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:20190719T085745Z
LOCATION:HG F 1
DTSTART;TZID=Europe/Stockholm:20190614T113000
DTEND;TZID=Europe/Stockholm:20190614T120000
UID:submissions.pasc-conference.org_PASC19_sess170_msa184@linklings.com
SUMMARY:Understanding and Optimizing Data-Movement in HPC Applications on 
 Heterogeneous Memory Systems
DESCRIPTION:Minisymposium\nComputer Science and Applied Mathematics, Physi
 cs\n\nUnderstanding and Optimizing Data-Movement in HPC Applications on He
 terogeneous Memory Systems\n\nMarkidis, Olshevsky, Chien, Rivas-Gomez, Pen
 g...\n\nCurrent top supercomputers feature a wide spectrum of diverse memo
 ry technologies, ranging from traditional DRAM to high bandwidth and on no
 de non-volatile memories. The diversity of memory technologies is expected
  to further increase in next generation supercomputers. These different me
 mory technologies differ by capacity, latency, bandwidth and might have hi
 gher performance in load than in store operations. Considering all these i
 ssues, how to map different application memory objects to different memori
 es for maximum performance is a daunting task. In this talk, I present the
  current effort in the EC-funded EPiGRAM-HS (Exascale ProGRAmming Models f
 or Heterogeneous Memories) project to address the challenge of high perfor
 mance data placement and movement in large-scale legacy applications. Taki
 ng one of the project pilot applications as example, the iPIC3D code [1], 
 I first describe the characterization of memory access pattern and size of
  different memory objects. I then discuss different mapping strategies and
  work in EPiGRAM-HS to simplify and automatize data placement and movement
  in large-scale HPC applications.<br />[1] Markidis, Stefano, Giovann
 i Lapenta and Rizwan-uddin. "Multi-scale simulations of plasma with iPIC3D
 ." <em>Mathematics and Computers in Simulation</em> 80.7 (2010):
  1509-1519.
END:VEVENT
END:VCALENDAR