Optimal power allocation and load distribution for multiple heterogeneous multicore server processors across clouds and data centers

For multiple heterogeneous multicore server processors across clouds and data centers, the aggregated performance of the cloud of clouds can be optimized by load distribution and balancing. Energy efficiency is one of the most important issues for large-scale server systems in current and future data centers. The multicore processor technology provides new levels of performance and energy efficiency. The present paper aims to develop power and performance constrained load distribution methods for cloud computing in current and future large-scale data centers. In particular, we address the problem of optimal power allocation and load distribution for multiple heterogeneous multicore server processors across clouds and data centers. Our strategy is to formulate optimal power allocation and load distribution for multiple servers in a cloud of clouds as optimization problems, i.e., power constrained performance optimization and performance constrained power optimization. Our research problems in large-scale data centers are well-defined multivariable optimization problems, which explore the power-performance tradeoff by fixing one factor and minimizing the other, from the perspective of optimal load distribution. It is clear that such power and performance optimization is important for a cloud computing provider to efficiently utilize all the available resources. We model a multicore server processor as a queuing system with multiple servers. Our optimization problems are solved for two different models of core speed, where one model assumes that a core runs at zero speed when it is idle, and the other model assumes that a core runs at a constant speed. Our results in this paper provide new theoretical insights into power management and performance optimization in data centers.

Performance evaluation of 3D printed miniature electromagnetic energy harvesters driven by air flow

As a renewable and non-polluting energy source, wind is used to produce electricity via large-diameter horizontal or vertical axis wind turbines. Such large wind turbines have been well designed and widely applied in industry. However, little attention has been paid to the design and development of miniature wind energy harvesters, which have great potential to be applied to the HVAC (heating, ventilating and air conditions) ventilation exhaust systems and household personal properties. In this work, 10 air-driven electromagnetic energy harvesters are fabricated using 3D printing technology. Parametric measurements are then conducted to study the effects of (1) the blade number, (2) its geometric size, (3) aspect ratio, presence or absence of (4) solid central shaft, (5) end plates, and (6) blade orientation. The maximum electrical power is 0.305 W. To demonstrate its practical application, the electricity generated is used to power 4 LED (light-emitting diode) lights. The maximum overall efficiency ηmax is approximately 6.59%. The cut-in and minimum operating Reynolds numbers are measured. The present study reveals that the 3D printed miniature energy harvesters provide a more efficient platform for harnessing ‘wind power’.

Attention and basic literacy and numeracy in children with Autism Spectrum Disorder: a one-year follow-up study

Little is known about the link between Executive Functioning (EF) and academic performance in children with Autism Spectrum Disorder (ASD) and how such links develop over time. This study examined word reading, basic mathematics, attention switching, sustained attention and their development. Two age, gender and perceptual IQ matched groups of cognitively able 7-12 year olds (ASD N = 40; typical developing [TYP] N = 40) were assessed at baseline and one year later, completing Word Reading and Numerical Operations tests and computerized tasks tapping attention switching and sustained attention. Children with ASD had similar word reading and numerical operations performance and similar development of these skills relative to TYP children. A delay in attention switching but similar development was found in children with ASD relative to TYP children. The EF tasks were correlated with reading and mathematics in ASD children only, however, in regression analyses these factors were not significant predictors of Time 2 reading and mathematics after accounting for Time 1 reading and mathematics scores. These findings indicate similar word reading and mathematics development but atypical attention profiles in cognitively able children with ASD. Implications for educators are discussed.

Optimal task placement with QoS constraints in geo-distributed data centers using DVFS

 With the rising demands on cloud services, the electricity consumption has been increasing drastically as the main operational expenditure (OPEX) to data center providers. The geographical heterogeneity of electricity prices motivates us to study the task placement problem over geo-distributed data centers. We exploit the dynamic frequency scaling technique and formulate an optimization problem that minimizes OPEX while guaranteeing the quality-of-service, i.e., the expected response time of tasks. Furthermore, an optimal solution is discovered for this formulated problem. The experimental results show that our proposal achieves much higher cost-efficiency than the traditional resizing scheme, i.e., by activating/deactivating certain servers in data centers.

On self-propagating methodological flaws in performance normalization for strength and power sports

Performance in strength and power sports is greatly affected by a variety of anthropometric factors. The goal of performance normalization is to factor out the effects of confounding factors and compute a canonical (normalized) performance measure from the observed absolute performance. Performance normalization is applied in the ranking of elite athletes, as well as in the early stages of youth talent selection. Consequently, it is crucial that the process is principled and fair. The corpus of previous work on this topic, which is significant, is uniform in the methodology adopted. Performance normalization is universally reduced to a regression task: the collected performance data are used to fit a regression function that is then used to scale future performances. The present article demonstrates that this approach is fundamentally flawed. It inherently creates a bias that unfairly penalizes athletes with certain allometric characteristics, and, by virtue of its adoption in the ranking and selection of elite athletes, propagates and strengthens this bias over time. The main flaws are shown to originate in the criteria for selecting the data used for regression, as well as in the manner in which the regression model is applied in normalization. This analysis brings into light the aforesaid methodological flaws and motivates further work on the development of principled methods, the foundations of which are also laid out in this work.

Modeling intermittent cycling performance in hypoxia using the critical power concept

PURPOSE: This study investigated the efficacy of an intermittent critical power model, termed the "work-balance" (W'BAL) model, during high-intensity exercise in hypoxia. METHODS: Eleven trained, male cyclists (mean ± SD; age 27 ± 6.6 yr, V[Combining Dot Above]O2peak 4.79 ± 0.56 L.min) completed a maximal ramp test and a 3 min "all-out" test to determine critical power (CP) and work performed above CP (W'). On another day an intermittent exercise test to task failure was performed. All procedures were performed in normoxia (NORM) and hypoxia (HYPO; FiO2 ≈ 0.155) in a single-blind, randomized and counter-balanced experimental design. The W'BAL model was used to calculate the minimum W' (W'BALmin) achieved during the intermittent test. W'BALmin in HYPO was also calculated using CP + W' derived in NORM (N+H). RESULTS: In HYPO there was an 18% decrease in V[Combining Dot Above]O2peak (4.79 ± 0.56 vs 3.93 ± 0.47 L.min ; P<0.001) and a 9% decrease in CP (347 ± 45 vs 316 ± 46 W; P<0.001). No significant change for W' occurred (13.4 ± 3.9 vs 13.7 ± 4.9 kJ; P=0.69; NORM vs HYPO). The change in V[Combining Dot Above]O2peak was significantly correlated with the change in CP (r = 0.72; P=0.01). There was no difference between NORM and HYPO for W'BALmin (1.1 ± 0.9 kJ vs 1.2 ± 0.6 kJ). The N+H analysis grossly overestimated W'BALmin (7.8 ± 3.4 kJ) compared with HYPO (P<0.001). CONCLUSION: The W'BAL model produced similar results in hypoxia and normoxia, but only when model parameters were determined under the same environmental conditions as the performance task. Application of the W'BAL model at altitude requires a modification of the model, or that CP and W' are measured at altitude.

Towards power consumption-delay tradeoff by workload allocation in cloud-fog computing

Fog computing, characterized by extending cloud computing to the edge of the network, has recently received considerable attention. The fog is not a substitute but a powerful complement to the cloud. It is worthy of studying the interplay and cooperation between the edge (fog) and the core (cloud). To address this issue, we study the tradeoff between power consumption and delay in a cloud-fog computing system. Specifically, we first mathematically formulate the workload allocation problem. After that, we develop an approximate solution to decompose the primal problem into three subproblems of corresponding subsystems, which can be independently solved. Finally, based on extensive simulations and numerical results, we show that by sacrificing modest computation resources to save communication bandwidth and reduce transmission latency, fog computing can significantly improve the performance of cloud computing.