Applying simulated evolution to scheduling in high level synthesis

Abstract

Listen

Translate article

Most scheduling algorithms in high level synthesis are greedy in nature and hence are vulnerable to local minimums in the design space. A novel scheduling algorithm is presented based on simulated evolution which incorporates probabilistic uphill moves to escape from local minimums. This algorithm uses local heuristics and simple cost functions, and relies on rapid iterations and effective design space exploration to obtain superior designs. >