BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20190719T085744Z
LOCATION:HG F 3
DTSTART;TZID=Europe/Stockholm:20190613T171500
DTEND;TZID=Europe/Stockholm:20190613T174500
UID:submissions.pasc-conference.org_PASC19_sess161_msa328@linklings.com
SUMMARY:Machine-Learning-Based Performance Modeling and Tuning for High-Pe
 rformance Computing
DESCRIPTION:Minisymposium\nComputer Science and Applied Mathematics\n\nMac
 hine-Learning-Based Performance Modeling and Tuning for High-Performance C
 omputing\n\nBalaprakash, Wild\n\nHeterogeneous nodes, many-core processors
 , deep memory hierarchies, energy efficiency demands, and performance vari
 ability make application and system management on high-performance computi
 ng systems an increasingly daunting task. Current strategies provided by o
 perating and runtime systems are mostly static and present several challen
 ges for porting and running applications at extreme-scale. The key challen
 ge consists in finding new proactive and predictive methodologies that wil
 l that support automated refinements of application mapping on extreme-sca
 le systems. In this talk, we will present our work on machine learning app
 roaches for modeling and tuning the performances of compute, communication
 , and I/O subsystems. In particular, we will focus on automated data-drive
 n performance modeling and Bayesian approaches for autotuning search. We w
 ill end the talk with perspectives and research challenges on designing re
 inforcement-learning-based self-improving systems that can observe, predic
 t, and optimize the overall performances of the applications and system au
 tomatically over time.
END:VEVENT
END:VCALENDAR