Abstract-We propose a new DVFS algorithm for enterprise systems that elevates performance as a first order control parameter and manages frequency and voltage as a function of performance requirements. We implement our algorithm on real Intel Westmere platform in Linux and demonstrate its ability to reduce the standard deviation from target performance by more than 90% over state of the art policies while reducing average power by 17%.

Abstract-Heterogeneous Multi-Processor Systems on a Chip (MPSoCs) are more complex from a thermal perspective compared to the homogeneous MPSoCs because of their inherent imbalance in power density. In this work we develop TempoMP, a new technique for thermal management of heterogeneous MPSoCs which leverages multi-parametric optimization along with our novel thermal predictor, Tempo. TempoMP is able to deliver locally optimal dynamic thermal management decisions to meet thermal constraints while minimizing power and maximizing performance. It leverages our Tempo predictor which, unlike the previous techniques, can estimate the impact of future power state changes at negligible overhead. Our experiments show that compared to the state of the art, Tempo can reduce the maximum prediction error by up to an order of magnitude. Our experiments with heterogeneous MPSoCs also show that TempoMP meets thermal constraints while reducing the average task lateness by 2.5X and energy-lateness product by 5X compared to the state of the art techniques. I. IntroductionHeterogeneous MPSoCs integrate cores of various performance and power characteristics to provide a better tradeoff with respect to performance, power and temperature by allowing customization of performance and power of the chip to match the requirements of the workload. uses ARMA model which has to be updated at run time to avoid inaccuracies due to workload changes, thus leading to overhead. A few techniques have been proposed that do not require online adaptation, such as [19] and [5], but both assume that previous thermal history is a good estimate of future thermal behavior, and thus neglect the impact of future power state changes on temperature. Proactive thermal management algorithms that leverage these predictors are heuristic in nature.A number of thermal management techniques have been proposed that leverage control theory and optimization. In [25], convex optimization is used to control the frequency of the cores on a homogeneous MPSoC to guaranty that thermal constraints are met. In [33], a linear quadratic regulator is used to solve the frequency assignment problem

Abstract-Traditionally CPU workload scheduling and fan control in multi-socket systems have been designed separately leading to less efficient solutions. In this paper we present Cool and Save, a cooling aware dynamic workload management strategy that is significantly more energy efficient than state-of-the art solutions in multi-socket CPU systems because it performs workload scheduling in tandem with controlling socket fan speeds. Our experimental results indicate that applying our scheme gives average fan energy savings of 73% concurrently with reducing the maximum fan speed by 53%, thus leading to lower vibrations and noise levels.