BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Eventi DIAG
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20171029T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RDATE:20181028T030000
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20180325T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.13018.field_data.0@oba.diag.uniroma1.it
DTSTAMP:20260408T053058Z
CREATED:20180221T142436Z
DESCRIPTION:High performance computing (HPC) resources and workloads are un
 dergoing tumultuous changes. HPC resources are growing more diverse with t
 he adoption of accelerators\; HPC workloads have increased in size by orde
 rs of magnitude. Despite these changes\, when assigning workload jobs to r
 esources\, HPC schedulers still rely on users to accurately anticipate the
 ir applications’ resource usage and remain stuck with the decades-old cent
 ralized scheduling model.In this talk we will discuss these ongoing change
 s and propose alternative models for HPC scheduling based on resource awar
 eness and fully hierarchical models. A key role in our models’ evaluation 
 is played by an emulator of a real open-source\, next-generation resource 
 management system. We will discuss the challenges of realistically mimicki
 ng the system's scheduling behavior. Our evaluation shows how our models i
 mprove scheduling scalability on a diverse set of synthetic and real-world
  workloads.This is joint work with Stephen Herbein and Michael Wyatt  at t
 he University of Delaware\, and Tapasya Patki\, Dong H. Ahn\, Don Lipari\,
  Thomas R.W. Scogland\, Marc Stearman\, Mark Grondona\, Jim Garlick\, Tama
 ra Dahlgren\, David Domyancic\, and Becky Springmeyer at the Lawrence Live
 rmore National Laboratory.Short vita:Michela Taufer is a Professor in Comp
 uter and Information Sciences and a J.P. Morgan Case Scholar at the Univer
 sity of Delaware\; she has a joint appointment in the Biomedical Departmen
 t and the Bioinformatics Program at the same university. She earned her un
 dergraduate degrees in Computer Engineering from the University of Padova 
 (Italy) and her doctoral degree in Computer Science from the Swiss Federal
  Institute of Technology or ETH (Switzerland). From 2003 to 2004 she was a
  La Jolla Interfaces in Science Training Program (LJIS) Postdoctoral Fello
 w at the University of California San Diego (UCSD) and The Scripps Researc
 h Institute (TSRI)\, where she worked on interdisciplinary projects in com
 puter systems and computational chemistry. Taufer’s research interests in 
 high performance computing include scientific applications\, scheduling an
 d reproducibility challenges\, and big data analytics. She has nearly 100 
 publications and delivered nearly 80 talks at various conferences and rese
 arch institutes. She is currently serving on the NSF Advisory Committee fo
 r Cyberinfrastructures (ACCI). She is a professional member of the IEEE an
 d a Distinguished Scientist of the ACM.
DTSTART;TZID=Europe/Paris:20180524T090000
DTEND;TZID=Europe/Paris:20180524T090000
LAST-MODIFIED:20191008T082902Z
LOCATION:Aula Magna
SUMMARY:Modeling the Next-Generation High Performance Schedulers - Michela 
 Taufer
URL;TYPE=URI:http://oba.diag.uniroma1.it/node/13018
END:VEVENT
END:VCALENDAR