{"id":"https://openalex.org/W4206414916","doi":"https://doi.org/10.1109/access.2021.3133100","title":"Performance Evaluation of INT8 Quantized Inference on Mobile GPUs","display_name":"Performance Evaluation of INT8 Quantized Inference on Mobile GPUs","publication_year":2021,"publication_date":"2021-01-01","ids":{"openalex":"https://openalex.org/W4206414916","doi":"https://doi.org/10.1109/access.2021.3133100"},"language":"en","primary_location":{"id":"doi:10.1109/access.2021.3133100","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2021.3133100","pdf_url":"https://ieeexplore.ieee.org/ielx7/6287639/6514899/09638444.pdf","source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://ieeexplore.ieee.org/ielx7/6287639/6514899/09638444.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5046043627","display_name":"Sumin Kim","orcid":"https://orcid.org/0000-0001-7747-2143"},"institutions":[{"id":"https://openalex.org/I124633538","display_name":"University of Seoul","ror":"https://ror.org/05en5nh73","country_code":"KR","type":"education","lineage":["https://openalex.org/I124633538"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Sumin Kim","raw_affiliation_strings":["Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea"],"raw_orcid":"https://orcid.org/0000-0001-7747-2143","affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea","institution_ids":["https://openalex.org/I124633538"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076301234","display_name":"Gunju Park","orcid":"https://orcid.org/0000-0002-6734-8648"},"institutions":[{"id":"https://openalex.org/I124633538","display_name":"University of Seoul","ror":"https://ror.org/05en5nh73","country_code":"KR","type":"education","lineage":["https://openalex.org/I124633538"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Gunju Park","raw_affiliation_strings":["Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea"],"raw_orcid":"https://orcid.org/0000-0002-6734-8648","affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea","institution_ids":["https://openalex.org/I124633538"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5029634385","display_name":"Youngmin Yi","orcid":"https://orcid.org/0000-0001-9802-2109"},"institutions":[{"id":"https://openalex.org/I124633538","display_name":"University of Seoul","ror":"https://ror.org/05en5nh73","country_code":"KR","type":"education","lineage":["https://openalex.org/I124633538"]}],"countries":["KR"],"is_corresponding":true,"raw_author_name":"Youngmin Yi","raw_affiliation_strings":["Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea. (e-mail: ymyi@uos.ac.kr)","Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea"],"raw_orcid":"https://orcid.org/0000-0001-9802-2109","affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea. (e-mail: ymyi@uos.ac.kr)","institution_ids":["https://openalex.org/I124633538"]},{"raw_affiliation_string":"Department of Electrical and Computer Engineering, University of Seoul, Seoulsiripdae-ro 163, Dongdaemoon-gu, Seoul, Korea","institution_ids":["https://openalex.org/I124633538"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5029634385"],"corresponding_institution_ids":["https://openalex.org/I124633538"],"apc_list":{"value":1850,"currency":"USD","value_usd":1850},"apc_paid":{"value":1850,"currency":"USD","value_usd":1850},"fwci":1.7457,"has_fulltext":true,"cited_by_count":25,"citation_normalized_percentile":{"value":0.87031747,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":99},"biblio":{"volume":"9","issue":null,"first_page":"164245","last_page":"164255"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998999834060669,"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":0.9998999834060669,"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":0.9932000041007996,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9848999977111816,"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/computer-science","display_name":"Computer science","score":0.7221871614456177},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.7133471965789795},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.709254801273346},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.6378390192985535},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.6048382520675659},{"id":"https://openalex.org/keywords/convolution","display_name":"Convolution (computer science)","score":0.5460491180419922},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.49822211265563965},{"id":"https://openalex.org/keywords/normalization","display_name":"Normalization (sociology)","score":0.4444522261619568},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3694472908973694},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.29367148876190186}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7221871614456177},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.7133471965789795},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.709254801273346},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.6378390192985535},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.6048382520675659},{"id":"https://openalex.org/C45347329","wikidata":"https://www.wikidata.org/wiki/Q5166604","display_name":"Convolution (computer science)","level":3,"score":0.5460491180419922},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.49822211265563965},{"id":"https://openalex.org/C136886441","wikidata":"https://www.wikidata.org/wiki/Q926129","display_name":"Normalization (sociology)","level":2,"score":0.4444522261619568},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3694472908973694},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.29367148876190186},{"id":"https://openalex.org/C144024400","wikidata":"https://www.wikidata.org/wiki/Q21201","display_name":"Sociology","level":0,"score":0.0},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.0},{"id":"https://openalex.org/C19165224","wikidata":"https://www.wikidata.org/wiki/Q23404","display_name":"Anthropology","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/access.2021.3133100","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2021.3133100","pdf_url":"https://ieeexplore.ieee.org/ielx7/6287639/6514899/09638444.pdf","source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},{"id":"pmh:oai:doaj.org/article:a9f8d8e79c5a430a98dbfac864894da4","is_oa":true,"landing_page_url":"https://doaj.org/article/a9f8d8e79c5a430a98dbfac864894da4","pdf_url":null,"source":{"id":"https://openalex.org/S4306401280","display_name":"DOAJ (DOAJ: Directory of Open Access Journals)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"IEEE Access, Vol 9, Pp 164245-164255 (2021)","raw_type":"article"}],"best_oa_location":{"id":"doi:10.1109/access.2021.3133100","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2021.3133100","pdf_url":"https://ieeexplore.ieee.org/ielx7/6287639/6514899/09638444.pdf","source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.5400000214576721}],"awards":[{"id":"https://openalex.org/G1601988010","display_name":null,"funder_award_id":"2017-0-00142","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"}],"funders":[{"id":"https://openalex.org/F4320328359","display_name":"Ministry of Science and ICT, South Korea","ror":"https://ror.org/01wpjm123"},{"id":"https://openalex.org/F4320335489","display_name":"Institute for Information and Communications Technology Promotion","ror":"https://ror.org/01g0hqq23"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4206414916.pdf","grobid_xml":"https://content.openalex.org/works/W4206414916.grobid-xml"},"referenced_works_count":42,"referenced_works":["https://openalex.org/W1836465849","https://openalex.org/W1884620995","https://openalex.org/W2149927643","https://openalex.org/W2172654076","https://openalex.org/W2626129225","https://openalex.org/W2741623780","https://openalex.org/W2766839578","https://openalex.org/W2809624076","https://openalex.org/W2885010062","https://openalex.org/W2886851211","https://openalex.org/W2889647155","https://openalex.org/W2904079411","https://openalex.org/W2904795057","https://openalex.org/W2907172645","https://openalex.org/W2922395136","https://openalex.org/W2931743911","https://openalex.org/W2955715866","https://openalex.org/W2963122961","https://openalex.org/W2963163009","https://openalex.org/W2971002981","https://openalex.org/W2998506323","https://openalex.org/W3015869703","https://openalex.org/W3097891912","https://openalex.org/W3101257755","https://openalex.org/W3105888187","https://openalex.org/W4236853429","https://openalex.org/W4237812669","https://openalex.org/W4295312788","https://openalex.org/W4297775537","https://openalex.org/W6638667902","https://openalex.org/W6737664043","https://openalex.org/W6741779227","https://openalex.org/W6745499037","https://openalex.org/W6752630882","https://openalex.org/W6753069482","https://openalex.org/W6753585227","https://openalex.org/W6754777574","https://openalex.org/W6757703747","https://openalex.org/W6758002897","https://openalex.org/W6765438313","https://openalex.org/W6766978945","https://openalex.org/W6784825895"],"related_works":["https://openalex.org/W2058965144","https://openalex.org/W2164382479","https://openalex.org/W2146343568","https://openalex.org/W98480971","https://openalex.org/W2150291671","https://openalex.org/W2013643406","https://openalex.org/W2027972911","https://openalex.org/W2157978810","https://openalex.org/W2097707447","https://openalex.org/W3157543420"],"abstract_inverted_index":{"During":[0],"the":[1,5,15,88,121,181],"past":[2],"several":[3],"years,":[4],"need":[6],"for":[7,27,51,129,166],"on-device":[8,29],"deep":[9,30],"learning":[10],"has":[11,20,35],"rapidly":[12],"increased,":[13],"and":[14,39,71,73,79,90,97,112,132,138,146,171,179],"performance":[16,89,122],"of":[17,92,123,195],"mobile":[18,52,83],"GPUs":[19,140],"significantly":[21],"increased.":[22],"As":[23],"a":[24,57],"viable":[25],"approach":[26],"efficient":[28],"learning,":[31],"INT8":[32,49,63,93,99,124,127,151,183],"quantized":[33,94,109,116,152],"inference":[34,153],"been":[36],"actively":[37],"studied":[38],"proposed":[40,182],"but":[41],"there":[42],"are":[43,108,115],"currently":[44],"few":[45],"frameworks":[46],"that":[47,60],"support":[48],"quantization":[50,64],"GPUs.":[53,84],"This":[54],"paper":[55],"presents":[56],"unified":[58],"framework":[59],"integrates":[61],"various":[62],"methods,":[65,125],"such":[66],"as":[67],"symmetric,":[68],"asymmetric,":[69],"per-layer,":[70],"per-channel,":[72],"discuss":[74,87],"their":[75],"impact":[76],"on":[77,81,134,141,155,168],"accuracy":[78,91,193],"efficiency":[80],"recent":[82],"Moreover,":[85],"we":[86],"Winograd":[95,100,184],"convolution":[96,101],"propose":[98],"with":[102,191],"F(2\u00d72,":[103],"3\u00d73),":[104],"where":[105],"weight":[106],"tensors":[107,114],"in":[110,117,189],"INT4":[111],"input":[113],"INT6.":[118],"We":[119],"evaluated":[120],"including":[126],"Winograd,":[128],"ResNet50,":[130],"MobileNet-v1,":[131],"VGG16":[133],"Mali":[135,169],"G52,":[136,170],"G72,":[137],"G76":[139],"Odroid":[142],"N2,":[143],"Galaxy":[144,147],"S9,":[145],"Note":[148],"10+,":[149],"respectively.":[150],"based":[154],"General":[156],"Matrix":[157],"Multiplication":[158],"(GEMM)":[159],"was":[160,172],"1.67\u00d7":[161],"faster":[162],"than":[163],"FP32":[164],"GEMM":[165],"ResNet50":[167],"further":[173],"accelerated":[174],"by":[175,180],"batch":[176],"normalization":[177],"folding":[178],"convolution,":[185],"achieving":[186],"2.45\u00d7":[187],"speedup":[188],"total":[190],"an":[192],"loss":[194],"only":[196],"-0.31%p.":[197]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":8},{"year":2023,"cited_by_count":5},{"year":2022,"cited_by_count":5}],"updated_date":"2026-05-06T08:25:59.206177","created_date":"2025-10-10T00:00:00"}