{"id":"https://openalex.org/W4417095767","doi":"https://doi.org/10.48550/arxiv.2506.04648","title":"FPSAttention: Training-Aware FP8 and Sparsity Co-Design for Fast Video Diffusion","display_name":"FPSAttention: Training-Aware FP8 and Sparsity Co-Design for Fast Video Diffusion","publication_year":2025,"publication_date":"2025-06-05","ids":{"openalex":"https://openalex.org/W4417095767","doi":"https://doi.org/10.48550/arxiv.2506.04648"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2506.04648","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.04648","pdf_url":"https://arxiv.org/pdf/2506.04648","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"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":null,"raw_type":"text"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2506.04648","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5089246231","display_name":"Akide Liu","orcid":"https://orcid.org/0000-0001-9870-8303"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Liu, Akide","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100358735","display_name":"Zeyu Zhang","orcid":"https://orcid.org/0009-0006-2834-3007"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zeyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064468010","display_name":"Z. Y. Li","orcid":"https://orcid.org/0000-0002-4652-9845"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zhexin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Bai, Xuehai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bai, Xuehai","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5045569962","display_name":"Yizeng Han","orcid":"https://orcid.org/0000-0001-5706-8784"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han, Yizeng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113292421","display_name":"Jiasheng Tang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Jiasheng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101807968","display_name":"Xing Yi","orcid":"https://orcid.org/0000-0003-3570-6946"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xing, Yuanjie","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5077973985","display_name":"Jing Wu","orcid":"https://orcid.org/0000-0001-8724-3686"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Jichao","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101785541","display_name":"Mingyang Yang","orcid":"https://orcid.org/0000-0002-1921-1217"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Mingyang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100625507","display_name":"Weihua Chen","orcid":"https://orcid.org/0000-0002-0548-330X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Weihua","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016341667","display_name":"Jiahao He","orcid":"https://orcid.org/0000-0003-3694-4989"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Jiahao","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086133186","display_name":"Yuanyu He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Yuanyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088134851","display_name":"Fan Wang","orcid":"https://orcid.org/0000-0002-2095-5256"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Fan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081525024","display_name":"Gholamreza Haffari","orcid":"https://orcid.org/0000-0001-7326-8380"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haffari, Gholamreza","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5076928390","display_name":"Bohan Zhuang","orcid":"https://orcid.org/0000-0002-0074-0303"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhuang, Bohan","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":15,"corresponding_author_ids":["https://openalex.org/A5089246231"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.5683000087738037,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.5683000087738037,"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/T11019","display_name":"Image Enhancement Techniques","score":0.07930000126361847,"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/T11165","display_name":"Image and Video Quality Assessment","score":0.07440000027418137,"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/speedup","display_name":"Speedup","score":0.8447999954223633},{"id":"https://openalex.org/keywords/granularity","display_name":"Granularity","score":0.7390000224113464},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.715499997138977},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.6642000079154968},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.5636000037193298},{"id":"https://openalex.org/keywords/noise-reduction","display_name":"Noise reduction","score":0.5092999935150146},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.40220001339912415},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.3756999969482422}],"concepts":[{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.8447999954223633},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.746399998664856},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.7390000224113464},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.715499997138977},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.6642000079154968},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.5636000037193298},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.5092999935150146},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5002999901771545},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.47519999742507935},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.40220001339912415},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3756999969482422},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3456999957561493},{"id":"https://openalex.org/C30814859","wikidata":"https://www.wikidata.org/wiki/Q4119603","display_name":"Video denoising","level":5,"score":0.3264999985694885},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.31360000371932983},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.296999990940094},{"id":"https://openalex.org/C29265498","wikidata":"https://www.wikidata.org/wiki/Q7047719","display_name":"Noise measurement","level":3,"score":0.2856999933719635},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2802000045776367},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.27639999985694885},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.2759000062942505},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.2727000117301941},{"id":"https://openalex.org/C199833920","wikidata":"https://www.wikidata.org/wiki/Q612536","display_name":"Vector quantization","level":2,"score":0.2687000036239624},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.2639999985694885},{"id":"https://openalex.org/C147297375","wikidata":"https://www.wikidata.org/wiki/Q6674930","display_name":"Look-ahead","level":2,"score":0.2630000114440918},{"id":"https://openalex.org/C122280245","wikidata":"https://www.wikidata.org/wiki/Q620622","display_name":"Kernel method","level":3,"score":0.2556999921798706},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.2513999938964844}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2506.04648","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.04648","pdf_url":"https://arxiv.org/pdf/2506.04648","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"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":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2506.04648","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.04648","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2506.04648","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.04648","pdf_url":"https://arxiv.org/pdf/2506.04648","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"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":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4417095767.pdf"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"generative":[1],"models":[2,148],"have":[3],"become":[4],"the":[5,53,77,109,113,152,174],"standard":[6],"for":[7,70,138,161,169],"producing":[8],"high-quality,":[9],"coherent":[10],"video":[11,71,170],"content,":[12],"yet":[13],"their":[14],"slow":[15],"inference":[16,33],"speeds":[17],"and":[18,28,68,99,119,130,146,149,164],"high":[19],"computational":[20],"demands":[21],"hinder":[22],"practical":[23],"deployment.":[24],"Although":[25],"both":[26,97],"quantization":[27,67,98],"sparsity":[29,69],"can":[30],"independently":[31],"accelerate":[32],"while":[34],"maintaining":[35],"generation":[36,171,181],"quality,":[37],"naively":[38],"combining":[39],"these":[40],"techniques":[41],"in":[42],"existing":[43],"training-free":[44],"approaches":[45],"leads":[46],"to":[47,52,108,173],"significant":[48],"performance":[49],"degradation":[50],"due":[51],"lack":[54],"of":[55,65],"joint":[56],"optimization.":[57],"We":[58],"introduce":[59],"FPSAttention,":[60],"a":[61,74,157,165],"novel":[62],"training-aware":[63],"co-design":[64],"FP8":[66],"generation,":[72],"with":[73,133],"focus":[75],"on":[76,143,151],"3D":[78,91],"bi-directional":[79],"attention":[80,162],"mechanism.":[81],"Our":[82],"approach":[83],"features":[84,137],"three":[85],"key":[86],"innovations:":[87],"1)":[88],"A":[89,102,123],"unified":[90],"tile-wise":[92],"granularity":[93],"that":[94,106,127],"simultaneously":[95],"supports":[96],"sparsity;":[100],"2)":[101],"denoising":[103,120],"step-aware":[104],"strategy":[105],"adapts":[107],"noise":[110],"schedule,":[111],"addressing":[112],"strong":[114],"correlation":[115],"between":[116],"quantization/sparsity":[117],"errors":[118],"steps;":[121],"3)":[122],"native,":[124],"hardware-friendly":[125],"kernel":[126,159],"leverages":[128],"FlashAttention":[129],"is":[131],"implemented":[132],"optimized":[134],"Hopper":[135],"architecture":[136],"highly":[139],"efficient":[140],"execution.":[141],"Trained":[142],"Wan2.1's":[144],"1.3B":[145],"14B":[147],"evaluated":[150],"VBench":[153],"benchmark,":[154],"FPSAttention":[155],"achieves":[156],"7.09x":[158],"speedup":[160,168],"operations":[163],"4.96x":[166],"end-to-end":[167],"compared":[172],"BF16":[175],"baseline":[176],"at":[177],"720p":[178],"resolution-without":[179],"sacrificing":[180],"quality.":[182]},"counts_by_year":[],"updated_date":"2026-04-21T08:09:41.155169","created_date":"2025-10-10T00:00:00"}