GA

Implementation of a simple Genetic Algorithm for scheduling optimization

TO RUN:

OfflineScheduler.exe [options]

[options] -a FILE Use FILE as library of architectures (please see bellow for formatting restrictions)

-c DECIMAL_NUM
    Set the crossover rate to DECIMAL_NUM
    
-d FILE
    Use FILE as the DFG file (usually contains a .aif extention)

-g INTEGER
    Run the genetic algorithm for INTEGER generations

-m DECIMAL_NUM
    Set the mutation rate to DECIMAL_NUM

-p INTEGER
    Set the Population size to INTEGER individuals

-t INTEGER
    Seed the random number generator with INTEGER

• These arguments are all optional and can be entered in any order.
• more options are currently in development

DEFAULT VALUES:

Default Values - Architecture Library file = input/architecture_library.txt - DFG = input/B1_10_5.aif

- Crossover rate = 0.85
- Mutation rate = 0.005

- Population size = 50
- Number of Generations = 500

Program Output - Error messages are currently being printed to stderr - All other program output is printed to stdout

CONSTRAINTS

The DFG should be generated by Ahmed's taskGenerator

The Crossover rate should be a decimal number between 0 and 1 The Mutation rate should be a decimal number between 0 and 1

The Population size should be a number between 2 and 10000 The number of generations should be greater than 1

ARCHITECTURE FILE FORMAT

The very first line in the file MUST be as follows: Num_Tasks= # # = the number different tasks that this file contains

The properties of each task must be entered in the following order:
<CONFIGURATION_TIME> <EXECUTION_TIME> <CONFIGURATION_POWER> <EXECUTION_POWER>

- each property must be an integer separated by one or more whitespaces

Lines beginning with # are ignored by the parser

All implementations of the same task MUST have the exact same name (eg. TASK2)

FUTURE IMPLEMENTATION IDEAS

Please see the tag \ FUTURE in the source code