BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201229Z
LOCATION:Second Floor Atrium
DTSTART;TZID=America/Chicago:20251118T080000
DTEND;TZID=America/Chicago:20251118T170000
UID:submissions.supercomputing.org_SC25_sess537_drs105@linklings.com
SUMMARY:Heterogeneous HPC Compute Continuum: A Roadmap for Workflow Mappin
 g and Scheduling From Sensor to Supercomputer
DESCRIPTION:Aasish Sharma (Georg-August-Universität Göttingen, Gesellschaf
 t für wissenschaftliche Datenverarbeitung mbH Göttingen)\n\nEfficient work
 load mapping and scheduling in heterogeneous HPC environments connecting f
 rom IoT, edge devices to cloud is essential for optimizing resource use, r
 educing makespan, and ensuring adaptability. This research explores advanc
 ed solutions addressing mapping and scheduling by investigating the gaps i
 n surveying the available tools and techniques that include classical opti
 mization methods, emerging AI-driven models, and hybrid quantum-inspired a
 pproaches.\n\nFor workflow-based workload mapping and scheduling, the stud
 y employs a proper system and workload modeling and evaluates mixed-intege
 r linear programming (MILP) for optimal assignment in smaller scenarios. I
 n larger environments, a graph neural networks and reinforcement learning 
 (GNN-RL) framework scales efficiently by learning adaptive policies reflec
 ting task dependencies and system characteristics. \n\nFor task-based work
 load mapping and scheduling, outlined integrated AI scheduler (IAIS) frame
 work dynamically manages resources in distributed, cloud, and HPC environm
 ents. IAIS combines recurrent neural networks (RNNs) and temporal convolut
 ional networks (TCNs) for predicting optimal task allocation. Enhanced wit
 h proximal policy optimization (PPO)-based reinforcement learning, IAIS ef
 fectively predicts throughput, minimizes latency, and maximizes resource u
 tilization. Complementary machine-learning models (e.g., simpler RNNs) fur
 ther expedite allocation of independent tasks, notably in cloud contexts.\
 n\nComparative evaluations indicate notable tools and techniques for optim
 ization performance, scalability, and resource efficiency applying IAIS, M
 ILP, and GNN-RL. Specifically, IAIS and GNN-RL demonstrate strong adaptabi
 lity and scalability within heterogeneous compute continuum environments, 
 laying the groundwork for future cognitive scheduling assistants capable o
 f real-time autonomous optimization.\n\nTag: Research & ACM SRC Posters\n\
 nRecording: Not Livestreamed, Not Recorded\n\nRegistration Category: Techn
 ical Program Reg Pass\n\nSession Chairs: Kento Sato (RIKEN Center for Comp
 utational Science (R-CCS)); Chris Schlipalius (Pawsey Supercomputing Resea
 rch Centre; Commonwealth Scientific and Industrial Research Organisation (
 CSIRO), Australia); and Anja Gerbes (Georg-August-Universität Göttingen)\n
 \n
END:VEVENT
END:VCALENDAR