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:20250626T234542Z
LOCATION:B314
DTSTART;TZID=America/New_York:20241117T153000
DTEND;TZID=America/New_York:20241117T160000
UID:submissions.supercomputing.org_SC24_sess741_ws_mlg105@linklings.com
SUMMARY:IRIS-GNN: Leveraging Graph Neural Networks for Scheduling on Truly
  Heterogeneous Runtime Systems
DESCRIPTION:Beau Johnston (Oak Ridge National Laboratory (ORNL)); Thibault
  de Boissiere (Canva Incorporated); and Mohammad Alaul Haque Monil, Narasi
 nga Rao Miniskar, Aaron Young, Seyong Lee, and Jeffrey S. Vetter (Oak Ridg
 e National Laboratory (ORNL))\n\nThe diversity of accelerators in computer
  systems poses significant challenges for software developers, such as man
 aging vendor-specific compiler toolchains, code fragmentation requiring di
 fferent kernel implementations, and performance portability issues. To add
 ress these, the Intelligent Runtime System (IRIS) was developed. IRIS work
 s across various systems, from smartphones to supercomputers, enabling aut
 omatic performance scaling based on available accelerators. Although IRIS 
 simplifies system details, optimal dynamic scheduling still requires user 
 input to understand workload structures. To address this, we introduce a n
 ew scheduling policy for IRIS, termed IRIS-GNN, which is the first IRIS hy
 brid policy that operates in conjunction with the dynamic policies. This p
 olicy employs a Graph-Neural Network (GNN) to conduct Graph Classification
  of any task graphs submitted to IRIS. This GNN analyzes the structure and
  attributes of the task graph, categorizing it as either locality, concurr
 ency, or mixed. This classification subsequently guides the selection of t
 he dynamic policy used by IRIS.\n\nTag: Artificial Intelligence/Machine Le
 arning, Graph Algorithms, Scalable Data Mining\n\nRegistration Category: W
 orkshop Reg Pass\n\nSession Chairs: Seung-Hwan Lim (Oak Ridge National Lab
 oratory (ORNL)), José Moreira (IBM), Catherine Schuman (University of Tenn
 essee), and Richard Vuduc (Georgia Institute of Technology)\n\n
END:VEVENT
END:VCALENDAR