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:20200129T163557Z
LOCATION:708
DTSTART;TZID=America/Denver:20191117T143000
DTEND;TZID=America/Denver:20191117T150000
UID:submissions.supercomputing.org_SC19_sess112_pec248@linklings.com
SUMMARY:Comparing GPU Power and Frequency Capping: A Case Study with the M
 uMMI Workflow
DESCRIPTION:Workshop\n\nComparing GPU Power and Frequency Capping: A Case 
 Study with the MuMMI Workflow\n\nPatki, Frye, Bhatia, Di Natale, Glosli...
 \n\nAccomplishing the goal of exascale computing under a potential power l
 imit requires HPC clusters to maximize both parallel efficiency and power 
 efficiency. As modern HPC systems embark on a trend toward extreme heterog
 eneity leveraging multiple GPUs per node, power management becomes even mo
 re challenging, especially when catering to scientific workflows with co-s
 cheduled components. The impact of managing GPU power on workflow performa
 nce and run-to-run reproducibility has not been adequately studied. \n\nIn
  this paper, we present a first-of-its-kind research to study the impact o
 f the two power management knobs that are available on NVIDIA Volta GPUs: 
 frequency capping and power capping. We analyzed performance and power met
 rics of GPU’s on a top-10 supercomputer by tuning these knobs for more tha
 n 5,300 runs in a scientific workflow. Our data found that GPU power cappi
 ng in a scientific workflow is an effective way of improving power efficie
 ncy while preserving performance, while GPU frequency capping is a demonst
 rably unpredictable way of reducing power consumption. Additionally, we id
 entified that frequency capping results in higher variation and anomalous 
 behavior on GPUs, which is counterintuitive to what has been observed in t
 he research conducted on CPUs.\n\nTag: Workshop Reg Pass, Extreme Scale Co
 mputing, Scalable Computing, Scientific Workflows\n\nRegistration Category
 : Workshop Reg Pass, Extreme Scale Computing, Scalable Computing, Scientif
 ic Workflows
URL:https://sc19.supercomputing.org/presentation/?id=pec248&sess=sess112
END:VEVENT
END:VCALENDAR