{"id":"https://openalex.org/W4409475001","doi":"https://doi.org/10.1109/lca.2025.3560786","title":"MixDiT: Accelerating Image Diffusion Transformer Inference With Mixed-Precision MX Quantization","display_name":"MixDiT: Accelerating Image Diffusion Transformer Inference With Mixed-Precision MX Quantization","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4409475001","doi":"https://doi.org/10.1109/lca.2025.3560786"},"language":"en","primary_location":{"id":"doi:10.1109/lca.2025.3560786","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2025.3560786","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"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 Computer Architecture Letters","raw_type":"journal-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":null,"display_name":"Daeun Kim","orcid":"https://orcid.org/0009-0006-1075-211X"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Daeun Kim","raw_affiliation_strings":["School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0009-0006-1075-211X","affiliations":[{"raw_affiliation_string":"School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101320389","display_name":"Jinwoo Hwang","orcid":null},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jinwoo Hwang","raw_affiliation_strings":["School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Changhun Oh","orcid":"https://orcid.org/0009-0009-9834-3848"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Changhun Oh","raw_affiliation_strings":["School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0009-0009-9834-3848","affiliations":[{"raw_affiliation_string":"School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5037553165","display_name":"Jongse Park","orcid":"https://orcid.org/0000-0002-6629-449X"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jongse Park","raw_affiliation_strings":["School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea"],"raw_orcid":"https://orcid.org/0000-0002-6629-449X","affiliations":[{"raw_affiliation_string":"School of Computing, Graduate School of AI Semiconductor, Korea Advanced Institute of Science and Technology, Daejeon, South Korea","institution_ids":["https://openalex.org/I157485424"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157485424"],"apc_list":null,"apc_paid":null,"fwci":1.059,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.7430932,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":97,"max":98},"biblio":{"volume":"24","issue":"1","first_page":"141","last_page":"144"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10688","display_name":"Image and Signal Denoising Methods","score":0.9779000282287598,"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/T10688","display_name":"Image and Signal Denoising Methods","score":0.9779000282287598,"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/T13114","display_name":"Image Processing Techniques and Applications","score":0.9726999998092651,"subfield":{"id":"https://openalex.org/subfields/2214","display_name":"Media Technology"},"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/T10052","display_name":"Medical Image Segmentation Techniques","score":0.9348999857902527,"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":0.6707535982131958},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6232129335403442},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6200273036956787},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.5490128397941589},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4286920726299286},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.36974167823791504},{"id":"https://openalex.org/keywords/computer-vision","display_name":"Computer vision","score":0.3656069040298462},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.24755015969276428},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.11689284443855286},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.09471717476844788}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6707535982131958},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6232129335403442},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6200273036956787},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.5490128397941589},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4286920726299286},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.36974167823791504},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.3656069040298462},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.24755015969276428},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.11689284443855286},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.09471717476844788}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/lca.2025.3560786","is_oa":false,"landing_page_url":"https://doi.org/10.1109/lca.2025.3560786","pdf_url":null,"source":{"id":"https://openalex.org/S17643076","display_name":"IEEE Computer Architecture Letters","issn_l":"1556-6056","issn":["1556-6056","1556-6064","2473-2575"],"is_oa":false,"is_in_doaj":false,"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 Computer Architecture Letters","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":10,"referenced_works":["https://openalex.org/W4380874652","https://openalex.org/W4390872297","https://openalex.org/W4401211881","https://openalex.org/W4404625217","https://openalex.org/W4405003117","https://openalex.org/W4407953477","https://openalex.org/W6862995742","https://openalex.org/W6868797372","https://openalex.org/W6869514076","https://openalex.org/W6870184360"],"related_works":["https://openalex.org/W2051487156","https://openalex.org/W2073681303","https://openalex.org/W2053286651","https://openalex.org/W2055243143","https://openalex.org/W2181743346","https://openalex.org/W2187401768","https://openalex.org/W4206178588","https://openalex.org/W3094491777","https://openalex.org/W3214715529","https://openalex.org/W4287635093"],"abstract_inverted_index":{"DiffusionTransformer":[0],"(DiT)":[1],"has":[2,47],"driven":[3],"significant":[4],"progress":[5],"in":[6,146],"image":[7],"generation":[8],"tasks.":[9],"However,":[10],"DiT":[11,55,88,93],"inferencing":[12,56],"is":[13],"notoriously":[14],"compute-intensive":[15],"and":[16,30,61,125],"incurs":[17],"long":[18],"latency":[19],"even":[20],"on":[21,33],"datacenter-scale":[22],"GPUs,":[23],"primarily":[24],"due":[25],"to":[26,37,86,101],"its":[27,38],"iterative":[28],"nature":[29],"heavy":[31],"reliance":[32],"GEMM":[34,107],"operations":[35],"inherent":[36],"encoder-based":[39],"structure.":[40],"To":[41,68,109],"address":[42],"the":[43,92,114],"challenge,":[44],"prior":[45],"work":[46],"explored":[48],"quantization,":[49],"but":[50],"achieving":[51],"low-precision":[52],"quantization":[53],"for":[54],"with":[57,143],"both":[58],"high":[59],"accuracy":[60],"substantial":[62],"speedup":[63,112,137],"remains":[64],"an":[65,75],"open":[66],"problem.":[67],"this":[69,71],"end,":[70],"paper":[72],"proposes":[73],"MixDiT,":[74],"algorithm-hardware":[76],"co-designed":[77],"acceleration":[78],"solution":[79],"that":[80,121,134],"exploits":[81],"mixed":[82],"Microscaling":[83],"(MX)":[84],"formats":[85],"quantize":[87],"activation":[89,94],"values.":[90],"MixDiTquantizes":[91],"tensors":[95],"by":[96],"selectively":[97],"applying":[98],"higher":[99],"precision":[100,128],"magnitude-based":[102],"outliers,":[103],"which":[104],"produce":[105],"mixed-precision":[106,115],"operations.":[108],"achieve":[110],"tangible":[111],"from":[113],"arithmetic,":[116],"we":[117],"design":[118],"a":[119,136],"MixDiTaccelerator":[120],"enables":[122],"precision-flexible":[123],"multiplications":[124],"efficient":[126],"MX":[127],"conversions.":[129],"Our":[130],"experimental":[131],"results":[132],"show":[133],"MixDiTdelivers":[135],"of":[138],"2.10\u20135.32\u00d7":[139],"over":[140],"RTX":[141],"3090,":[142],"no":[144],"loss":[145],"FID.":[147]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2025-10-10T00:00:00"}