{"685258":{"#nid":"685258","#data":{"type":"event","title":"PhD Proposal by Rafael Oliveira  ","body":[{"value":"\u003Cp\u003E\u003Cstrong\u003ETitle: End-to-End Support for In-Network Acceleration of Dynamically Chained Parametric Tasks\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ERafael Oliveira \u0026nbsp;\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003EPh.D. Student in Computer Science \u0026nbsp;\u003C\/p\u003E\u003Cp\u003ESchool of Computer Science \u0026nbsp;\u003C\/p\u003E\u003Cp\u003ECollege of Computing \u0026nbsp;\u003C\/p\u003E\u003Cp\u003EGeorgia Institute of Technology \u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003EDate: Tuesday, September 30th, 2025 \u0026nbsp;\u003C\/p\u003E\u003Cp\u003ETime: 3:00pm - 4:30pm ET \u0026nbsp;\u003C\/p\u003E\u003Cp\u003ELocation: KACB 3100, or\u003Ca href=\u0022https:\/\/gatech.zoom.us\/j\/98975861466?pwd=YmlMWVU2V1hqL01rbE1KeVZhYnVVQT09\u0022 title=\u0022https:\/\/gatech.zoom.us\/j\/98975861466?pwd=YmlMWVU2V1hqL01rbE1KeVZhYnVVQT09\u0022\u003E\u0026nbsp;Zoom\u003C\/a\u003E\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ECommittee: \u0026nbsp;\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003EDr. Ada Gavrilovska \u0026nbsp;(advisor)\u0026nbsp;\u2013 College of Computing, Georgia Institute of Technology \u0026nbsp;\u003C\/p\u003E\u003Cp\u003EDr. Calton Pu\u0026nbsp;\u2013 College of Computing, Georgia Institute of Technology \u0026nbsp;\u003C\/p\u003E\u003Cp\u003EDr. Santosh Pande \u2013 College of Computing, Georgia Institute of Technology \u0026nbsp;\u003C\/p\u003E\u003Cp\u003EDr. Alex Daglis \u2013 College of Computing, Georgia Institute of Technology \u0026nbsp;\u003C\/p\u003E\u003Cp\u003EDr. Hadi Esmaeilzadeh \u2013 Computer Science and Engineering, UCSD \u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EAbstract:\u003C\/strong\u003E Modern cloud infrastructure faces a fundamental inefficiency: while modern software scales through shared resources\u2014from dynamically linked libraries to containerized microservices\u2014SmartNICs allocate dedicated hardware accelerators per tenant,\u0026nbsp;evoking the pre-DLL era where common functions were duplicated in every program. This wastes precious hardware resources in multi-tenant environments where applications often need the same computations (AES encryption, Kalman filters) with different parameters.\u003C\/p\u003E\u003Cp\u003EThis thesis introduces parametric Compute Unit (CU) sharing\u2014enabling multiple tenants to share hardware accelerators just as software shares libraries. The key challenge is managing parameters at line rate: unlike software\u0027s flexible memory, hardware requires deterministic, cycle-accurate parameter delivery across dynamically composed processing chains.\u003C\/p\u003E\u003Cp\u003EWe present three systems that collectively solve this challenge: Vazado\u0026nbsp;enables dynamic chaining of CUs with incompatible signatures through hardware-compiler co-design, achieving 13.19\u00d7 energy efficiency over 48-core CPUs while using just 10% of FPGA resources. ErdTree\u0026nbsp;introduces hierarchical topic trees that reduce memory footprints by 52% through automatic parameter deduplication and hardware-friendly FIFO caching. COMPREX\u0026nbsp;accelerates NIC-CPU transfers 7\u00d7 through domain-specific compression.\u003C\/p\u003E\u003Cp\u003ETogether, these systems demonstrate that efficient multi-tenant acceleration is achievable without sacrificing programmability\u2014enabling SmartNICs to finally deliver on their promise of scalable, shared acceleration for cloud infrastructure.\u003C\/p\u003E","summary":"","format":"limited_html"}],"field_subtitle":"","field_summary":[{"value":"\u003Cp\u003E\u003Cstrong\u003EEnd-to-End Support for In-Network Acceleration of Dynamically Chained Parametric Tasks\u003C\/strong\u003E\u003C\/p\u003E","format":"limited_html"}],"field_summary_sentence":[{"value":"End-to-End Support for In-Network Acceleration of Dynamically Chained Parametric Tasks"}],"uid":"27707","created_gmt":"2025-09-25 11:55:32","changed_gmt":"2025-09-25 11:56:28","author":"Tatianna Richardson","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2025-09-30T15:00:00-04:00","event_time_end":"2025-09-30T16:30:00-04:00","event_time_end_last":"2025-09-30T16:30:00-04:00","gmt_time_start":"2025-09-30 19:00:00","gmt_time_end":"2025-09-30 20:30:00","gmt_time_end_last":"2025-09-30 20:30:00","rrule":null,"timezone":"America\/New_York"},"location":" KACB 3100, or Zoom","extras":[],"groups":[{"id":"221981","name":"Graduate Studies"}],"categories":[],"keywords":[{"id":"102851","name":"Phd proposal"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1788","name":"Other\/Miscellaneous"}],"invited_audience":[{"id":"78771","name":"Public"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}