BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20211208T164413Z
LOCATION:3014
DTSTART;TZID=America/Los_Angeles:20211209T110000
DTEND;TZID=America/Los_Angeles:20211209T113000
UID:dac_DAC 2021_sess159_RESEARCH1055@linklings.com
SUMMARY:Evolved Neural-Hardware-Compiler Co-Design
DESCRIPTION:Research Manuscript\n\nEvolved Neural-Hardware-Compiler Co-Des
 ign\n\nLin\n\nHardware architecture search is an efficient tool to enable 
 specialized accelerator designs for neural networks. Current search framew
 orks focus on sizing the architectural hyper-parameters while neglect sear
 ching the PE connectivity and compiler mappings.\nTo tackle this challenge
 , we propose ENHCS that holistically searches the neural/hardware architec
 tures and compiler mapping in one optimization loop. ENHCS composes highly
  matched architectures together with efficient mapping for given neural ne
 tworks. With the same computation resources, EHSS can rival the human desi
 gn Eyeriss by 4.4X EDP reduction with 2.7% accuracy improvement on ImageNe
 t and offers 3.5X EDP reduction than only searching the architectural sizi
 ng.\n\nTopic: Design\n\nKeyword: AI/ML System Design\n\nVirtual Program: P
 resented In-Person
END:VEVENT
END:VCALENDAR