|
Abstract:
|
Two high speed implementations of the genetic algorithm (GA) are described and their performances are evaluated on a highly constrained DNA Code Word Library Generation test case problem. The first is a distributed, or multi-deme, Island Model GA coded in C that uses the Message Passing Interface (MPI) protocol and runs on multiple processors in a cluster. The second is a single population GA coded in VHDL that implements both the GA and the fitness function evaluator in hardware on a single Field Programmable Logic Array (FPGA) chip. While the distributed GA is generally applicable to many problem types, the hardware GA is especially applicable to problems characterized by a fitness function requiring the calculation of a matrix of relatively simple integer-only or Boolean logic functions that can be efficiently implemented in a hardware systolic array.
|
