{"id":"https://openalex.org/W7164149133","doi":"https://doi.org/10.1109/fccm68464.2026.00065","title":"TileFuse: Fused Mixed-Precision Kernels for Quantized LLM Inference on AMD XDNA2 NPUs","display_name":"TileFuse: Fused Mixed-Precision Kernels for Quantized LLM Inference on AMD XDNA2 NPUs","publication_year":2026,"publication_date":"2026-05-13","ids":{"openalex":"https://openalex.org/W7164149133","doi":"https://doi.org/10.1109/fccm68464.2026.00065"},"language":null,"primary_location":{"id":"doi:10.1109/fccm68464.2026.00065","is_oa":false,"landing_page_url":"https://doi.org/10.1109/fccm68464.2026.00065","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE 34th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM)","raw_type":"proceedings-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/A5138330249","display_name":"Wesley Pang","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wesley Pang","raw_affiliation_strings":["University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120467490","display_name":"Gregory Jun","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gregory Jun","raw_affiliation_strings":["University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138390008","display_name":"Feiyang Liu","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Feiyang Liu","raw_affiliation_strings":["University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5056321228","display_name":"Deming Chen","orcid":"https://orcid.org/0000-0002-3016-0270"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Deming Chen","raw_affiliation_strings":["University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign,Department of Electrical and Computer Engineering","institution_ids":["https://openalex.org/I157725225"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157725225"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.79064334,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"277","last_page":"277"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12254","display_name":"Machine Learning in Bioinformatics","score":0.18469999730587006,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},"topics":[{"id":"https://openalex.org/T12254","display_name":"Machine Learning in Bioinformatics","score":0.18469999730587006,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T10222","display_name":"Genomics and Chromatin Dynamics","score":0.12540000677108765,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.10379999876022339,"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/inference","display_name":"Inference","score":0.3864000141620636},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.3255999982357025},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.32440000772476196},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.31470000743865967},{"id":"https://openalex.org/keywords/signal-processing","display_name":"Signal processing","score":0.2614000141620636}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4828000068664551},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.48190000653266907},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4307999908924103},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.413100004196167},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.3864000141620636},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3255999982357025},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.32440000772476196},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.31470000743865967},{"id":"https://openalex.org/C104267543","wikidata":"https://www.wikidata.org/wiki/Q208163","display_name":"Signal processing","level":3,"score":0.2614000141620636},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.2549000084400177}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/fccm68464.2026.00065","is_oa":false,"landing_page_url":"https://doi.org/10.1109/fccm68464.2026.00065","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2026 IEEE 34th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":2,"referenced_works":["https://openalex.org/W4406650295","https://openalex.org/W4410810249"],"related_works":[],"abstract_inverted_index":{"On-device":[0],"LLM":[1,89],"inference":[2,90],"increasingly":[3],"relies":[4],"on":[5],"low-bit":[6],"quantization,":[7],"but":[8],"practical":[9],"formats":[10],"such":[11],"as":[12],"W4A16":[13],"do":[14],"not":[15],"map":[16],"cleanly":[17],"to":[18,78],"existing":[19],"client":[20],"NPU":[21],"software":[22],"stacks.":[23],"We":[24],"present":[25],"TileFuse,":[26],"a":[27,61],"fused":[28],"mixed-precision":[29],"kernel":[30],"library":[31],"for":[32],"AMD":[33],"XDNA2":[34],"NPUs":[35],"that":[36],"directly":[37],"consumes":[38],"quantized":[39],"weights,":[40],"performs":[41],"on-the-fly":[42],"dequantization,":[43],"and":[44,60,76],"executes":[45],"GEMM/GEMV":[46],"without":[47],"materializing":[48],"BF16":[49],"weights":[50],"in":[51,87,91],"DRAM.":[52],"TileFuse":[53,70],"combines":[54],"offline":[55],"interleaved":[56],"pre-tiling,":[57],"optimized":[58],"microkernels,":[59],"memory-core-assisted":[62],"2D":[63],"weight-distribution":[64],"dataflow.":[65],"Across":[66],"Ryzen":[67],"AI":[68],"platforms,":[69],"achieves":[71],"over":[72,83],"2\u00d7":[73],"kernel-level":[74],"gains":[75],"up":[77],"1.9\u00d7":[79],"faster":[80],"prefilling":[81],"with":[82,93],"50%":[84],"lower":[85],"energy":[86],"end-to-end":[88],"comparison":[92],"the":[94],"iGPU":[95],"baseline.":[96]},"counts_by_year":[],"updated_date":"2026-06-26T08:34:08.712188","created_date":"2026-06-11T00:00:00"}