|
Title:
|
EXTRACTING KNOWLEDGE TO PREDICT TSP PERFORMANCE ORDER |
|
Author(s):
|
Paula Cecilia Fritzsche , Dolores Rexachs , Emilio Luque |
|
ISBN:
|
978-972-8924-40-9 |
|
Editors:
|
Jörg Roth, Jairo Gutiérrez and Ajith P. Abraham (series editors: Piet Kommers, Pedro Isaías and Nian-Shing Chen) |
|
Year:
|
2007 |
|
Edition:
|
Single |
|
Type:
|
Full Paper |
|
First Page:
|
65 |
|
Last Page:
|
72 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Parallel distributed architectures are essential for solving large-scale scientific and engineering problems. Its increasing
use has generated the need for performance prediction for both deterministic applications and non-deterministic
applications.
The parallel performance prediction of data-dependent applications is an extremely challenging problem because for a
specific issue the input data sets may cause variability in execution times. Some examples of this kind of applications are
the sorting algorithms, the searching algorithms, the traveling salesman problem (TSP), and practical problems that can
be formulated as TSP problems.
The development of a new practical prediction methodology to estimate the execution time of data-dependent parallel
applications is the primary target of this study. The entire methodology consists of several stages: the hypotheses
formulation, the composition of experiments, the execution of the studied application, the knowledge discovery in
databases (KDD) process, the understanding of the model, the quality evaluation, and finally the comparison module.
Three different parallel TSP algorithms are used to show the usefulness of the proposed methodology. The experimental
results are quite promising; the capacity of prediction is greater than 75%. |
|
|
|
|
|
|
