BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260422T143140Z
LOCATION:B309
DTSTART;TZID=America/New_York:20241119T140000
DTEND;TZID=America/New_York:20241119T143000
UID:submissions.supercomputing.org_SC24_sess403_pap462@linklings.com
SUMMARY:Moirae: Generating High-Performance Composite Stencil Programs wit
 h Global Optimizations
DESCRIPTION:Xiaoyan Liu, Xinyu Yang, Kejie Ma, Shanghao Liu, Kaige Zhang, 
 Hailong Yang, Yi Liu, Zhongzhi Luan, and Depei Qian (Beihang University)\n
 \nStencil computation is one of the most universal computation motifs in s
 cientific applications such as weather prediction. Due to the complexity o
 f scientific simulation, the stencil computation can contain a set of comp
 lex stencil operations that form a directed acyclic graph (referred to com
 posite stencil). Unfortunately, most existing stencil optimizations and co
 mpilers only focus on intra-stencil operation, and cannot fully explore th
 e performance improvement potential of composite stencils in nowadays appl
 ications. To this end, we propose Moirae, a framework that explores a nove
 l optimization space and generates high-performance code for composite ste
 ncils. We first propose a lightweight cost model with a fine-grained analy
 sis of memory access behavior to predict the performance. Based on the cos
 t model, we propose an evolutionary search method to find a high-performan
 ce optimization, leveraging a search space pruning method with stencil dom
 ain knowledge. Experimental results show that Moirae can outperform the st
 ate-of-the-art composite stencil compilers.\n\nTag: Accelerators, Compiler
 s, Embedded and/or Reconfigurable Systems, Linear Algebra, Performance Eva
 luation and/or Optimization Tools\n\nRegistration Category: Tech Program R
 eg Pass\n\nSession Chair: Keita Teranishi (Oak Ridge National Laboratory (
 ORNL))\n\n
END:VEVENT
END:VCALENDAR
