{"id":"https://openalex.org/W4403577798","doi":"https://doi.org/10.1145/3627673.3680054","title":"A Real-Time Adaptive Multi-Stream GPU System For Online Approximate Nearest Neighborhood Search","display_name":"A Real-Time Adaptive Multi-Stream GPU System For Online Approximate Nearest Neighborhood Search","publication_year":2024,"publication_date":"2024-10-20","ids":{"openalex":"https://openalex.org/W4403577798","doi":"https://doi.org/10.1145/3627673.3680054"},"language":"en","primary_location":{"id":"doi:10.1145/3627673.3680054","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3627673.3680054","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Information and Knowledge Management","raw_type":"proceedings-article"},"type":"conference-paper","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/A5113444955","display_name":"Yiping Sun","orcid":"https://orcid.org/0009-0000-9559-8268"},"institutions":[{"id":"https://openalex.org/I862669128","display_name":"Xiaomi (China)","ror":"https://ror.org/029f7bn57","country_code":"CN","type":"company","lineage":["https://openalex.org/I862669128"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yiping Sun","raw_affiliation_strings":["Xiaohongshu Inc, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0000-9559-8268","affiliations":[{"raw_affiliation_string":"Xiaohongshu Inc, Shanghai, China","institution_ids":["https://openalex.org/I862669128"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102195711","display_name":"Yang Shi","orcid":null},"institutions":[{"id":"https://openalex.org/I862669128","display_name":"Xiaomi (China)","ror":"https://ror.org/029f7bn57","country_code":"CN","type":"company","lineage":["https://openalex.org/I862669128"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yang Shi","raw_affiliation_strings":["Xiaohongshu Inc, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0009-2042-5405","affiliations":[{"raw_affiliation_string":"Xiaohongshu Inc, Shanghai, China","institution_ids":["https://openalex.org/I862669128"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5111256217","display_name":"Jiaolong Du","orcid":null},"institutions":[{"id":"https://openalex.org/I862669128","display_name":"Xiaomi (China)","ror":"https://ror.org/029f7bn57","country_code":"CN","type":"company","lineage":["https://openalex.org/I862669128"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiaolong Du","raw_affiliation_strings":["Xiaohongshu Inc, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0006-7993-4755","affiliations":[{"raw_affiliation_string":"Xiaohongshu Inc, Shanghai, China","institution_ids":["https://openalex.org/I862669128"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I862669128"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":7,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"4906","last_page":"4913"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.9998000264167786,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.9998000264167786,"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/T10191","display_name":"Robotics and Sensor-Based Localization","score":0.9915000200271606,"subfield":{"id":"https://openalex.org/subfields/2202","display_name":"Aerospace Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10331","display_name":"Video Surveillance and Tracking Methods","score":0.9854000210762024,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7983837723731995},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.42069652676582336},{"id":"https://openalex.org/keywords/real-time-computing","display_name":"Real-time computing","score":0.34544312953948975},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.2584019899368286}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7983837723731995},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.42069652676582336},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.34544312953948975},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.2584019899368286}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3627673.3680054","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3627673.3680054","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Information and Knowledge Management","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":17,"referenced_works":["https://openalex.org/W2039820867","https://openalex.org/W2099253838","https://openalex.org/W2124509324","https://openalex.org/W2140348470","https://openalex.org/W2165489143","https://openalex.org/W2323693848","https://openalex.org/W2952307259","https://openalex.org/W2963469388","https://openalex.org/W2998702515","https://openalex.org/W3028864969","https://openalex.org/W3092853178","https://openalex.org/W3131109761","https://openalex.org/W3196481040","https://openalex.org/W4306317504","https://openalex.org/W4384697339","https://openalex.org/W4386798157","https://openalex.org/W4387321071"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W4399354997","https://openalex.org/W3062287","https://openalex.org/W2380390332","https://openalex.org/W2742145873","https://openalex.org/W2023572661","https://openalex.org/W4245975140","https://openalex.org/W2532592438"],"abstract_inverted_index":{"In":[0,95],"recent":[1],"years,":[2],"Approximate":[3],"Nearest":[4],"Neighbor":[5],"Search":[6],"(ANNS)":[7],"has":[8,235,254],"played":[9],"a":[10,29,100,149,153,176,190,196],"pivotal":[11],"role":[12],"in":[13,20,126,172,239],"modern":[14],"search":[15,242],"and":[16,131,138,214,243,253],"recommendation":[17,244],"systems,":[18,245],"especially":[19],"emerging":[21],"LLM":[22],"applications":[23,61],"like":[24],"Retrieval-Augmented":[25],"Generation.":[26],"There":[27],"is":[28],"growing":[30],"exploration":[31],"into":[32],"harnessing":[33],"the":[34,42,56,122],"parallel":[35,178],"computing":[36],"capabilities":[37],"of":[38,45,59,66,248,250],"GPUs":[39],"to":[40,82,88,203,230],"meet":[41],"substantial":[43],"demands":[44],"ANNS.":[46],"However,":[47],"existing":[48,127,184],"systems":[49,73,130,185],"primarily":[50],"focus":[51],"on":[52,91,135,146,159],"offline":[53],"scenarios,":[54],"overlooking":[55],"distinct":[57],"requirements":[58],"online":[60],"that":[62,186],"necessitate":[63],"real-time":[64,85,123,144,169],"insertion":[65,86,124,156,171],"new":[67],"vectors.":[68],"This":[69],"limitation":[70],"renders":[71],"such":[72],"inefficient":[74],"for":[75],"real-world":[76,240],"scenarios.":[77],"Moreover,":[78],"previous":[79],"architectures":[80],"struggled":[81],"effectively":[83,218],"support":[84],"due":[87],"their":[89,133],"reliance":[90,134],"serial":[92],"execution":[93,179,208],"streams.":[94],"this":[96],"paper,":[97],"we":[98,151,174],"introduce":[99,152,175],"novel":[101],"Real-Time":[102],"Adaptive":[103],"Multi-Stream":[104],"GPU":[105,128],"ANNS":[106,129],"System":[107],"(RTAMS-GANNS).":[108],"Our":[109,193],"architecture":[110],"achieves":[111],"its":[112],"objectives":[113],"through":[114],"three":[115],"key":[116],"advancements:":[117],"1)":[118],"We":[119],"initially":[120],"examined":[121],"mechanisms":[125],"discovered":[132],"repetitive":[136],"copying":[137],"memory":[139,160],"allocation,":[140],"which":[141,162,181],"significantly":[142],"hinders":[143],"effectiveness":[145],"GPUs.":[147],"As":[148],"solution,":[150],"dynamic":[154,197],"vector":[155,170],"algorithm":[157],"based":[158],"blocks,":[161],"includes":[163],"in-place":[164],"rearrangement.":[165],"2)":[166],"To":[167],"enable":[168],"parallel,":[173],"multi-stream":[177],"mode,":[180],"differs":[182],"from":[183],"operate":[187],"serially":[188],"within":[189],"single":[191],"stream.":[192],"system":[194,234],"utilizes":[195],"resource":[198],"pool,":[199],"allowing":[200],"multiple":[201],"streams":[202],"execute":[204],"concurrently":[205],"without":[206],"additional":[207],"blocking.":[209],"3)":[210],"Through":[211],"extensive":[212],"experiments":[213],"comparisons,":[215],"our":[216],"approach":[217],"handles":[219],"varying":[220],"QPS":[221],"levels":[222],"across":[223],"different":[224],"datasets,":[225],"reducing":[226],"latency":[227],"by":[228],"up":[229],"40%-80%.":[231],"The":[232],"proposed":[233],"also":[236],"been":[237],"deployed":[238],"industrial":[241],"serving":[246],"hundreds":[247],"millions":[249],"users":[251],"daily,":[252],"achieved":[255],"significant":[256],"results.":[257]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":6}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}