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:20210916T132455Z
LOCATION:Mère Royaume
DTSTART;TZID=Europe/Stockholm:20210709T150000
DTEND;TZID=Europe/Stockholm:20210709T153000
UID:submissions.pasc-conference.org_PASC21_sess115_msa126@linklings.com
SUMMARY:Stateful Dataflow Multigraphs: A Data-Centric Model for Performanc
 e Portability on Heterogeneous Architectures
DESCRIPTION:Minisymposium\n\nStateful Dataflow Multigraphs: A Data-Centric
  Model for Performance Portability on Heterogeneous Architectures\n\nBen-N
 un\n\nThe ubiquity of accelerators in high-performance computing has drive
 n programming complexity beyond the skill-set of the average domain scient
 ist. To maintain performance portability in the future, it is imperative t
 o decouple architecture-specific programming paradigms from the underlying
  scientific computations. The talk discusses the Stateful DataFlow multiGr
 aph (SDFG), a data-centric intermediate representation that enables separa
 ting program definition from its optimization. By combining fine-grained d
 ata dependencies with high-level control-flow, SDFGs are both expressive a
 nd amenable to program transformations, such as tiling and double-bufferin
 g. These transformations are applied to the SDFG in an interactive process
 , using extensible pattern matching, graph rewriting, and a graphical user
  interface. We demonstrate SDFGs on CPUs, GPUs, and FPGAs over various mot
 ifs - from fundamental computational kernels to graph analytics. We show t
 hat SDFGs deliver competitive performance, allowing domain scientists to d
 evelop applications naturally and port them to approach peak hardware perf
 ormance without modifying the original scientific code.\n\nDomain: CS and 
 Math, Climate and Weather, Engineering
END:VEVENT
END:VCALENDAR
