Comparative Framework for the Analysis of Thermal and Resource Management Algorithms for Multi-Core Architectures

Moeez Akmal1, Muhammad Sarmad Saeed1, Muhammad Usama Sardar2, Hareem Shafi1, Osman Hasan1, Heba Khdr3, Jorg Henkel3
1National University of Sciences and Technology, Islamabad, Pakistan, 2Technische Universität Dresden, 3Karlsruhe Institute of Technology, Karlsruhe, Germany


Abstract

With current trends in hardware design, on-chip multi-core architectures have emerged as a new paradigm for executing highly parallel and resource demanding applications. However, inefficient resource management and thermal issues in these multi-core architectures may lead to performance degradation. To cater for these issues, several resource and thermal management algorithms have been proposed. These algorithms are usually analyzed independently and under different environmental settings, which makes a fair comparison between various thermal or resource management algorithms very difficult. In order to overcome this problem, this paper presents a generic framework for analyzing various resource and thermal management algorithms under similar settings. The proposed framework is developed using C++ and has been successfully used to compare mDTM and DsRem in this paper.