Software level green computing for large scale systems

Abstract

Energy conservation has become a critical issue in modern system electronic devices. Energy wastage in electronic devices occurs in both hardware and software components. Software drives the hardware thus decisions taken during software design and development have significant impact on energy consumption of a computing system. Green Computing addresses energy conservation by application of different techniques at software and hardware level. Energy efficient compiler is a software level green computing technique. Besides compiler optimization, an effective scheduling approach makes efficient use of resources to directly impact the green aspect. Therefore, focus of this paper is identification of energy conservation measures for software level and there utilization at compiler and scheduler. A Distributed Green Compiler (DGC) is presented in this research that is hardware independent and uses an existing distributed compiler. It distributes source code of software over a network, reshapes binary code by applying green strategies during code transformation at compile time and gives green suggestion to software programmer for energy conservation. For scheduling, Distributed Interactive Engineering Toolbox (DIET) scheduler is used and a new algorithm is proposed for the DIET scheduler. The proposed algorithm introduces green aspect in scheduler to effectually make use of resources in such a way that consumption of power and carbon dioxide emission is reduced. Performance analysis of proposed compiler shows that it conserves energy clock cycles up to 40% by applying few green strategies. © 2012 Fakhar et al.