_name":"Shape analysis (program analysis)","score":0.3269999921321869},{"id":"https://openalex.org/keywords/shape-optimization","display_name":"Shape optimization","score":0.32670000195503235}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7484999895095825},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.6406999826431274},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.5756000280380249},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.5634999871253967},{"id":"https://openalex.org/C2778648169","wikidata":"https://www.wikidata.org/wiki/Q967768","display_name":"Compatibility (geochemistry)","level":2,"score":0.5604000091552734},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.45980000495910645},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.3898000121116638},{"id":"https://openalex.org/C112604564","wikidata":"https://www.wikidata.org/wiki/Q7489226","display_name":"Shape analysis (program analysis)","level":3,"score":0.3269999921321869},{"id":"https://openalex.org/C29513896","wikidata":"https://www.wikidata.org/wiki/Q7489239","display_name":"Shape optimization","level":3,"score":0.32670000195503235},{"id":"https://openalex.org/C50897621","wikidata":"https://www.wikidata.org/wiki/Q2665508","display_name":"Hybrid system","level":2,"score":0.3149000108242035},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.3100999891757965},{"id":"https://openalex.org/C107568181","wikidata":"https://www.wikidata.org/wiki/Q5319000","display_name":"Dynamic priority scheduling","level":3,"score":0.3100999891757965},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.30090001225471497},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.2849000096321106},{"id":"https://openalex.org/C179603123","wikidata":"https://www.wikidata.org/wiki/Q1941921","display_name":"Modeling language","level":3,"score":0.2840999960899353},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2827000021934509},{"id":"https://openalex.org/C2984822820","wikidata":"https://www.wikidata.org/wiki/Q1123036","display_name":"Processor scheduling","level":3,"score":0.26429998874664307},{"id":"https://openalex.org/C175309249","wikidata":"https://www.wikidata.org/wiki/Q725864","display_name":"Pipeline transport","level":2,"score":0.25519999861717224}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3750720.3757289","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3750720.3757289","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Workshop Proceedings of the 54th International Conference on Parallel Processing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3750720.3757289","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3750720.3757289","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Workshop Proceedings of the 54th International Conference on Parallel Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":5,"referenced_works":["https://openalex.org/W2005838647","https://openalex.org/W3161395920","https://openalex.org/W4381785750","https://openalex.org/W4394998528","https://openalex.org/W4408903581"],"related_works":[],"abstract_inverted_index":{"Dynamic":[0],"shape":[1,18,44,66,79,88,126,139],"models,":[2,106],"such":[3],"as":[4],"Large":[5],"Language":[6],"Models":[7],"(LLMs),":[8],"pose":[9],"significant":[10],"challenges":[11],"for":[12,42,124,134],"compiler":[13],"optimization":[14],"due":[15],"to":[16,85,137],"runtime":[17],"variability":[19],"and":[20,51,64,99,111],"limited":[21],"scheduling":[22],"support.":[23],"TVM":[24,83,142],"provides":[25],"an":[26],"end-to-end":[27],"compilation":[28],"pipeline":[29],"with":[30,146],"auto-tuning":[31],"via":[32],"its":[33,36],"MetaSchedule.":[34],"However,":[35],"current":[37],"design":[38],"is":[39],"primarily":[40],"optimized":[41],"static":[43],"models.":[45,89],"This":[46],"work":[47,129],"addresses":[48],"these":[49],"limitations":[50],"proposes":[52],"practical":[53,132],"solutions.":[54],"We":[55],"introduce":[56],"three":[57],"methods:":[58],"shape-aware":[59],"prioritized":[60],"scheduling,":[61,63],"hybrid":[62],"compile-time":[65,78],"binding":[67,80],"-":[68],"all":[69],"of":[70],"which":[71],"enable":[72],"symbolic":[73],"shape\u2013aware":[74],"tuning.":[75],"In":[76],"particular,":[77],"extends":[81],"the":[82,104,113],"MetaSchedule":[84],"support":[86],"dynamic":[87,125,138],"The":[90],"proposed":[91],"approaches":[92],"are":[93],"evaluated":[94],"on":[95],"both":[96],"desktop":[97],"PCs":[98],"resource-constrained":[100],"embedded":[101],"systems.":[102],"For":[103],"tested":[105],"including":[107],"Whisper-tiny,":[108],"GPT-2,":[109],"Llama-3.2-1B,":[110],"Qwen2.5-0.5B,":[112],"experimental":[114],"results":[115],"demonstrate":[116],"that":[117],"our":[118],"methods":[119],"significantly":[120],"improve":[121],"execution":[122],"performance":[123],"workloads.":[127],"Our":[128],"introduces":[130],"a":[131],"solution":[133],"applying":[135],"auto-scheduling":[136],"models":[140],"in":[141],"while":[143],"maintaining":[144],"compatibility":[145],"existing":[147],"optimizations.":[148]},"counts_by_year":[],"updated_date":"2025-12-21T02:06:08.432651","created_date":"2025-12-21T00:00:00"}