BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240116T191659Z
LOCATION:603
DTSTART;TZID=America/Denver:20231112T093100
DTEND;TZID=America/Denver:20231112T094600
UID:submissions.supercomputing.org_SC23_sess420_ws_ss109@linklings.com
SUMMARY:Comparing Power Signatures of HPC Workloads: Machine Learning vs S
 imulation
DESCRIPTION:Workshop\n\nAnish Govind (University of California, San Diego 
 (UCSD)) and Sridutt Bhalachandra, Zhengji Zhao, Ermal Rrapaj, Brian Austin
 , and Hai Ah Nam (Lawrence Berkeley National Laboratory (LBNL))\n\nPower i
 s a limiting factor for supercomputers limiting their scale and operation.
  Characterizing the power signatures of different application types can en
 able data centers to operate efficiently, even when power constrained. Thi
 s paper investigates power profiles of diverse scientific applications, sp
 anning both traditional simulations and modern machine learning (ML) runni
 ng on the Perlmutter supercomputer at the National Energy Research Scienti
 fic Computing Center (NERSC). Our findings indicate that traditional simul
 ations typically consume more power on average than ML workloads. Furtherm
 ore, ML applications exhibit periodic power fluctuations attributed to epo
 ch transitions during training. Finally, we discuss the potential implicat
 ions of the research insights toward automatic demand response (ADR) and c
 onsiderations for designing future systems.\n\nTag: Energy Efficiency, Gre
 en Computing, Sustainability\n\nRegistration Category: Workshop Reg Pass\n
 \nSession Chairs: Kimmo Koski (CSC – IT Center for Science Ltd, Finland); 
 James H. Rogers (Oak Ridge National Laboratory (ORNL)); Fumiyoshi Shoji (R
 IKEN, Center for Computational Science); William W. Thigpen (NASA); Michèl
 e Weiland (University of Edinburgh, EPCC); and Mike Woodacre (Hewlett Pack
 ard Enterprise (HPE))
END:VEVENT
END:VCALENDAR