{"id":"https://openalex.org/W7167096080","doi":"https://doi.org/10.1145/3797905.3800540","title":"EPLoN: Exploiting Efficient Parallelism with Selective Rematerialization for Lightning Attention on Ascend NPU","display_name":"EPLoN: Exploiting Efficient Parallelism with Selective Rematerialization for Lightning Attention on Ascend NPU","publication_year":2026,"publication_date":"2026-07-02","ids":{"openalex":"https://openalex.org/W7167096080","doi":"https://doi.org/10.1145/3797905.3800540"},"language":null,"primary_location":{"id":"doi:10.1145/3797905.3800540","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3797905.3800540","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 40th ACM International Conference on Supercomputing","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3797905.3800540","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100657207","display_name":"Han Bao","orcid":"https://orcid.org/0000-0003-3490-0974"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Han Bao","raw_affiliation_strings":["Huawei Technologies, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-3490-0974","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139912757","display_name":"Zhenfeng Su","orcid":"https://orcid.org/0009-0002-3298-8706"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhenfeng Su","raw_affiliation_strings":["Huawei Technologies, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-3298-8706","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120469226","display_name":"\u0410.\u0412. \u0421\u0435\u0442\u044f\u0435\u0432","orcid":"https://orcid.org/0009-0003-4030-4711"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Alexander Setyaev","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0009-0003-4030-4711","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114601886","display_name":"Stanislav Kamenev","orcid":null},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Stanislav Kamenev","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0009-0005-1165-6421","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087783841","display_name":"Alexander Gneushev","orcid":null},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Alexander Gneushev","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0009-0001-1817-782X","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101715904","display_name":"Kang Zhao","orcid":"https://orcid.org/0000-0002-3528-9081"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kang Zhao","raw_affiliation_strings":["Huawei Technologies, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-3528-9081","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010808858","display_name":"Junmin Xiao","orcid":"https://orcid.org/0000-0003-0457-4709"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210090176","display_name":"Institute of Computing Technology","ror":"https://ror.org/0090r4d87","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210090176"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Junmin Xiao","raw_affiliation_strings":["Institute of Computing Technology of Chinese Academy of Sciences, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-0457-4709","affiliations":[{"raw_affiliation_string":"Institute of Computing Technology of Chinese Academy of Sciences, Beijing, China","institution_ids":["https://openalex.org/I19820366","https://openalex.org/I4210090176"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5104320958","display_name":"Haoran Lin","orcid":null},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haoran Lin","raw_affiliation_strings":["Huawei Technologies, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0000-7806-1864","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139904046","display_name":"Anastasiya Bistrigova","orcid":"https://orcid.org/0009-0006-1302-1697"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Anastasiya Bistrigova","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0009-0006-1302-1697","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005872644","display_name":"S.N. Buzykanov","orcid":"https://orcid.org/0009-0005-3405-0705"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Sergey Buzykanov","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0009-0005-3405-0705","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022744098","display_name":"Evgeny Tetin","orcid":"https://orcid.org/0000-0001-6878-8330"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Evgeny Tetin","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0000-0001-6878-8330","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006750799","display_name":"Guangming Tan","orcid":"https://orcid.org/0000-0002-6361-5948"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210090176","display_name":"Institute of Computing Technology","ror":"https://ror.org/0090r4d87","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210090176"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guangming Tan","raw_affiliation_strings":["Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-6361-5948","affiliations":[{"raw_affiliation_string":"Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China","institution_ids":["https://openalex.org/I19820366","https://openalex.org/I4210090176"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114973604","display_name":"Boxiao Liu","orcid":"https://orcid.org/0000-0002-9792-1361"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Boxiao Liu","raw_affiliation_strings":["Huawei Technologies, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-9792-1361","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063242080","display_name":"Xueyi Zou","orcid":null},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xueyi Zou","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0009-0004-7565-7098","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021124418","display_name":"Zhenhua Dong","orcid":"https://orcid.org/0000-0002-2231-4663"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhenhua Dong","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0000-0002-2231-4663","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044370695","display_name":"C. C. Korikov","orcid":"https://orcid.org/0000-0001-6227-5542"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Constantine Korikov","raw_affiliation_strings":["Huawei Technologies, Shenzhen, China"],"raw_orcid":"https://orcid.org/0000-0001-6227-5542","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Shenzhen, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021890706","display_name":"Xianzhi Yu","orcid":"https://orcid.org/0000-0002-1497-5525"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xianzhi Yu","raw_affiliation_strings":["Huawei Technologies, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-1497-5525","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5075481740","display_name":"Zhongzhe Hu","orcid":"https://orcid.org/0000-0002-6708-3942"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhongzhe Hu","raw_affiliation_strings":["Huawei Technologies, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-6708-3942","affiliations":[{"raw_affiliation_string":"Huawei Technologies, Beijing, China","institution_ids":["https://openalex.org/I2250955327"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"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":"1040","last_page":"1052"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.4571000039577484,"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"}},"topics":[{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.4571000039577484,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.11299999803304672,"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"}},{"id":"https://openalex.org/T14347","display_name":"Big Data and Digital Economy","score":0.11289999634027481,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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":0.7720999717712402},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.5853000283241272},{"id":"https://openalex.org/keywords/multithreading","display_name":"Multithreading","score":0.47870001196861267},{"id":"https://openalex.org/keywords/parsec","display_name":"Parsec","score":0.41530001163482666},{"id":"https://openalex.org/keywords/parallelism","display_name":"Parallelism (grammar)","score":0.4081000089645386},{"id":"https://openalex.org/keywords/implementation","display_name":"Implementation","score":0.3968000113964081},{"id":"https://openalex.org/keywords/instruction-level-parallelism","display_name":"Instruction-level parallelism","score":0.3944999873638153},{"id":"https://openalex.org/keywords/computational-complexity-theory","display_name":"Computational complexity theory","score":0.37540000677108765}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8270999789237976},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.7720999717712402},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.7060999870300293},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.5853000283241272},{"id":"https://openalex.org/C201410400","wikidata":"https://www.wikidata.org/wiki/Q1064412","display_name":"Multithreading","level":3,"score":0.47870001196861267},{"id":"https://openalex.org/C44060867","wikidata":"https://www.wikidata.org/wiki/Q12129","display_name":"Parsec","level":3,"score":0.41530001163482666},{"id":"https://openalex.org/C2781172179","wikidata":"https://www.wikidata.org/wiki/Q853109","display_name":"Parallelism (grammar)","level":2,"score":0.4081000089645386},{"id":"https://openalex.org/C26713055","wikidata":"https://www.wikidata.org/wiki/Q245962","display_name":"Implementation","level":2,"score":0.3968000113964081},{"id":"https://openalex.org/C140763907","wikidata":"https://www.wikidata.org/wiki/Q2714055","display_name":"Instruction-level parallelism","level":3,"score":0.3944999873638153},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.37540000677108765},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.34610000252723694},{"id":"https://openalex.org/C61483411","wikidata":"https://www.wikidata.org/wiki/Q3124522","display_name":"Data parallelism","level":3,"score":0.326200008392334},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.32519999146461487},{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.31839999556541443},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2912999987602234},{"id":"https://openalex.org/C164620267","wikidata":"https://www.wikidata.org/wiki/Q376953","display_name":"Adder","level":3,"score":0.2757999897003174},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.26750001311302185},{"id":"https://openalex.org/C69398868","wikidata":"https://www.wikidata.org/wiki/Q129052","display_name":"Lightning (connector)","level":3,"score":0.26649999618530273},{"id":"https://openalex.org/C106515295","wikidata":"https://www.wikidata.org/wiki/Q26806595","display_name":"Parallel processing","level":2,"score":0.2590000033378601}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3797905.3800540","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3797905.3800540","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 40th ACM International Conference on Supercomputing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3797905.3800540","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3797905.3800540","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 40th ACM International Conference on Supercomputing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":5,"referenced_works":["https://openalex.org/W4389524555","https://openalex.org/W4401857375","https://openalex.org/W4415797154","https://openalex.org/W7133214447","https://openalex.org/W7133227460"],"related_works":[],"abstract_inverted_index":{"The":[0],"quadratic":[1],"computational":[2,124],"complexity":[3],"of":[4,45,75,105,122,131,141],"softmax":[5],"attention":[6],"presents":[7,71],"a":[8,26,72,93,110,129,137],"fundamental":[9],"bottleneck":[10],"to":[11,16,40,133],"scaling":[12],"modern":[13],"language":[14],"models":[15],"long":[17],"sequences.":[18],"While":[19],"the":[20,43,85,103,106,114],"proposed":[21],"Lightning":[22,66,76],"Attention":[23,67,77],"mechanism":[24],"offers":[25],"linear-complexity":[27],"alternative,":[28],"its":[29],"state-of-the-art":[30,115],"implementations":[31],"remain":[32],"predominantly":[33],"optimized":[34,78],"for":[35,65,79],"GPU":[36,121],"architectures":[37],"and":[38,99,136],"fail":[39],"fully":[41],"leverage":[42],"capabilities":[44],"alternative":[46],"accelerators":[47],"such":[48],"as":[49],"Ascend":[50,81],"NPUs.":[51,82],"To":[52],"bridge":[53],"this":[54],"gap,":[55],"we":[56],"propose":[57],"EPLoN":[58,70,83],"(Exploiting":[59],"Efficient":[60],"Parallelism":[61],"with":[62,92],"Selective":[63],"Rematerialization":[64],"on":[68,97,118],"NPU).":[69],"high-performance":[73],"implementation":[74],"heterogeneous":[80],"reformulates":[84],"algorithm,":[86],"introducing":[87],"an":[88,119],"efficient":[89],"parallelism":[90],"scheme":[91],"rematerialization":[94],"strategy":[95],"based":[96],"inter-":[98],"intra-core":[100],"that":[101],"maximizes":[102],"utilization":[104],"NPU":[107],"architecture.":[108],"In":[109],"cross-architectural":[111],"comparison":[112],"against":[113],"FlashLinearAttention":[116],"(FLA)":[117],"Nvidia":[120],"comparable":[123],"capacity,":[125],"our":[126],"evaluation":[127],"achieves":[128],"speedup":[130,140],"up":[132],"3.39":[134],"\u00d7":[135],"geometric":[138],"mean":[139],"1.73":[142],"\u00d7,":[143],"while":[144],"reducing":[145],"peak":[146],"memory":[147],"consumption":[148],"by":[149],"approximately":[150],"33%.":[151]},"counts_by_year":[],"updated_date":"2026-07-22T07:51:19.307946","created_date":"2026-07-03T00:00:00"}
