BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230124T171527Z
LOCATION:C147-154
DTSTART;TZID=America/Chicago:20221114T105200
DTEND;TZID=America/Chicago:20221114T111400
UID:submissions.supercomputing.org_SC22_sess452_ws_pawatm104@linklings.com
SUMMARY:Task Fusion in Distributed Runtimes
DESCRIPTION:Workshop\n\nTask Fusion in Distributed Runtimes\n\nSundram, Ai
 ken, Lee\n\nWe present distributed task fusion, a run-time optimization fo
 r task-based runtimes operating on parallel and heterogeneous systems.  Di
 stributed task fusion dynamically performs an efficient buffering, analysi
 s, and fusion of asynchronously-evaluated distributed operations, reducing
  the overheads inherent to scheduling distributed tasks in implicitly para
 llel frameworks and runtimes. We identify the constraints under which dist
 ributed task fusion is permissible and describe an implementation in Legat
 e, a domain-agnostic library for constructing portable and scalable task-b
 ased libraries.  We present performance results using cuNumeric, a Legate 
 library that enables scalable execution of NumPy pipelines on parallel and
  heterogeneous systems.  We realize speedups up to 1.5x with task fusion e
 nabled on up to 32 P100 GPUs, thus demonstrating efficient execution of pi
 pelines involving many successive fine-grained tasks.  Finally, we discuss
  potential future work, including complementary optimizations that could r
 esult in additional performance improvements.\n\nSession Format: Recorded\
 n\nTag: Applications, Architectures, Heterogeneous Systems, Hierarchical P
 arallelism, Parallel Programming Languages and Models, Performance, Perfor
 mance Portability, Scientific Computing\n\nRegistration Category: Workshop
  Reg Pass
END:VEVENT
END:VCALENDAR