High Level Synthesis using Learning Automata Genetic Algorithm

High-level synthesis consists of many interdependent tasks such as scheduling, allocation and binding. All tasks in high-level synthesis are NP-complete and the design objectives are in conflict for nature, most of the already proposed approaches are not efficient in the exploration of the design space and not effective in the identification of different trade-offs. For these reasons, genetic algorithms can be considered as good candidates to tackle such difficult explorations. A new algorithm that named Learning Automata Genetic Algorithm (LAGA) is used in this paper to perform scheduling and allocation concurrently. This algorithm is based on the Genetic Algorithm, the difference is that the Learning Process is added to the Genetic Algorithm. This strategy can complete the scheduling and the allocation effectively in the high-level synthesis under certain time and resource constraints. This algorithm is implemented in C language and is tested finally on a number of DSP benchmarks, and the test results then are compared with those obtained from four other different techniques which are commonly used in high-level synthesis. The experimental results show that the high-level synthesis using the LAGA algorithm is very effective, especially under the area constraint.