{"id":"https://openalex.org/W4404411111","doi":"https://doi.org/10.1109/tc.2024.3500362","title":"SADIMM: Accelerating Sparse Attention Using DIMM-Based Near-Memory Processing","display_name":"SADIMM: Accelerating Sparse Attention Using DIMM-Based Near-Memory Processing","publication_year":2024,"publication_date":"2024-11-15","ids":{"openalex":"https://openalex.org/W4404411111","doi":"https://doi.org/10.1109/tc.2024.3500362"},"language":"en","primary_location":{"id":"doi:10.1109/tc.2024.3500362","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tc.2024.3500362","pdf_url":null,"source":{"id":"https://openalex.org/S157670870","display_name":"IEEE Transactions on Computers","issn_l":"0018-9340","issn":["0018-9340","1557-9956","2326-3814"],"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 Transactions on Computers","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":"https://openalex.org/A5064127766","display_name":"Huize Li","orcid":"https://orcid.org/0000-0002-8710-4472"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":true,"raw_author_name":"Huize Li","raw_affiliation_strings":["School of Computing, National University of Singapore, Singapore"],"affiliations":[{"raw_affiliation_string":"School of Computing, National University of Singapore, Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101745478","display_name":"Dan Chen","orcid":"https://orcid.org/0000-0003-4068-1667"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Dan Chen","raw_affiliation_strings":["School of Computing, National University of Singapore, Singapore"],"affiliations":[{"raw_affiliation_string":"School of Computing, National University of Singapore, Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5049237676","display_name":"Tulika Mitra","orcid":"https://orcid.org/0000-0003-4136-4188"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Tulika Mitra","raw_affiliation_strings":["School of Computing, National University of Singapore, Singapore"],"affiliations":[{"raw_affiliation_string":"School of Computing, National University of Singapore, Singapore","institution_ids":["https://openalex.org/I165932596"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5064127766"],"corresponding_institution_ids":["https://openalex.org/I165932596"],"apc_list":null,"apc_paid":null,"fwci":0.8383,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.74090525,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":97},"biblio":{"volume":"74","issue":"2","first_page":"542","last_page":"554"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9847000241279602,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9847000241279602,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic 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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.965399980545044,"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":0.9613999724388123,"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":0.4988727569580078},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.4063279330730438}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4988727569580078},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4063279330730438}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tc.2024.3500362","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tc.2024.3500362","pdf_url":null,"source":{"id":"https://openalex.org/S157670870","display_name":"IEEE Transactions on Computers","issn_l":"0018-9340","issn":["0018-9340","1557-9956","2326-3814"],"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 Transactions on Computers","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":40,"referenced_works":["https://openalex.org/W2034861439","https://openalex.org/W2113235308","https://openalex.org/W2605347906","https://openalex.org/W2795118915","https://openalex.org/W2923014074","https://openalex.org/W2963323070","https://openalex.org/W2979826702","https://openalex.org/W3017024317","https://openalex.org/W3043023836","https://openalex.org/W3111721177","https://openalex.org/W3146763006","https://openalex.org/W3158275024","https://openalex.org/W3192841685","https://openalex.org/W3199934250","https://openalex.org/W3206453033","https://openalex.org/W3207399097","https://openalex.org/W4214612132","https://openalex.org/W4214686755","https://openalex.org/W4280496502","https://openalex.org/W4281687131","https://openalex.org/W4281731825","https://openalex.org/W4293023446","https://openalex.org/W4293024045","https://openalex.org/W4380881178","https://openalex.org/W4385245566","https://openalex.org/W4386763995","https://openalex.org/W4386764227","https://openalex.org/W4390241252","https://openalex.org/W4393407269","https://openalex.org/W4394891385","https://openalex.org/W4404134025","https://openalex.org/W6674147819","https://openalex.org/W6676984168","https://openalex.org/W6719768283","https://openalex.org/W6727099177","https://openalex.org/W6755207826","https://openalex.org/W6776048684","https://openalex.org/W6796814978","https://openalex.org/W6838322825","https://openalex.org/W6854308872"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Self-attention":[0],"mechanism":[1],"is":[2,89],"the":[3,65,77,84,97,132,167,179],"performance":[4,44,57],"bottleneck":[5],"of":[6,99,169],"Transformer":[7,20],"based":[8],"language":[9],"models.":[10],"In":[11,118,148],"response,":[12],"researchers":[13],"have":[14],"proposed":[15],"sparse":[16,23,52,81,115],"attention":[17,24,53,116,133],"to":[18,59,75,93,126,174],"expedite":[19],"execution.":[21],"However,":[22],"involves":[25,136],"massive":[26],"random":[27],"access,":[28],"rendering":[29],"it":[30],"as":[31,38],"a":[32,111,122,152],"memory-intensive":[33,47],"kernel.":[34],"Memory-based":[35],"architectures,":[36],"such":[37],"<i>near-memory":[39],"processing</i>":[40],"(NMP),":[41],"demonstrate":[42,188],"notable":[43],"enhancements":[45],"in":[46,80],"applications.":[48],"Nonetheless,":[49],"existing":[50],"NMP-based":[51,114],"accelerators":[54],"face":[55],"suboptimal":[56],"due":[58],"hardware":[60,66,146],"and":[61,185,199],"software":[62,85],"challenges.":[63],"On":[64,83],"front,":[67],"current":[68],"solutions":[69],"employ":[70],"homogeneous":[71],"logic":[72,139],"integration,":[73],"struggling":[74],"support":[76,128],"diverse":[78],"operations":[79,130],"attention.":[82],"side,":[86],"token-based":[87],"dataflow":[88],"commonly":[90],"adopted,":[91],"leading":[92],"load":[94,164],"imbalance":[95],"after":[96,166],"pruning":[98,168],"weakly":[100,170],"connected":[101,171],"tokens.":[102,172],"To":[103],"address":[104],"these":[105],"challenges,":[106],"this":[107],"paper":[108],"introduces":[109],"SADIMM,":[110],"hardware-software":[112],"co-designed":[113],"accelerator.":[117],"hardware,":[119],"we":[120,150],"propose":[121],"heterogeneous":[123],"integration":[124],"approach":[125,162],"efficiently":[127],"various":[129],"within":[131],"mechanism.":[134],"This":[135,161],"employing":[137],"different":[138,142],"units":[140],"for":[141],"operations,":[143],"thereby":[144],"improving":[145],"efficiency.":[147],"software,":[149],"implement":[151],"dimension-based":[153],"dataflow,":[154],"dividing":[155],"input":[156],"sequences":[157],"by":[158],"model":[159],"dimensions.":[160],"achieves":[163,191],"balancing":[165],"Compared":[173],"NVIDIA":[175],"RTX":[176],"A6000":[177],"GPU,":[178],"experimental":[180],"results":[181],"on":[182],"BERT,":[183],"BART,":[184],"GPT-2":[186],"models":[187],"that":[189],"SADIMM":[190],"48<inline-formula><tex-math":[192],"notation=\"LaTeX\">$\\boldsymbol{\\times}$</tex-math></inline-formula>,":[193,195,201,203],"35<inline-formula><tex-math":[194],"37<inline-formula><tex-math":[196],"notation=\"LaTeX\">$\\boldsymbol{\\times}$</tex-math></inline-formula>":[197,205],"speedups":[198],"194<inline-formula><tex-math":[200],"202<inline-formula><tex-math":[202],"191<inline-formula><tex-math":[204],"energy":[206],"efficiency":[207],"improvement,":[208],"respectively.":[209]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":1}],"updated_date":"2026-03-27T14:29:43.386196","created_date":"2024-11-16T00:00:00"}