University of Texas at Arlington
Heterogeneous multi-core processor is demonstrated to be more efficient than its homogeneous counterpart due to its ability to meet different resource requirements of the applications. One of the challenges of designing a heterogeneous multi-core is how to schedule programs to the core that can execute them most efficiently. This paper presents a method to map application to its optimum core by analyzing the micro-architecture independent characteristics of that application. The proposed method exploits data/instruction reuse distance and true dependency distance to derive switching gains and ranks for each configuration, and maps the application to the optimum core accordingly.