{"id":"https://openalex.org/W7225635269","doi":"https://doi.org/10.5281/zenodo.24071602","title":"Eliminating the Autograd Memory Wall via Continuous-to-Discrete Sobolev Space Friedrichs Mollification in Fixed-Point Silicon","display_name":"Eliminating the Autograd Memory Wall via Continuous-to-Discrete Sobolev Space Friedrichs Mollification in Fixed-Point Silicon","publication_year":2026,"publication_date":"2026-11-09","ids":{"openalex":"https://openalex.org/W7225635269","doi":"https://doi.org/10.5281/zenodo.24071602"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.24071602","doi":"https://doi.org/10.5281/zenodo.24071602","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24071602","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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/10.5281/zenodo.24071602","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5161344411","display_name":"Jonathan R. Ament","orcid":"https://orcid.org/0009-0005-9058-9991","observed_orcids":["https://orcid.org/0009-0005-9058-9991"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ament, Jonathan R.","raw_affiliation_strings":["Mavericks Alpha Signal"],"raw_orcid":"https://orcid.org/0009-0005-9058-9991","affiliations":[{"raw_affiliation_string":"Mavericks Alpha Signal","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5161344411"],"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/T10904","display_name":"Embedded Systems Design Techniques","score":0.8291000127792358,"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/T10904","display_name":"Embedded Systems Design Techniques","score":0.8291000127792358,"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/T11697","display_name":"Numerical Methods and Algorithms","score":0.08730000257492065,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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/T10363","display_name":"Low-power high-performance VLSI design","score":0.019500000402331352,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/reverse-mode-automatic-differentiation","display_name":"reverse-mode differentiation","score":0.7670000195503235},{"id":"https://openalex.org/keywords/memory-wall","display_name":"memory wall","score":0.5189999938011169},{"id":"https://openalex.org/keywords/analytic-derivatives","display_name":"analytical derivatives","score":0.5600000023841858},{"id":"https://openalex.org/keywords/fpga-accelerator","display_name":"FPGA accelerator","score":0.4059999883174896},{"id":"https://openalex.org/keywords/systemverilog","display_name":"SystemVerilog","score":0.5630000233650208},{"id":"https://openalex.org/keywords/hardware-in-the-loop-simulation","display_name":"hardware-in-the-loop","score":0.39800000190734863},{"id":"https://openalex.org/keywords/energy-efficient-inference","display_name":"energy-efficient inference","score":0.37599998712539673},{"id":"https://openalex.org/keywords/ip-xact","display_name":"IP-XACT","score":0.421999990940094}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5934000015258789},{"id":"https://openalex.org/C68043766","wikidata":"https://www.wikidata.org/wiki/Q267416","display_name":"Static random-access memory","level":2,"score":0.5293999910354614},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.515500009059906},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.5095000267028809},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.4733999967575073},{"id":"https://openalex.org/C77390884","wikidata":"https://www.wikidata.org/wiki/Q217302","display_name":"Application-specific integrated circuit","level":2,"score":0.4449000060558319},{"id":"https://openalex.org/C76752949","wikidata":"https://www.wikidata.org/wiki/Q7607499","display_name":"Stencil","level":2,"score":0.3978999853134155},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.3756999969482422},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.3736000061035156},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.36480000615119934},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.358599990606308},{"id":"https://openalex.org/C46891859","wikidata":"https://www.wikidata.org/wiki/Q1061546","display_name":"Magnetoresistive random-access memory","level":3,"score":0.35569998621940613},{"id":"https://openalex.org/C100800780","wikidata":"https://www.wikidata.org/wiki/Q1175867","display_name":"Memory controller","level":3,"score":0.3495999872684479},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.3441999852657318},{"id":"https://openalex.org/C149810388","wikidata":"https://www.wikidata.org/wiki/Q5374873","display_name":"Emulation","level":2,"score":0.3000999987125397},{"id":"https://openalex.org/C178693496","wikidata":"https://www.wikidata.org/wiki/Q911691","display_name":"Clock rate","level":3,"score":0.27959999442100525},{"id":"https://openalex.org/C46362747","wikidata":"https://www.wikidata.org/wiki/Q173431","display_name":"CMOS","level":2,"score":0.2777999937534332},{"id":"https://openalex.org/C136085584","wikidata":"https://www.wikidata.org/wiki/Q910289","display_name":"Overlay","level":2,"score":0.2702000141143799},{"id":"https://openalex.org/C2776638159","wikidata":"https://www.wikidata.org/wiki/Q18343761","display_name":"Memory cell","level":4,"score":0.267300009727478},{"id":"https://openalex.org/C82687282","wikidata":"https://www.wikidata.org/wiki/Q66221","display_name":"Auxiliary memory","level":2,"score":0.2646999955177307},{"id":"https://openalex.org/C2078106","wikidata":"https://www.wikidata.org/wiki/Q14906620","display_name":"Accumulator (cryptography)","level":2,"score":0.25699999928474426},{"id":"https://openalex.org/C19221803","wikidata":"https://www.wikidata.org/wiki/Q4668679","display_name":"Hydraulic accumulator","level":2,"score":0.25200000405311584},{"id":"https://openalex.org/C131017901","wikidata":"https://www.wikidata.org/wiki/Q170451","display_name":"Logic gate","level":2,"score":0.2515000104904175}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.24071602","doi":"https://doi.org/10.5281/zenodo.24071602","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24071602","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},{"id":"doi:10.5281/zenodo.24071603","doi":"https://doi.org/10.5281/zenodo.24071603","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24071603","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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:10.5281/zenodo.24071602","doi":"https://doi.org/10.5281/zenodo.24071602","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24071602","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7224611519","https://openalex.org/W7056644860","https://openalex.org/W7225006928","https://openalex.org/W63649654","https://openalex.org/W2909730704","https://openalex.org/W2904793694","https://openalex.org/W7161498673","https://openalex.org/W4401500797","https://openalex.org/W2777316529","https://openalex.org/W2948888782"],"abstract_inverted_index":{"Abstract\u2014Reverse-mode":[0],"automatic":[1],"differentiation":[2],"requires":[3],"storing":[4],"gigabytes":[5],"of":[6],"intermediate":[7],"activations":[8],"in":[9,29,311],"off-chip":[10],"DRAM,":[11],"creating":[12],"a":[13,21,87,97,102,110,119,128,134,142,259],"severe":[14],"memory":[15,44],"and":[16,46,75,141,180,417],"energy":[17],"bottleneck.":[18],"We":[19],"present":[20],"fixed-point":[22,94],"SystemVerilog":[23,307],"co-processor":[24],"that":[25],"computes":[26],"analytical":[27],"derivatives":[28],"one":[30],"forward":[31],"pass":[32],"via":[33],"discrete":[34],"Sobolev-space":[35,82],"Friedrichs":[36,84],"mollification,":[37],"achieving":[38],"264.0":[39,154],"ns":[40,155,157,214,225],"FPGA":[41,152,196],"latency,":[42],"91.25%":[43,247],"compression,":[45],"18.4":[47,271],"mW":[48,272],"power.":[49],"Microarchitectural":[50],"&":[51,244,292,375,434],"Hardware":[52,58],"Verification":[53],"Summary":[54],"(Dossier":[55],"Telemetry):":[56],"The":[57],"Memory":[59,243],"Engine":[60],"(omega_mollifier_core,":[61],"Model":[62],"ID:":[63],"OMEGA-HME-Q130,":[64],"packaged":[65],"under":[66,443],"IEEE":[67,76,373,386],"1685-2022":[68],"IP-XACT":[69],"with":[70,96,264],"cryptographic":[71],"SHA3-512":[72],"ipBinderHash:":[73],"4A7F902C0B5E8D1F":[74],"1735-2014":[77],"v2":[78],"encryption)":[79],"projects":[80],"continuous":[81,337],"W_0^{1,p}(Omega)":[83],"mollification":[85],"onto":[86],"synchronous":[88],"8-stage,":[89],"66-cycle":[90],"signed":[91,111],"Q1.30":[92],"(INT32)":[93],"pipeline":[95],"7-tap":[98],"symmetric":[99],"stencil":[100],"ROM,":[101],"1,024-leaf":[103],"parallel":[104],"Wallace":[105],"Multiply-Accumulate":[106],"(MAC)":[107],"reduction":[108],"tree,":[109],"48-bit":[112],"saturating":[113],"accumulator":[114],"(18":[115],"overflow":[116],"guard":[117],"bits),":[118],"16":[120,227],"KB":[121],"on-die":[122],"6T":[123],"circular":[124],"SRAM":[125],"ring":[126],"buffer,":[127],"128-bit":[129],"AMBA":[130],"AXI4-Stream":[131],"router":[132],"(mcp_axi4_router),":[133],"credential-gated":[135],"5-state":[136],"MESI+":[137],"cache":[138],"coherence":[139],"FSM,":[140],"50.0":[143],"ms":[144],"hardware":[145,371],"supervisory":[146],"circuit":[147],"breaker":[148],"(sre_watchdog_shield).":[149],"\u2022":[150,207,242,290,339],"Deterministic":[151],"Latency:":[153],"(\u00b10.4":[156],"jitter,":[158],"66":[159],"clock":[160],"cycles":[161],"at":[162,200,216,426],"250.0":[163],"MHz)":[164],"reported":[165],"across":[166,226],"500":[167],"bare-metal":[168],"hardware-in-the-loop":[169],"(HIL)":[170],"runs":[171],"on":[172],"commercial":[173,370],"AMD":[174],"Virtex":[175],"UltraScale+":[176],"(VU13P,":[177],"Alveo":[178],"UL3524)":[179],"Versal":[181],"(Alveo":[182],"V80LL)":[183],"FPGAs":[184],"(11,420":[185],"LUTs,":[186],"<15%":[187],"fabric":[188],"utilization)\u2014outperforming":[189],"the":[190,363,385],"October":[191],"2026":[192,442],"STAC-ML":[193],"Markets":[194],"(Inference)":[195],"benchmark":[197],"leader":[198],"(XLRA260312":[199],"1,500\u20131,890":[201],"ns)":[202],"by":[203],"5.68x":[204],"to":[205,258,334,384,420],"7.15x.":[206],"Predictive":[208],"5nm":[209],"FinFET":[210,240],"ASIC":[211],"Complexity:":[212],"55.0":[213],"execution":[215],"1.200":[217],"GHz":[218],"(833.3":[219],"ps":[220],"period,":[221],"WNS":[222],"\u2265":[223,331],"+0.000":[224],"PVT":[228],"corners),":[229],"0.42":[230],"mm\u00b2":[231],"macro":[232],"area":[233],"(1,000":[234],"\u00b5m":[235],"\u00d7":[236,327],"420":[237],"\u00b5m),":[238],"48,192,408":[239],"transistors.":[241],"SWaP-C":[245],"Compression:":[246],"host":[248],"VRAM":[249],"compression":[250],"(collapsing":[251],"24.0":[252],"GB":[253,261],"dense":[254],"FP32":[255],"activation":[256,269],"buffering":[257],"2.1":[260],"static":[262],"floor":[263],"0.00":[265],"Bytes":[266],"external":[267],"DRAM":[268],"tape);":[270],"active":[273],"core":[274],"power":[275],"(0.34":[276],"pJ/MAC,":[277],"4.8576":[278],"nJ":[279],"per":[280],"inference,":[281],"+7.2\u00b0C":[282],"steady-state":[283],"thermal":[284],"rise":[285],"vs.":[286],"+42.5\u00b0C":[287],"GPU":[288],"baseline).":[289],"Formal":[291],"Numerical":[293],"Verification:":[294],"512/512":[295],"randomized":[296],"UVM":[297],"1.2":[298],"burst":[299],"tests":[300],"passed":[301],"(98.7%":[302],"functional":[303],"coverage),":[304],"134/134":[305],"formal":[306],"Assertions":[308],"(SVAs)":[309],"verified":[310],"Cadence":[312],"JasperGold,":[313],"bit-exact":[314],"parity":[315],"(expected.npy":[316],"==":[317],"actual.npy,":[318],"SHA-256:":[319],"a99d06c2b5ebff8c03601571efca345137b5568b376eee6d0f4a4ac6fd859e82),":[320],"Mean":[321],"Squared":[322],"Error":[323],"(MSE)":[324],"\u2264":[325],"3.47":[326],"10\u207b\u00b9\u2070,":[328],"Cosine":[329],"Similarity":[330],"0.9999999991":[332],"relative":[333],"IEEE-754":[335],"FP64":[336],"baselines.":[338],"Export":[340],"Control":[341],"Classification:":[342],"BIS":[343],"15":[344],"C.F.R.":[345],"Part":[346],"774":[347],"Supp.":[348],"1":[349],"ECCN":[350],"3A090":[351],"Total":[352],"Processing":[353],"Performance":[354],"TPP":[355,365],"=":[356],"78.6432":[357],"(ASIC)":[358],"/":[359,431],"16.3840":[360],"(FPGA)\u201420.3x":[361],"below":[362],"1,600":[364],"control":[366],"threshold":[367],"(Non-Controlled":[368],"EAR99":[369],"IP).":[372],"Pre-Submission":[374],"Patent":[376,409,413],"Priority":[377],"Notice:":[378],"This":[379],"work":[380],"has":[381],"been":[382],"submitted":[383],"for":[387],"possible":[388],"publication":[389],"(IEEE":[390],"Computer":[391],"Architecture":[392],"Letters).":[393],"Copyright":[394],"may":[395,404],"be":[396,407],"transferred":[397],"without":[398],"notice,":[399],"after":[400],"which":[401],"this":[402],"version":[403],"no":[405],"longer":[406],"accessible.":[408],"Pending:":[410],"U.S.":[411],"Provisional":[412],"Application":[414],"Nos.":[415],"64/127,500":[416],"64/128,324":[418],"(Assigned":[419],"Mavericks":[421],"Alpha":[422],"Signal":[423],"LLC,":[424],"recorded":[425],"USPTO":[427],"EPAS":[428],"Reel":[429],"075667":[430],"Frames":[432],"0383":[433],"0567;":[435],"Foreign":[436],"Filing":[437],"Licenses":[438],"Granted":[439],"Sept.":[440],"23,":[441],"35":[444],"U.S.C.":[445],"\u00a7":[446],"184).":[447],"Master":[448],"Corporate":[449],"Docket:":[450],"TAS-2026-HME-NP03":[451],"|":[452,456],"SAM.gov":[453],"UEI:":[454],"RA2FZ1JUGCK8":[455],"Permanent":[457],"DLA":[458],"CAGE":[459],"Code:":[460],"26U94.":[461]},"counts_by_year":[],"updated_date":"2026-11-11T12:16:26.678958","created_date":"2026-11-10T00:00:00"}
