{"id":"https://openalex.org/W4214668489","doi":"https://doi.org/10.1145/3503222.3507754","title":"GPUReplay: a 50-KB GPU stack for client ML","display_name":"GPUReplay: a 50-KB GPU stack for client ML","publication_year":2022,"publication_date":"2022-02-22","ids":{"openalex":"https://openalex.org/W4214668489","doi":"https://doi.org/10.1145/3503222.3507754"},"language":"en","primary_location":{"id":"doi:10.1145/3503222.3507754","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3503222.3507754","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3503222.3507754","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3503222.3507754","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5025182744","display_name":"Heejin Park","orcid":"https://orcid.org/0000-0001-8142-7505"},"institutions":[{"id":"https://openalex.org/I219193219","display_name":"Purdue University West Lafayette","ror":"https://ror.org/02dqehb95","country_code":"US","type":"education","lineage":["https://openalex.org/I219193219"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Heejin Park","raw_affiliation_strings":["Purdue University, USA"],"raw_orcid":"https://orcid.org/0000-0001-8142-7505","affiliations":[{"raw_affiliation_string":"Purdue University, USA","institution_ids":["https://openalex.org/I219193219"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5025585492","display_name":"Felix Xiaozhu Lin","orcid":"https://orcid.org/0000-0002-1615-6419"},"institutions":[{"id":"https://openalex.org/I51556381","display_name":"University of Virginia","ror":"https://ror.org/0153tk833","country_code":"US","type":"education","lineage":["https://openalex.org/I51556381"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Felix Xiaozhu Lin","raw_affiliation_strings":["University of Virginia, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Virginia, USA","institution_ids":["https://openalex.org/I51556381"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.647,"has_fulltext":true,"cited_by_count":9,"citation_normalized_percentile":{"value":0.77423212,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"157","last_page":"170"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9997000098228455,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9997000098228455,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9987999796867371,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9957000017166138,"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":0.8723998069763184},{"id":"https://openalex.org/keywords/executable","display_name":"Executable","score":0.6822671890258789},{"id":"https://openalex.org/keywords/stack","display_name":"Stack (abstract data type)","score":0.5523562431335449},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.5503302812576294},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4563676714897156},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.44252967834472656},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.4389810562133789},{"id":"https://openalex.org/keywords/compiler","display_name":"Compiler","score":0.41386184096336365},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.41352778673171997},{"id":"https://openalex.org/keywords/general-purpose-computing-on-graphics-processing-units","display_name":"General-purpose computing on graphics processing units","score":0.41134893894195557},{"id":"https://openalex.org/keywords/embedded-system","display_name":"Embedded system","score":0.3687337636947632},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.23329293727874756},{"id":"https://openalex.org/keywords/graphics","display_name":"Graphics","score":0.14790648221969604}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8723998069763184},{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.6822671890258789},{"id":"https://openalex.org/C9395851","wikidata":"https://www.wikidata.org/wiki/Q177929","display_name":"Stack (abstract data type)","level":2,"score":0.5523562431335449},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.5503302812576294},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4563676714897156},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.44252967834472656},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.4389810562133789},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.41386184096336365},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.41352778673171997},{"id":"https://openalex.org/C50630238","wikidata":"https://www.wikidata.org/wiki/Q971505","display_name":"General-purpose computing on graphics processing units","level":3,"score":0.41134893894195557},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3687337636947632},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.23329293727874756},{"id":"https://openalex.org/C21442007","wikidata":"https://www.wikidata.org/wiki/Q1027879","display_name":"Graphics","level":2,"score":0.14790648221969604},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3503222.3507754","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3503222.3507754","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3503222.3507754","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3503222.3507754","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3503222.3507754","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3503222.3507754","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1852190989","display_name":"Collaborative Research: CNS Core: Medium: Understanding and Strengthening Memory Security for Non-Volatile Memory","funder_award_id":"2106893","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G5952027803","display_name":"CAREER: A Trustworthy and Verifiable Software Backplane for the Cloud Edge","funder_award_id":"1846102","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8345018822","display_name":"SPX: Write Once, Run on Anything: Verified, Tuned Accelerator Kernels from High Level Specifications","funder_award_id":"1919197","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4214668489.pdf","grobid_xml":"https://content.openalex.org/works/W4214668489.grobid-xml"},"referenced_works_count":25,"referenced_works":["https://openalex.org/W1624569747","https://openalex.org/W1976792530","https://openalex.org/W2068034551","https://openalex.org/W2093832531","https://openalex.org/W2124365587","https://openalex.org/W2140839850","https://openalex.org/W2142892618","https://openalex.org/W2362396924","https://openalex.org/W2734941459","https://openalex.org/W2793790797","https://openalex.org/W2903650079","https://openalex.org/W2930957133","https://openalex.org/W2931092525","https://openalex.org/W2931743911","https://openalex.org/W2966011700","https://openalex.org/W2979832172","https://openalex.org/W3010640493","https://openalex.org/W3015584356","https://openalex.org/W3016075089","https://openalex.org/W3097981673","https://openalex.org/W3102175148","https://openalex.org/W3133174490","https://openalex.org/W4236658661","https://openalex.org/W4238623774","https://openalex.org/W4251828973"],"related_works":["https://openalex.org/W2357088637","https://openalex.org/W3192364216","https://openalex.org/W1963859303","https://openalex.org/W2364044215","https://openalex.org/W2389600408","https://openalex.org/W240129890","https://openalex.org/W3048701459","https://openalex.org/W2149078538","https://openalex.org/W2080146221","https://openalex.org/W2370314112"],"abstract_inverted_index":{"GPUReplay":[0,119],"(GR)":[1],"is":[2,31,70,80,144],"a":[3,20,71,95,122],"novel":[4],"way":[5],"for":[6,24,138],"deploying":[7],"GPU-accelerated":[8],"computation":[9],"on":[10,36,48],"mobile":[11],"and":[12,27,44,63,100,102,132],"embedded":[13],"devices.":[14],"It":[15,79],"addresses":[16],"high":[17],"complexity":[18],"of":[19,42,74,84,114,124],"modern":[21],"GPU":[22,34,39,77,126,128],"stack":[23,40],"deployment":[25],"ease":[26],"security.":[28],"The":[29,67,142],"idea":[30],"to":[32,65,104,111],"record":[33],"executions":[35,47,89],"the":[37,46,75],"full":[38],"ahead":[41],"time":[43],"replay":[45],"new":[49],"input":[50],"at":[51,146],"run":[52],"time.":[53],"We":[54,116],"address":[55],"key":[56],"challenges":[57],"towards":[58],"making":[59],"GR":[60],"feasible,":[61],"sound,":[62],"practical":[64],"use.":[66],"resultant":[68],"replayer":[69],"drop-in":[72],"replacement":[73],"original":[76],"stack.":[78],"tiny":[81],"(50":[82],"KB":[83],"executable),":[85],"robust":[86],"(replaying":[87],"long":[88],"without":[90],"divergence),":[91],"portable":[92],"(running":[93],"in":[94,98],"commodity":[96],"OS,":[97],"TEE,":[99],"baremetal),":[101],"quick":[103],"launch":[105],"(speeding":[106],"up":[107,110],"startup":[108],"by":[109],"two":[112],"orders":[113],"magnitude).":[115],"show":[117],"that":[118],"works":[120],"with":[121],"variety":[123],"integrated":[125],"hardware,":[127],"APIs,":[129],"ML":[130],"frameworks,":[131],"33":[133],"neural":[134],"network":[135],"(NN)":[136],"implementations":[137],"inference":[139],"or":[140],"training.":[141],"code":[143],"available":[145],"https://github.com/bakhi/GPUReplay.":[147]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":2},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":1}],"updated_date":"2026-08-06T08:24:18.245995","created_date":"2025-10-10T00:00:00"}