Hardware and software partitioning using genetic algorithm in image processing application
Loading...
Date
2018-06
Authors
Loo, Fang Hean
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In this project, HW-SW Partitioning is used as a process to map each task of image
processing application to be executed either in software (Hard Processor System, HPS)
or hardware (Field Programmable Gate Array, FPGA). The framework for HW-SW
Partitioning using Genetic Algorithm (GA) is developed in MATLAB. Total ten different
combinations of GA parameters are used to test the developed framework. The GA
parameters such as population size, crossover percentage and mutation percentage are
varied to get the optimum combination of GA parameters. Three different HW/SW
Partitioned Solutions are generated and the HW resources spent by first, second, and third
solutions must not exceed the constraint value, Q = 341, Q = 681, and Q = 1022
respectively. The HW resource spent in HW/SW Partitioned Solution 1 (Q = 341) is 77.97%
lesser than pure hardware solution. It is 4.65% faster than pure hardware solution and
26.6% faster than pure software solution. The HW resource spent in HW/SW Partitioned
Solution 2 (Q = 681) is 50.29% lesser than pure hardware solution. It is 8.51% faster than
pure hardware solution and 29.61% faster than pure software solution. The HW resource
spent in HW/SW Partitioned Solution 3 (Q = 1022) is 45.01% lesser than pure hardware
solution. It is 10.09% faster than pure hardware solution and 30.83% faster than pure
software solution. Future work of this project is to implement the HW-SW partitioned
solution in Altera DE1-SoC.