Association rules learning technique for knowledge mining about scheduling algorithm performance

With the advent of increasingly higher numbers of processors on-chip, task scheduling has become an important concern in system design, and research in this area has produced substantial and diversified knowledge. As a result, the efficient management and taping of this knowledge has become a concern in itself. This paper addresses the issue of how to effectively extract performance information about a scheduling algorithm in the context of a set of applications, by learning the association rules between the applications´ attributes and the algorithms´ performance metrics. The new methodology that is presented serves to both increase the designer´s knowledge about a particular scheduling algorithm and compare algorithms.