{"id":"https://openalex.org/W7134979112","doi":"https://doi.org/10.1109/asp-dac66049.2026.11420624","title":"SpAct-NDP: Efficient LLM Inference via Sparse Activation on NDP-GPU Heterogeneous Architecture","display_name":"SpAct-NDP: Efficient LLM Inference via Sparse Activation on NDP-GPU Heterogeneous Architecture","publication_year":2026,"publication_date":"2026-01-19","ids":{"openalex":"https://openalex.org/W7134979112","doi":"https://doi.org/10.1109/asp-dac66049.2026.11420624"},"language":null,"primary_location":{"id":"doi:10.1109/asp-dac66049.2026.11420624","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asp-dac66049.2026.11420624","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 31st Asia and South Pacific Design Automation Conference (ASP-DAC)","raw_type":"proceedings-article"},"type":"conference-paper","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/A5101746186","display_name":"Jiaming Xu","orcid":"https://orcid.org/0000-0001-6104-4742"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiaming Xu","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5073502048","display_name":"Tongxin Xie","orcid":"https://orcid.org/0009-0005-1620-8927"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tongxin Xie","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128790601","display_name":"Yongkang Zhou","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yongkang Zhou","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021616613","display_name":"Jinhao Li","orcid":"https://orcid.org/0000-0002-4812-3320"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jinhao Li","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102585583","display_name":"Yaoxiu Lian","orcid":"https://orcid.org/0009-0007-7858-5132"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yaoxiu Lian","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127971384","display_name":"Zhenhua Zhu","orcid":null},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhenhua Zhu","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128764896","display_name":"Yu Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yu Wang","raw_affiliation_strings":["Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5128714238","display_name":"Guohao Dai","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guohao Dai","raw_affiliation_strings":["Shanghai Jiao Tong University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University","institution_ids":["https://openalex.org/I183067930"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"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":"568","last_page":"574"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12326","display_name":"Network Packet Processing and Optimization","score":0.3102000057697296,"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/T12326","display_name":"Network Packet Processing and Optimization","score":0.3102000057697296,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.14990000426769257,"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/T11181","display_name":"Advanced Data Storage Technologies","score":0.029500000178813934,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/inference","display_name":"Inference","score":0.44359999895095825},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.2890999913215637},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.2806999981403351},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2515000104904175},{"id":"https://openalex.org/keywords/domain","display_name":"Domain (mathematical analysis)","score":0.25040000677108765}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6470999717712402},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.44359999895095825},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43130001425743103},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.29089999198913574},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.2890999913215637},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2806999981403351},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2565000057220459},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2515000104904175},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.25040000677108765},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.24619999527931213}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/asp-dac66049.2026.11420624","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asp-dac66049.2026.11420624","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 31st Asia and South Pacific Design Automation Conference (ASP-DAC)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.4726899266242981,"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy"}],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320327803","display_name":"Shanghai Rising-Star Program","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":15,"referenced_works":["https://openalex.org/W2962834855","https://openalex.org/W3189166979","https://openalex.org/W4221001402","https://openalex.org/W4308083526","https://openalex.org/W4387321091","https://openalex.org/W4392427708","https://openalex.org/W4395106409","https://openalex.org/W4400985603","https://openalex.org/W4401211704","https://openalex.org/W4402671659","https://openalex.org/W4404401018","https://openalex.org/W4405440056","https://openalex.org/W4409248815","https://openalex.org/W4409282445","https://openalex.org/W4411486604"],"related_works":[],"abstract_inverted_index":{"Sparse":[0],"activation":[1,6,174,292],"is":[2,59,224],"caused":[3],"by":[4,34,180],"the":[5,11,53,60,64,84,90,127,136,142,155,162,163,170,182,192,209,217,220,234,241,284],"function":[7],"(e.g.,":[8],"ReLU)":[9],"in":[10,31],"feed-forward":[12],"network":[13],"(FFN)":[14],"of":[15,74,106,112,121,126,172,211,219],"large":[16],"language":[17],"models":[18],"(LLMs),":[19],"and":[20,39,55,67,80,185,232,244,255,269,274,277,297],"recently":[21],"emerges":[22],"as":[23],"a":[24],"promising":[25],"method":[26],"for":[27,63,99,146,227,288],"LLM":[28,75,148,289],"inference":[29,76,149],"acceleration":[30],"resource-constrained":[32],"scenarios":[33],"effectively":[35],"reducing":[36,249],"computational":[37],"workload":[38,184,205],"memory":[40,110,156,188],"requirements":[41],"with":[42,77,150,229,283,290],">80":[43],"%":[44],"predicted":[45,129,221],"dynamic":[46,56,91],"sparsity.":[47],"In":[48],"this":[49],"paper,":[50],"we":[51,139,160,196,213],"identify":[52],"heavy":[54],"data":[57],"transfer":[58],"primary":[61],"reason":[62],"significant":[65],"synchronization":[66],"poor":[68],"GPU":[69],"utilization":[70,179],"during":[71,109,119,131,158,194],"decoding":[72],"phase":[73],"sparse":[78,92,128,151,173,222,245,291],"activation,":[79,93],"propose":[81,197,233],"to":[82,88,175,202,240,266],"apply":[83],"near-data-processing":[85],"(NDP)":[86],"architecture":[87,145],"handle":[89],"while":[94],"addressing":[95],"three":[96],"critical":[97],"challenges":[98],"further":[100],"NDP-GPU":[101,132],"collaboration":[102,210],"optimization.":[103],"(1)":[104,153],"Under-utilization":[105],"DRAM":[107,177],"bandwidth":[108,178],"access":[111,157],"NDP.":[113,122],"(2)":[114,190],"Workload":[115],"imbalance":[116],"across":[117],"channels":[118],"computation":[120,193],"(3)":[123,207],"Time-consuming":[124],"parsing":[125,218,237],"pattern":[130,223,246],"collaboration.":[133],"To":[134],"tackle":[135],"above":[137],"challenges,":[138],"present":[140],"SpAct-NDP,":[141],"NDPGPU":[143],"heterogeneous":[144],"efficient":[147],"activation.":[152],"For":[154,191,208],"NDP,":[159,195],"design":[161],"specific":[164],"sparsity-aware":[165],"weight":[166],"mapping":[167],"strategy":[168],"considering":[169],"characteristics":[171],"improve":[176],"balancing":[181],"bank":[183],"eliminating":[186],"redundant":[187],"access.":[189],"two-level":[198],"heuristic":[199],"scheduling":[200],"system":[201],"achieve":[203],"channel-wise":[204],"balance.":[206],"NDP-GPU,":[212],"point":[214],"out":[215],"that":[216,262],"more":[225],"suitable":[226],"GPUs":[228],"high":[230],"parallelism":[231],"request-weight":[235],"pair":[236],"mechanism":[238],"according":[239],"input":[242],"requests":[243],"on":[247,293],"GPU,":[248],"$\\sim":[250,256],"3":[251],"\\times$":[252,258,268,271,276,279],"execution":[253],"time":[254],"9":[257],"memory.":[259],"Experiments":[260],"show":[261],"SpActNDP":[263],"achieves":[264],"up":[265],"$2.17":[267],"$1.92":[270],"end-to-end":[272],"speedup":[273],"$1.53":[275],"$1.45":[278],"energy":[280],"efficiency":[281],"compared":[282],"SOTA":[285],"software":[286],"frameworks":[287],"NVIDIA":[294,298],"RTX":[295],"3090":[296],"Tesla":[299],"A100.":[300]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2026-03-12T00:00:00"}