{"id":"https://openalex.org/W7148243161","doi":"https://doi.org/10.1109/hipcw66559.2025.00114","title":"Accelerating Quantized Deep Learning on Arm with JIT-Compiled INT8 Matmul Kernels","display_name":"Accelerating Quantized Deep Learning on Arm with JIT-Compiled INT8 Matmul Kernels","publication_year":2025,"publication_date":"2025-12-17","ids":{"openalex":"https://openalex.org/W7148243161","doi":"https://doi.org/10.1109/hipcw66559.2025.00114"},"language":null,"primary_location":{"id":"doi:10.1109/hipcw66559.2025.00114","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hipcw66559.2025.00114","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE 32nd International Conference on High Performance Computing, Data and Analytics Workshop (HiPCW)","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/A5065411417","display_name":"Shreyas Shankar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shreyas Shankar","raw_affiliation_strings":["Fujitsu Research of India,Bangalore,India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fujitsu Research of India,Bangalore,India","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132798823","display_name":"Abhishek Jain","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Abhishek Jain","raw_affiliation_strings":["Fujitsu Research of India,Bangalore,India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fujitsu Research of India,Bangalore,India","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132805841","display_name":"Ashish Chopra","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ashish Chopra","raw_affiliation_strings":["Fujitsu Research of India,Bangalore,India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fujitsu Research of India,Bangalore,India","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5116479487","display_name":"Masahiro Doteguchi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Masahiro Doteguchi","raw_affiliation_strings":["Fujitsu Research of India,Bangalore,India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fujitsu Research of India,Bangalore,India","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112225690","display_name":"A. Nukariya","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Atsushi Nukariya","raw_affiliation_strings":["Fujitsu Research of India,Bangalore,India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fujitsu Research of India,Bangalore,India","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5132827667","display_name":"Priyanka Sharma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Priyanka Sharma","raw_affiliation_strings":["Fujitsu Research of India,Bangalore,India"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fujitsu Research of India,Bangalore,India","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"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":"319","last_page":"320"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.1875,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.1875,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.17829999327659607,"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/T11448","display_name":"Face recognition and analysis","score":0.06530000269412994,"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/deep-learning","display_name":"Deep learning","score":0.6128000020980835},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.3312000036239624},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.3255000114440918},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3102000057697296},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.29350000619888306},{"id":"https://openalex.org/keywords/autoencoder","display_name":"Autoencoder","score":0.2867000102996826}],"concepts":[{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6331999897956848},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.6128000020980835},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.491100013256073},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.43209999799728394},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.34880000352859497},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.3312000036239624},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.3255000114440918},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3102000057697296},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.29350000619888306},{"id":"https://openalex.org/C101738243","wikidata":"https://www.wikidata.org/wiki/Q786435","display_name":"Autoencoder","level":3,"score":0.2867000102996826},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.2759999930858612},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.2750999927520752},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.2667999863624573},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.2551000118255615}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/hipcw66559.2025.00114","is_oa":false,"landing_page_url":"https://doi.org/10.1109/hipcw66559.2025.00114","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE 32nd International Conference on High Performance Computing, Data and Analytics Workshop (HiPCW)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321034","display_name":"New Energy and Industrial Technology Development Organization","ror":"https://ror.org/0055k7a87"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Quantization":[0],"enables":[1,91],"efficient":[2,25],"CPU":[3],"inference":[4,50,162],"by":[5],"reducing":[6],"memory":[7,126],"and":[8,56,67,95,104,118,125,139,165,191,200],"compute":[9,124],"costs":[10],"through":[11],"INT8":[12,35,49,143,170,181],"precision,":[13],"a":[14,75],"key":[15],"optimization":[16,96],"for":[17,47,60,149],"deploying":[18],"large":[19],"models":[20],"like":[21],"LLMs":[22],"on":[23,37,51,69,115,174,185],"energy":[24],"systems.":[26,176,202],"Since":[27],"matrix":[28,82],"multiplication":[29,83],"dominates":[30],"deep":[31,182],"learning":[32,183],"workloads,":[33],"optimizing":[34],"matmul":[36,138],"CPUs":[38],"yields":[39],"substantial":[40],"performance":[41,66],"gains.":[42],"While":[43],"the":[44,146,153],"existing":[45,142],"kernels":[46,85,144],"quantized":[48],"Arm":[52,186],"platform":[53],"is":[54],"scarce":[55],"lacks":[57],"architectural":[58],"tuning":[59],"varying":[61],"shapes":[62,117],"threads":[63],"thus":[64],"limiting":[65],"scalability":[68,190],"multi-core":[70],"CPUs.":[71],"This":[72,177],"work":[73,178],"introduces":[74],"suite":[76],"of":[77,123],"highly":[78],"optimized":[79],"8-bit":[80],"JIT":[81],"(GEMM)":[84],"using":[86],"oneDNN.":[87],"Just-In-Time":[88],"(JIT)":[89],"compilation":[90],"runtime":[92],"code":[93],"generation,":[94],"tailored":[97],"to":[98,133,159],"specific":[99],"workload":[100],"parameters,":[101],"providing":[102],"flexibility":[103],"better":[105],"hardware":[106],"utilization.":[107],"Our":[108],"kernel":[109,130,147],"supports":[110],"variable":[111],"data":[112],"blocking":[113],"depending":[114],"input":[116],"threading,":[119],"achieving":[120],"optimal":[121],"utilization":[122],"resources.":[127],"The":[128],"proposed":[129],"delivers":[131],"up":[132,158],"7x":[134],"speedup":[135],"over":[136,141],"FP32":[137,164],"4x":[140,160],"at":[145],"level":[148],"LLM-relevant":[150],"shapes.":[151],"At":[152],"model":[154],"level,":[155],"it":[156],"achieves":[157],"faster":[161,167],"than":[163,168],"2.5x":[166],"current":[169],"implementations":[171],"in":[172],"PyTorch":[173],"Arm-based":[175],"substantially":[179],"accelerates":[180],"workloads":[184],"HPC":[187],"platforms,":[188],"improving":[189],"efficiency":[192],"across":[193],"domains":[194],"such":[195],"as":[196],"NLP,":[197],"computer":[198],"vision,":[199],"recommendation":[201]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2026-04-03T00:00:00"}