Title :
Operation scheduling for parallel functional units using genetic algorithms
Author :
Zeitlhofer, Th ; Wess, B.
Author_Institution :
INTHFT, Wien Univ., Austria
Abstract :
We describe a new and efficient approach to solve the scheduling problem for VLIW architectures. The scheduling times of the operations are used as the problems parameters. This in conjunction with a pruning technique based on critical path analysis leads to a significant reduction of search space complexity. A genetic algorithm is used to search for valid schedules of a given length. The genetic algorithm uses a fitness vector that guides the genetic operators crossover and mutation resulting in a fast convergence towards near perfect solutions. The proposed method is also applicable to the problem of register allocation by using a different fitness function. Another advantage of the genetic algorithm approach is that usually a great number of equally performing schedules is obtained allowing for further optimization subject to arbitrary constraints
Keywords :
computational complexity; critical path analysis; digital signal processing chips; directed graphs; genetic algorithms; parallel architectures; processor scheduling; VLIW architectures; critical path analysis; data dependence graph; directed acyclic graph; fast convergence; fitness vector; genetic algorithms; genetic operators crossover; mutation; near perfect solutions; operation scheduling; parallel functional units; problem parameters; pruning technique; register allocation; scheduling times; search space complexity reduction; Biological cells; Compaction; Constraint optimization; Digital signal processing; Genetic algorithms; Genetic mutations; Modems; Processor scheduling; Registers; VLIW;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1999. Proceedings., 1999 IEEE International Conference on
Conference_Location :
Phoenix, AZ
Print_ISBN :
0-7803-5041-3
DOI :
10.1109/ICASSP.1999.758319