Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Vectorization and Parallelization of Irregular Problems

Summary: Vectorization and Parallelization
of Irregular Problems
via Graph Coloring
Hans-Christian Hege Hinnerk St¨uben 1
1Freie Universit¨at Berlin, Institut f¨ur Theoretische Physik, Arnimallee 14,
D-1000 Berlin 33
Efficient implementations of irregular problems on vector and parallel archi-
tectures generally are hard to realize. An important class of irregular prob-
lems are Gauß-Seidel iteration schemes applied to irregular data sets. The
unstructured data dependences arising there prevent restructuring compilers
from generating efficient code for vector or parallel machines. It is shown,
how to structure the data dependences by decomposing the data set using
graph coloring techniques and by specifying a particular execution order al-
ready on the algorithm level. Methods to master the irregularities originating
from different types of tasks are proposed. An example of application is given
and possible future developments are mentioned.
1 Introduction 1
2 Preliminaries 2


Source: Andrzejak, Artur - Konrad-Zuse-Zentrum für Informationstechnik Berlin


Collections: Computer Technologies and Information Sciences