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:20240626T180033Z
LOCATION:3001\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20240625T170000
DTEND;TZID=America/Los_Angeles:20240625T171500
UID:dac_DAC 2024_sess105_RESEARCH089@linklings.com
SUMMARY:G2PM: Performance Modeling for ACAP Architecture with Dual-Tiered 
 Graph Representation Learning
DESCRIPTION:Research Manuscript\n\nTuo Dai, Bizhao Shi, and Guojie Luo (Pe
 king University)\n\nPerformance estimation is a crucial component in the o
 ptimization processes of accelerator development on the Versal ACAP archit
 ecture. \nHowever, existing approaches present limitations - they are eith
 er too slow to facilitate efficient iterations, or they lack the necessary
  accuracy due to the specific AIE array architecture and two-level program
 ming model of Versal ACAP. \nTo tackle this challenge, we propose G$^2$PM,
  a performance modeling technique based on a hierarchical graph representa
 tion centered on the AIE array.\nMore specifically, we employ a hierarchic
 al graph neural network to identify features of both kernel programs and d
 ataflow programs, taking into account the hardware and software characteri
 stics of the Versal ACAP architecture. \nIn our evaluations, our method de
 monstrates significant improvements, achieving a mean error rate of less t
 han 1.6\% and providing a speed-up factor of 4165$\times$ compared to the 
 simulation-based method.\n\nTopic: AI\n\nKeyword: AI/ML Application and In
 frastructure\n\nSession Chairs: Hongxiang Fan (Imperial College London; Sa
 msung AI Center, UK) and Xiaoxuan Yang (University of Virginia, Stanford U
 niversity)
END:VEVENT
END:VCALENDAR