Publication Details
Tuning Parallel Performance of Global Image Processing Operators
Dvořák Václav, prof. Ing., DrSc.
Parallel performance, FFT, FWT, Performance tuning
Parallel performance and power consumption is all that usually mattersin custom computing for image processing tasks. Tuning the performanceon a particular parallel platform is not easy and can be aided bysimulation. This paper investigates performance of two global imageprocessing operators, FFT and FWT, on distributed (shared) memoryarchitecture. As the chip multiprocessors and processor arrays arebecoming ever more popular, understanding their parallel performance isessential. Variations in processor count, clock rate, link speed, busbandwidth, cache line size, cache size, as well as mapping softwarecomponents to processors influence performance a great deal and areanalyzed in the paper.
@inproceedings{BUT13977,
author="Vladimír {Kutálek} and Václav {Dvořák}",
title="Tuning Parallel Performance of Global Image Processing Operators",
booktitle="Proceedings of 37th International Conference MOSIS´03 Modelling and Simulation of Systems",
year="2003",
pages="57--64",
address="Ostrava",
isbn="80-85988-86-0"
}