BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Estimating and Exploiting Potential Parallelism by Source-level De
 pendence Profiling - Jonathan Mak (Computer Laboratory)
DTSTART:20100813T141500Z
DTEND:20100813T151500Z
UID:TALK25663@talks.cam.ac.uk
CONTACT:Dominic Orchard
DESCRIPTION:Manual parallelization of programs is known to be difficult an
 d error-prone\, and there are currently few ways to measure the amount of 
 potential parallelism in the original sequential code. We present an exten
 sion of Embla\, a Valgrind-based dependence profiler that links dynamic de
 pendences back to source code. This new tool estimates potential task-leve
 l parallelism in a sequential program and helps programmers exploit it at 
 the source level. Using the popular fork-join model\, our tool provides a 
 realistic estimate of potential speed-up for parallelization with framewor
 ks like Cilk\, TBB or OpenMP 3.0. Estimates can be given for several diffe
 rent parallelization models\, varying in programmer effort and capabilitie
 s required of the underlying implementation. Our tool also outputs source-
 level dependence information to aid the parallelization of programs with l
 ots of inherent parallelism\, as well as critical paths to suggest algorit
 hmic rewrites of programs with little of it. We validate our claims by run
 ning our tool over serial elisions of sample Cilk programs\, finding addit
 ional inherent parallelism not exploited by the Cilk code\, as well as ove
 r serial C benchmarks where the profiling results suggest parallelism-enha
 ncing algorithmic rewrites.\nJoint work with Karl-Filip Faxén 
LOCATION:GS15\, Computer Laboratory
END:VEVENT
END:VCALENDAR