[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-32157":3,"doc-seo-32157":27},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"file_id":15,"file_url":16,"file_type":17,"file_size":18,"view_count":11,"is_deleted":4,"is_public":19,"is_downloadable":19,"audit_status":19,"page_count":20,"language":21,"language_code":22,"table_of_contents":23,"faqs":24,"seo_title":13,"seo_description":14,"update_tm":25,"read_time":26},32157,3848291630094,"Emma Wilson","https://eur-avatar.wpscdn.com/davatar_085a072bc5b1113ac321206ff7593b45",8,"Research & Report","Estimation of Energy-dissipation Lower-bounds for Neuromorphic Learning-in-memory","Neuromorphic optimizers using local, parallel updates are constrained by three energy bottlenecks: memory read access in compute-in-memory, precision-driven memory writes in learning-in-memory, and data movement between short-term and long-term memories during consolidation. This work derives theoretical estimates of the energy-to-solution lower bound for an ideal optimizer that matches memory update and consolidation dynamics to optimization/annealing dynamics. The resulting model-agnostic energy-efficiency bounds depend on update operations, parameter count, convergence speed, and solution precision, and are applied to large-scale AI workloads.","cbCaivbacOILY0v8","https://ap.wps.com/l/cbCaivbacOILY0v8","pdf",3067666,1,17,"English","en","# Introduction\n## Memory and energy bottlenecks in AI training\n## From CIM to learning-in-memory: addressing update and consolidation walls\n# Energy-to-solution lower-bound estimation","[{\"question\":\"What energy bottlenecks motivate learning-in-memory neuromorphic optimization?\",\"answer\":\"The paper frames energy limits around three performance walls: memory-wall from repeated memory read access, update-wall from repeated memory writes at optimization precision, and consolidation-wall from repeated transfers between short- and long-term memory.\"},{\"question\":\"How does the ideal neuromorphic optimizer in the paper achieve tighter energy-efficiency estimates?\",\"answer\":\"It modulates the energy barrier of physical memories so that memory update and memory consolidation dynamics align with the optimization or annealing dynamics of learning.\"},{\"question\":\"What quantities determine the derived energy-to-solution lower bounds?\",\"answer\":\"The estimates are model-agnostic and depend on the number of model-update operations (OPS), model size in parameter count, speed of convergence, and the precision required for the solution.\"}]",1780952469,43,{"code":4,"msg":28,"data":29},"ok",{"site_id":30,"language":22,"slug":31,"title":13,"keywords":32,"description":14,"schema_data":33,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":25},105,"estimation-of-energy-dissipation-lower-bounds-for-neuromorphic-learning-in-memory","",{"@graph":34,"@context":84},[35,52,67],{"@type":36,"itemListElement":37},"BreadcrumbList",[38,42,46,49],{"item":39,"name":40,"@type":41,"position":19},"https://docshare.wps.com","Home","ListItem",{"item":43,"name":44,"@type":41,"position":45},"https://docshare.wps.com/document/","Document",2,{"item":47,"name":12,"@type":41,"position":48},"https://docshare.wps.com/document/research-report/",3,{"item":50,"name":13,"@type":41,"position":51},"https://docshare.wps.com/document/estimation-of-energy-dissipation-lower-bounds-for-neuromorphic-learning-in-memory/32157/",4,{"url":50,"name":13,"@type":53,"author":54,"headline":13,"publisher":56,"fileFormat":59,"description":14,"dateModified":60,"datePublished":61,"encodingFormat":59,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":55},"Person",{"url":39,"name":57,"@type":58},"DocShare","Organization","application/pdf","2026-06-13","2026-06-08",true,{"@type":64,"interactionType":65,"userInteractionCount":11},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What energy bottlenecks motivate learning-in-memory neuromorphic optimization?","Question",{"text":74,"@type":75},"The paper frames energy limits around three performance walls: memory-wall from repeated memory read access, update-wall from repeated memory writes at optimization precision, and consolidation-wall from repeated transfers between short- and long-term memory.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does the ideal neuromorphic optimizer in the paper achieve tighter energy-efficiency estimates?",{"text":79,"@type":75},"It modulates the energy barrier of physical memories so that memory update and memory consolidation dynamics align with the optimization or annealing dynamics of learning.",{"name":81,"@type":72,"acceptedAnswer":82},"What quantities determine the derived energy-to-solution lower bounds?",{"text":83,"@type":75},"The estimates are model-agnostic and depend on the number of model-update operations (OPS), model size in parameter count, speed of convergence, and the precision required for the solution.","https://schema.org",{"og:url":50,"og:type":86,"og:title":13,"og:site_name":57,"og:description":14},"article",{"robots":88,"canonical":50},"index,follow",{"doc_id":7,"site_id":30}]