|
Title:
|
PERFORMANCE PREDICTION OF FUTURE GENERATION COMPUTER SYSTEMS |
|
Author(s):
|
Andreas Boklund , Stefan Christiernin |
|
ISBN:
|
972-8924-09-7 |
|
Editors:
|
Nuno Guimarães, Pedro Isaías and Ambrosio Goikoetxea |
|
Year:
|
2006 |
|
Edition:
|
Single |
|
Keywords:
|
Evolution, performance, evaluation, prediction, Gustafson-Barsis and Amdahl. |
|
Type:
|
Full Paper |
|
First Page:
|
42 |
|
Last Page:
|
48 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Choosing the correct parallel algorithm is often a challenging task. Choosing the optimal algorithm not only for current, but also for future computer systems has traditionally been deemed as impossible. In this paper we present a new model for Evolutionary Performance Prediction that is based on, among others, Amdahls, Gustafson-Barsis and Moores laws. We also show how the variables can be determined both theoretically and experimentally through a number of examples. Finally we conclude the article with a discussion of practical limitations and validity of the proposed model. |
|
|
|
|
|
|
