{"id":"https://openalex.org/W7225834873","doi":"https://doi.org/10.48550/arxiv.2611.13741","title":"Retargetable Compilation for Tensor Kernels via a Unified Layout Abstraction","display_name":"Retargetable Compilation for Tensor Kernels via a Unified Layout Abstraction","publication_year":2026,"publication_date":"2026-11-10","ids":{"openalex":"https://openalex.org/W7225834873","doi":"https://doi.org/10.48550/arxiv.2611.13741"},"language":"en","primary_location":{"id":"doi:10.48550/arxiv.2611.13741","doi":"https://doi.org/10.48550/arxiv.2611.13741","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13741","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2611.13741","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5155275055","display_name":"Likai Ma","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ma, Likai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004459947","display_name":"Renze Chen","orcid":"https://orcid.org/0000-0001-5938-7965","observed_orcids":["https://orcid.org/0000-0001-5938-7965"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Renze","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100777513","display_name":"Tianqi Li","orcid":"https://orcid.org/0009-0005-1454-9058","observed_orcids":["https://orcid.org/0009-0005-1454-9058","https://orcid.org/0000-0001-6688-3115"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Tianqi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161474176","display_name":"Yuhao Luo","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Yuhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161461804","display_name":"Youwei Xiao","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Youwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161454628","display_name":"Size Zheng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Size","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161420515","display_name":"Xiuhong Li","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Xiuhong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5161465238","display_name":"Yun Liang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Yun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"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":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9951000213623047,"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":0.9951000213623047,"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/T10904","display_name":"Embedded Systems Design Techniques","score":0.002899999963119626,"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/T10126","display_name":"Logic, programming, and type systems","score":0.0003000000142492354,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8267999887466431},{"id":"https://openalex.org/C55526617","wikidata":"https://www.wikidata.org/wiki/Q719375","display_name":"Operand","level":2,"score":0.8242999911308289},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.7710999846458435},{"id":"https://openalex.org/C170723468","wikidata":"https://www.wikidata.org/wiki/Q182933","display_name":"x86","level":3,"score":0.7107999920845032},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.5852000117301941},{"id":"https://openalex.org/C164155591","wikidata":"https://www.wikidata.org/wiki/Q2067766","display_name":"Satisfiability modulo theories","level":2,"score":0.5795000195503235},{"id":"https://openalex.org/C2780575108","wikidata":"https://www.wikidata.org/wiki/Q7316652","display_name":"Retargeting","level":2,"score":0.5654000043869019},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.5160999894142151},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5138999819755554},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.453000009059906},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.37700000405311584},{"id":"https://openalex.org/C2778770139","wikidata":"https://www.wikidata.org/wiki/Q1966904","display_name":"Solver","level":2,"score":0.3700000047683716},{"id":"https://openalex.org/C73564150","wikidata":"https://www.wikidata.org/wiki/Q11417093","display_name":"Instruction scheduling","level":5,"score":0.36309999227523804},{"id":"https://openalex.org/C188985296","wikidata":"https://www.wikidata.org/wiki/Q868954","display_name":"Page layout","level":2,"score":0.36239999532699585},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.35030001401901245},{"id":"https://openalex.org/C155281189","wikidata":"https://www.wikidata.org/wiki/Q3518150","display_name":"Tensor (intrinsic definition)","level":2,"score":0.3393000066280365},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.30880001187324524},{"id":"https://openalex.org/C2982832238","wikidata":"https://www.wikidata.org/wiki/Q5531640","display_name":"General purpose","level":2,"score":0.30660000443458557},{"id":"https://openalex.org/C190902152","wikidata":"https://www.wikidata.org/wiki/Q1325106","display_name":"Optimizing compiler","level":3,"score":0.301800012588501},{"id":"https://openalex.org/C150202949","wikidata":"https://www.wikidata.org/wiki/Q107602","display_name":"Pointer (user interface)","level":2,"score":0.2937000095844269},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.29089999198913574},{"id":"https://openalex.org/C2765594","wikidata":"https://www.wikidata.org/wiki/Q2624187","display_name":"Integrated circuit layout","level":3,"score":0.2743000090122223},{"id":"https://openalex.org/C202491316","wikidata":"https://www.wikidata.org/wiki/Q272683","display_name":"Instruction set","level":2,"score":0.2669999897480011},{"id":"https://openalex.org/C164620267","wikidata":"https://www.wikidata.org/wiki/Q376953","display_name":"Adder","level":3,"score":0.2669000029563904},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.2522999942302704}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2611.13741","doi":"https://doi.org/10.48550/arxiv.2611.13741","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13741","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2611.13741","doi":"https://doi.org/10.48550/arxiv.2611.13741","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13741","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Modern":[0],"hardware":[1,135],"accelerators":[2],"achieve":[3],"high":[4],"performance":[5],"through":[6],"specialized":[7],"instructions":[8,16,50,188],"coupled":[9],"with":[10],"tightly":[11],"managed":[12],"memory":[13,141],"hierarchies.":[14],"These":[15],"impose":[17],"strict":[18],"layout":[19,40,65,76,102,116,127,146,162,179],"requirements":[20,66,103],"on":[21,208,211],"their":[22,190],"operands:":[23],"where":[24],"tensor":[25,129],"fragments":[26],"reside,":[27],"how":[28,33,54],"they":[29,58],"are":[30,154,164],"organized,":[31],"and":[32,53,70,78,140,148,181,189,201,210,213,220,222,229],"parallel":[34,138],"units":[35],"access":[36],"them.":[37],"Such":[38],"instruction-visible":[39,115],"is":[41,124],"critical":[42],"to":[43,81,134],"kernel":[44],"performance,":[45],"as":[46,117],"it":[47],"determines":[48],"whether":[49],"execute":[51],"correctly":[52],"much":[55],"data":[56],"movement":[57],"incur.":[59],"However,":[60],"existing":[61],"compilers":[62],"either":[63],"scatter":[64],"across":[67,104,158,196],"handwritten":[68],"programs":[69],"local":[71],"scheduling":[72],"rules,":[73],"or":[74],"provide":[75],"representations":[77],"conversions":[79],"specific":[80],"a":[82,97,110,125],"single":[83],"class":[84],"of":[85,96],"targets.":[86],"Both":[87],"approaches":[88],"expose":[89],"the":[90,94,159,174,178,185,223],"same":[91],"abstraction":[92],"gap:":[93],"absence":[95],"unified":[98],"model":[99],"for":[100],"instruction":[101,119,168],"backends.":[105],"This":[106],"paper":[107],"presents":[108],"Arknife,":[109],"retargetable":[111,224],"compiler":[112,225],"that":[113],"treats":[114],"first-class":[118],"semantics.":[120],"Its":[121],"central":[122],"idea":[123],"constraint-based":[126],"algebra:":[128],"layouts":[130],"map":[131],"logical":[132],"axes":[133,136],"spanning":[137],"levels":[139],"spaces.":[142],"Instruction":[143],"operands":[144],"declare":[145],"requirements,":[147],"target-provided":[149],"repairs":[150],"resolve":[151],"conflicts.":[152],"Layouts":[153],"then":[155],"solved":[156],"globally":[157],"program.":[160],"ISA-specific":[161],"facts":[163],"supplied":[165],"by":[166,218,227],"target-side":[167],"descriptions":[169],"rather":[170],"than":[171],"hardcoded":[172],"in":[173],"compiler.":[175],"Retargeting":[176],"reuses":[177],"algebra":[180],"solver":[182],"while":[183],"defining":[184],"new":[186],"target's":[187],"operand":[191],"requirements.":[192],"We":[193],"evaluate":[194],"Arknife":[195,205],"x86":[197,212],"CPU,":[198],"Hexagon":[199,214],"NPU,":[200],"NVIDIA":[202],"A100":[203],"GPU.":[204],"matches":[206],"cuBLAS":[207],"A100,":[209],"outperforms":[215],"vendor-optimized":[216],"libraries":[217],"$2.17\\times$":[219],"$3.78\\times$":[221],"MISAAL":[226],"$5.08\\times$":[228],"$4.53\\times$,":[230],"respectively.":[231]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-12T00:00:00"}
