BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Sabre//Sabre VObject 4.6.1//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Europe/Zurich
X-LIC-LOCATION:Europe/Zurich
TZURL:http://tzurl.org/zoneinfo/Europe/Zurich
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19810329T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19961027T030000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:news202@dmi.unibas.ch
DTSTAMP;TZID=Europe/Zurich:20181113T175225
DTSTART;TZID=Europe/Zurich:20180621T100000
SUMMARY:Computer Science Colloquium: Prof. Allen D. Malony
DESCRIPTION:Abstract: Since the beginning of ``high-performance'' parallel 
 computing\, observingand analyzing performance for purposes of finding bot
 tlenecks andidentifying opportunities for improvement has been at the hear
 t ofdelivering the performance potential of next-generation scalable syste
 ms.Interestingly\, it is the ever-changing parallel computing landscape th
 at isthe main driver of requirements for parallel performance technology a
 nd theimprovements necessary beyond the current state-of-the-art.  Indeed
 \, thedevelopment and application of our TAU Performance System over many 
 yearslargely follows an evolutionary path of addressing measurement and an
 alysisproblems in new parallel machines and programming environments.Howev
 er\, the outlook to future parallel systems with high degrees ofconcurrenc
 y\, heterogeneous components\, dynamic runtime environments\,asynchronous 
 execution\, and power constraints suggests a new perspectivewill be needed
  on the role of performance observation and analysis inrespect to tool tec
 hnology integration and performance optimizationmethods.  The reliance on
  post-mortem analysis of application-level("1st person") performance measu
 rements is prohibitive forexascale-class machines because of the performan
 ce data volume\, theprimitive basis for performance data attribution\, and
  the fundamentalproblem of performance variation that will exist.  Instea
 d\, it will beimportant to provide introspection support across the exasca
 le softwarestack to understand how system ("3rd person") resources are use
 d duringexecution.  Furthermore\, the opportunity to couple a global perf
 ormanceintrospection capability (a "performance backplane") with onlineper
 formance decision analytics inspires the concept of an autonomicperformanc
 e system that can feed back policy-based decisions to guidethe computation
  to better states of execution.The talk will explore these issues by givin
 g a brief retrospective onperformance tool evolution\, setting the stage f
 or current researchprojects where a new performance perspective is being p
 ursued.  It willalso speculate on what might be included in next-generati
 on parallelsystems hardware\, specifically to make the exascale machines m
 oreperformance-aware and dynamically-adaptive. Speaker Bio: Allen D. Malon
 y is a Professor in the Department of Computer andInformation Science at t
 he University of Oregon (UO) where he directsparallel computing research p
 rojects\, notably the TAU parallelperformance system project.  He has ext
 ensive experience in performancebenchmarking and characterization of high-
 performance computing systems\,and has developed performance evaluation to
 ols for a range of parallelmachines during the last 30 years.  Malony is 
 also interested incomputational and data science.  He is the Director of 
 the new OregonAdvanced Computing Institute for Science and Society (OACISS
 ) at UO.Malony was awarded the NSF National Young Investigator award\, was
  aFulbright Research Scholar to The Netherlands and Austria\, and received
 the Alexander von Humboldt Research Award for Senior U.S. Scientists bythe
  Alexander von Humboldt Foundation.  Last year he was theFulbright-Tocque
 ville Distinguished Chair to France.  Recently\, he wasaward the Fulbrigh
 t for the Future by the Franco-Americaine FulbrightCommission.  Malony is
  the CEO of ParaTools\, Inc.\, which he founded withDr. Sameer Shende in 2
 004.  ParaTools SAS is a French company theystarted in 2014\, and is whol
 ly-owned company by ParaTools\, Inc.ParaTools specializes in performance a
 nalysis and engineering\, HPCapplications and optimization\, and parallel 
 software\, hardware\, andtools.
X-ALT-DESC:\n<b>Abstract:</b> Since the beginning of ``high-performance'' p
 arallel computing\, observing<br />and analyzing performance for purposes 
 of finding bottlenecks and<br />identifying opportunities for improvement 
 has been at the heart of<br />delivering the performance potential of next
 -generation scalable systems.<br />Interestingly\, it is the ever-changing
  parallel computing landscape that is<br />the main driver of requirements
  for parallel performance technology and the<br />improvements necessary b
 eyond the current state-of-the-art.&nbsp\; Indeed\, the<br />development a
 nd application of our TAU Performance System over many years<br />largely 
 follows an evolutionary path of addressing measurement and analysis<br />p
 roblems in new parallel machines and programming environments.<br /><br />
 However\, the outlook to future parallel systems with high degrees of<br /
 >concurrency\, heterogeneous components\, dynamic runtime environments\,<b
 r />asynchronous execution\, and power constraints suggests a new perspect
 ive<br />will be needed on the role of performance observation and analysi
 s in<br />respect to tool technology integration and performance optimizat
 ion<br />methods.&nbsp\; The reliance on post-mortem analysis of applicati
 on-level<br />(&quot\;1st person&quot\;) performance measurements is prohi
 bitive for<br />exascale-class machines because of the performance data vo
 lume\, the<br />primitive basis for performance data attribution\, and the
  fundamental<br />problem of performance variation that will exist.&nbsp\;
  Instead\, it will be<br />important to provide introspection support acro
 ss the exascale software<br />stack to understand how system (&quot\;3rd p
 erson&quot\;) resources are used during<br />execution.&nbsp\; Furthermore
 \, the opportunity to couple a global performance<br />introspection capab
 ility (a &quot\;performance backplane&quot\;) with online<br />performance
  decision analytics inspires the concept of an autonomic<br />performance 
 system that can feed back policy-based decisions to guide<br />the computa
 tion to better states of execution.<br /><br />The talk will explore these
  issues by giving a brief retrospective on<br />performance tool evolution
 \, setting the stage for current research<br />projects where a new perfor
 mance perspective is being pursued.&nbsp\; It will<br />also speculate on 
 what might be included in next-generation parallel<br />systems hardware\,
  specifically to make the exascale machines more<br />performance-aware an
 d dynamically-adaptive. <br /><br /><b>Speaker Bio: </b>Allen D. Malony is
  a Professor in the Department of Computer and<br />Information Science at
  the University of Oregon (UO) where he directs<br />parallel computing re
 search projects\, notably the TAU parallel<br />performance system project
 .&nbsp\; He has extensive experience in performance<br />benchmarking and 
 characterization of high-performance computing systems\,<br />and has deve
 loped performance evaluation tools for a range of parallel<br />machines d
 uring the last 30 years.&nbsp\; Malony is also interested in<br />computat
 ional and data science.&nbsp\; He is the Director of the new Oregon<br />A
 dvanced Computing Institute for Science and Society (OACISS) at UO.<br />M
 alony was awarded the NSF National Young Investigator award\, was a<br />F
 ulbright Research Scholar to The Netherlands and Austria\, and received<br
  />the Alexander von Humboldt Research Award for Senior U.S. Scientists by
 <br />the Alexander von Humboldt Foundation.&nbsp\; Last year he was the<b
 r />Fulbright-Tocqueville Distinguished Chair to France.&nbsp\; Recently\,
  he was<br />award the Fulbright for the Future by the Franco-Americaine F
 ulbright<br />Commission.&nbsp\; Malony is the CEO of ParaTools\, Inc.\, w
 hich he founded with<br />Dr. Sameer Shende in 2004.&nbsp\; ParaTools SAS 
 is a French company they<br />started in 2014\, and is wholly-owned compan
 y by ParaTools\, Inc.<br />ParaTools specializes in performance analysis a
 nd engineering\, HPC<br />applications and optimization\, and parallel sof
 tware\, hardware\, and<br />tools.<br /><br />
DTEND;TZID=Europe/Zurich:20180621T120000
END:VEVENT
END:VCALENDAR
