Algorithms are presented for extracting synchronization-free
parallelism available in arbitrarily nested parameterized loops.
The parallelism is represented with synchronization-free trees
of dependent operations. Sets representing trees can be
described with non-linear expressions. The main idea is to
firstly extract sources of synchronization-free trees and next
to generate parallel code based on a while loop. Experimental
results are presented exposing speed-up and efficiency of
parallel programs written in the OpenMP standard on the basis of
code generated by the algorithms proposed.
%0 Book Section
%1 Bielecki2008-or
%A Bielecki, Wlodzimierz
%A Beletska, Anna
%A Palkowski, Marek
%A Pietro, Pierluigi San
%B Algorithms and Architectures for Parallel Processing
%D 2008
%I Springer Berlin Heidelberg
%K Polyhedral_model To_Read
%P 185--195
%T Finding Synchronization-Free Parallelism Represented with
Trees of Dependent Operations
%X Algorithms are presented for extracting synchronization-free
parallelism available in arbitrarily nested parameterized loops.
The parallelism is represented with synchronization-free trees
of dependent operations. Sets representing trees can be
described with non-linear expressions. The main idea is to
firstly extract sources of synchronization-free trees and next
to generate parallel code based on a while loop. Experimental
results are presented exposing speed-up and efficiency of
parallel programs written in the OpenMP standard on the basis of
code generated by the algorithms proposed.
@incollection{Bielecki2008-or,
abstract = {Algorithms are presented for extracting synchronization-free
parallelism available in arbitrarily nested parameterized loops.
The parallelism is represented with synchronization-free trees
of dependent operations. Sets representing trees can be
described with non-linear expressions. The main idea is to
firstly extract sources of synchronization-free trees and next
to generate parallel code based on a while loop. Experimental
results are presented exposing speed-up and efficiency of
parallel programs written in the OpenMP standard on the basis of
code generated by the algorithms proposed.},
added-at = {2015-04-11T18:41:09.000+0200},
author = {Bielecki, Wlodzimierz and Beletska, Anna and Palkowski, Marek and Pietro, Pierluigi San},
biburl = {https://www.bibsonomy.org/bibtex/20f9751ac59c2b1368116d9915d389533/christophv},
booktitle = {Algorithms and Architectures for Parallel Processing},
interhash = {e2d2d632f2865f9ca01ef08d320fa158},
intrahash = {0f9751ac59c2b1368116d9915d389533},
keywords = {Polyhedral_model To_Read},
pages = {185--195},
publisher = {Springer Berlin Heidelberg},
series = {Lecture Notes in Computer Science},
timestamp = {2015-04-11T18:41:09.000+0200},
title = {Finding {Synchronization-Free} Parallelism Represented with
Trees of Dependent Operations},
year = 2008
}