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:20260522T150117Z
LOCATION:D175
DTSTART;TZID=America/Chicago:20181112T170000
DTEND;TZID=America/Chicago:20181112T170700
UID:submissions.supercomputing.org_SC18_sess176_ws_llvmlt101@linklings.com
SUMMARY:Compiler Optimization for Heterogeneous Locality and Homogeneous P
 arallelism in OpenCL and LLVM
DESCRIPTION:Dorit Nuzman (Intel Corporation); Michael Zuckerman (Mobileye,
  an Intel Company); and Ayal Zaks (Intel Corporation)\n\nHeterogeneous pla
 tforms may include accelerators such as Digital Signal Processors (DSP’s) 
 that employ SW-controlled scratch-pad memories instead of, or in addition 
 to standard HW-cached memory. Controlling scratch-pads efficiently typical
 ly requires tiling and pipelining loops, thereby optimizing for memory loc
 ality rather than parallelism as a primary objective. On the other hand, a
 chieving high performance on CPU’s and GPU’s typically requires optimizing
  for data-level parallelism as a primary objective, compromising locality.
  In this lightning talk, we show how OpenCL and LLVM can be used to achiev
 e both target-dependent locality and target-independent parallelism. Such 
 an approach facilitates the development of optimized software for DSP acce
 lerators while enabling its efficient execution on standard servers. Follo
 wing the work of Tian et al., our approach leverages automatic compiler op
 timization and relies purely on OpenCL, including its device-side enqueue 
 capability and SPIR-V format.\n\nTag: Program Transformation, Programming 
 Systems\n\nRegistration Category: Workshop Reg Pass\n\n
END:VEVENT
END:VCALENDAR