Power consumption is an important constraint in achieving efficient execution on High Performance Computing Multicore Systems. As the number of cores available on a chip continues to increase, the importance of power consumption will continue to grow. In order to achieve improved performance on multicore systems scientific applications must make use of efficient methods for reducing power consumption and must further be refined to achieve reduced execution time. In this dissertation, we introduce a performance modeling framework, E-AMOM, to enable improved execution of scientific applications on parallel multicore systems with regards to a limited power budget. We develop models for each application based upon performance hardware counters. Our models utilize different performance counters for each application and for each performance component (runtime, system power consumption, CPU power consumption, and memory power consumption) that are selected via our performance-tuned principal component analysis method. Models developed through E-AMOM provide insight into the performance characteristics of each application that affect performance for each component on a parallel multicore system. Our models are more than 92% accurate across both Hybrid (MPI/OpenMP) and MPI implementations for six scientific applications. E-AMOM includes an optimization component that utilizes our models to employ run-time Dynamic Voltage and Frequency Scaling (DVFS) and Dynamic Concurrency Throttling to reduce power consumption of the scientific applications. Further, we optimize our applications based upon insights provided by the performance models to reduce runtime of the applications. Our methods and techniques are able to save up to 18% in energy consumption for Hybrid (MPI/OpenMP) and MPI scientific applications and reduce the runtime of the applications up to 11% on parallel multicore systems.

Nowadays there is no need to convince anyone about the necessity of renewable energy. One of the most promising ways to obtain it is the wind power. Countries like Denmark, Germany or Spain proved that, while professionally managed, it can cover a substantial part of the overall energy demand. One of the main and specific problems related to the wind power management — development of the accurate power prediction models. Nowadays State-Of-Art systems provide predictions for a single wind turbine, wind farm or a group of them. However, the spatio-temporal propagation of the errors is not adequately considered. In this paper the potential for improving modern wind power prediction tool WPPT, based on the spatio-temporal propagation of the errors, is examined. Several statistical models (Linear, Threshold, Varying-coefficient and Conditional Parametric) capturing the cross-dependency of the errors, obtained in different parts of the country, are presented. The analysis is based on the weather forecast information and wind power prediction errors obtained for the territory of Denmark in the year 2004. / Vienas iš perspektyviausių bei labiausiai plėtojamų atsinaujinančių energijos šaltinių - vėjas. Tokios Europos Sąjungos šalys kaip Danija, Vokietija bei Ispanija savo patirtimi įrodė, jog tinkamai valdomas bei vystomas vėjo ūkis gali padengti svarią šalies energijos paklausos dalį. Pagal Europos Sąjungos direktyvą 2001/77/EC Lietuva yra įsipareigojusi iki 2010 m. pasiekti, kad elektros energijos gamyba iš atsinaujinančių energijos išteklių sudarytų 7% suvartojamos elektros energijos. Šių įsipareigojimų įvykdymui Lietuvos vyriausybės priimtu nutarimu yra nustatyta atsinaujinančių energijos išteklių naudojimo skatinimo tvarka, pagal kurią numatyta palaipsniui plėsti vėjo energijos naudojimą šalyje. Planuojama, kad iki 2010 m. bus pastatyta 200 MW bendros galios vėjo elektrinių, kurios gamins apie 2,2% visos suvartojamos elektros energijos [Marčiukaitis, 2007]. Didėjant vėjo energijos daliai energetikos sistemoje, Lietuvoje ateityje kils sistemos balansavimo problemų dėl nuolatinių vėjo jėgainių galios svyravimų. Kaip rodo kitų šalių patirtis, vėjo elektrinių galios prognozė yra efektyvi priemonė, leidžianti išspręsti šias problemas. Šiame darbe pristatyti keletas statistinių modelių bei metodų, skirtų išgaunamos vėjo energijos prognozėms gerinti. Analizė bei modeliavimas atlikti nagrinėjant Danijos WPPT (Wind Power Prediction Tool) duomenis bei meteorologines prognozes. Pagrindinis darbo tikslas - modifikuoti WPPT, atsižvelgiant į vėjo krypties bei stiprio įtaką energijos... [toliau žr. visą tekstą]

The single-core performance trend predicted by Moore's law has been impeded in recent years partly due to the limitations imposed by increasing processor power demands. One way to mitigate this limitation in performance improvement is the introduction of multi-core and multi-processor computation. Another approach to increasing the performance-per-Watt metric is to utilize the processor's power more efficiently. In a single-core system, the processor cannot sustainably dissipate more than the nominal Thermal Design Power (TDP) limit determined for the processor at design time. Therefore it is important to understand and manage the power demands of the processes being executed. This principle also applies to multi-core and multi-processor environments. In a multi-processor environment, knowing the power demands of the workload, the power management unit can schedule the workload to a processor based on the state of each processor and process in the most efficient way. This is an example of the knapsack problem. Another approach, also applicable to multi-cores, could be to reduce the core's power by reducing its working voltage and frequency, leading to mitigation of the power bursts, lending more headroom to other cores, and keeping the total power under the TDP limit. The information collected from the execution of the software running on the processor (i.e. the workload) is the key to determining the actions needed with regards to power management at any given time. This work comprises two different approaches in improving the comprehension of software power demands as it executes on the processor. In the first part of this work, the effects of software data on power is analysed. It is important to be able to model the power based on the instructions it comprises, however, to the best of our knowledge, no work exists in which the effects of the values being processed has been investigated with regards to processor power. Creating a power model capable of accurately reflecting the power demands of the software at any given time is a problem addressed by previous research. The software power model can be used in processor simulation environments as well as in the processor itself to create an estimated power dissipation without the need to physically measure the power. In the first part of this research, the effects of software data on power is investigated. In order to collect the data required as part of this research, a profiler tool has been developed by the author and used in this part of the research as well as the second part. The second part of this work focuses on the development of processor power throughout time during the execution of the software. Understanding the power demands of the processor at any given time is important to maintain and manage processor power. Additionally, acquiring an insight into the future power demands of the software can help the system with scheduling planning ahead of time, in order to prepare for any high-power section of the code as well as to plan to use the available power headroom as a result of an upcoming low-power section. In this part of our work, a new hierarchical approach to software phase classification is developed. Software phase classification problem focuses on determining the behaviour of the software at any given time slice by assigning the time slice to one of pre-determined software phases. Each phase is assumed to have known behaviour which was previously measured and instrumented based on previously observed instances of the phase, or by utilizing a model capable of estimating the behaviour of each phase. Using a two-tiered hierarchical clustering approach, our proposed phase classification methodology incorporates the recent performance behaviour of the software in order to determine the power phase. We focused on determining the power phase using the performance information because the real processor power is not usually available without the need for added hardware, while there exists a large number of different performance counters available on most modern processors. Additionally, based on our observations, the relation between performance phases and power behaviour is highly predictable. This method is shown to provide robust results with a low amount of noise compared to other methods, while providing a high enough timing accuracy for the processor to act on. To the best of our knowledge, no other existing work is able to provide both timing accuracy and reduced noise compared to our work. Software phase classification can be used to control the processor power based on the software's phase at any given time, but it does not provide future insight into the progression of the workload. Finally, we developed and compared several phase prediction methodologies based on phase precursors and phase locality concepts. Phase precursor-based methods rely on detecting the precursors observed before the software enters a certain phase, while phase locality methods rely on the locality principle, which postulates a high probability for the current software behaviour to be observed in the near-future. The phase classification, as well as phase prediction methodologies was shown to be able to reduce the power bursts within a workload in order to provide a more smooth power trace. As the bursts are removed from one workload's power trace, the multi-core processor power headroom can be confidently utilized for another process. / Graduate

abstract: With a ground-based Doppler lidar on the upwind side of a wind farm in the Tehachapi Pass of California, radial wind velocity measurements were collected for repeating sector sweeps, scanning up to 10 kilometers away. This region consisted of complex terrain, with the scans made between mountains. The dataset was utilized for techniques being studied for short-term forecasting of wind power by correlating changes in energy content and of turbulence intensity by tracking spatial variance, in the wind ahead of a wind farm. A ramp event was also captured and its propagation was tracked. Orthogonal horizontal wind vectors were retrieved from the radial velocity using a sector Velocity Azimuth Display method. Streamlines were plotted to determine the potential sites for a correlation of upstream wind speed with wind speed at downstream locations near the wind farm. A "virtual wind turbine" was "placed" in locations along the streamline by using the time-series velocity data at the location as the input to a modeled wind turbine, to determine the extractable energy content at that location. The relationship between this time-dependent energy content upstream and near the wind farm was studied. By correlating the energy content with each upstream location based on a time shift estimated according to advection at the mean wind speed, several fits were evaluated. A prediction of the downstream energy content was produced by shifting the power output in time and applying the best-fit function. This method made predictions of the power near the wind farm several minutes in advance. Predictions were also made up to an hour in advance for a large ramp event. The Magnitude Absolute Error and Standard Deviation are presented for the predictions based on each selected upstream location. / Dissertation/Thesis / Masters Thesis Mechanical Engineering 2014

CoordenaÃÃo de AperfeiÃoamento de NÃvel Superior / A constÃncia dos ventos brasileiros, e a necessidade de amenizar a demanda das grandes cidades encontraram na energia eÃlica uma forte parceira. Como uma alternativa na produÃÃo de energia elÃtrica, condomÃnios e prÃdios modernos, alÃm de algumas aplicaÃÃes rurais, tÃm recorrido Ãs turbinas eÃlicas de pequeno porte como uma alternativa para sanar suas necessidades. Contudo, a maioria da tecnologia encontrada no mercado à importada e nÃo foi desenvolvida exclusivamente para aplicaÃÃes no Brasil. A ferramenta mais importante na aerodinÃmica experimental à o tÃnel de vento. Experimentos controlados em escala fornecem um grande nÃmero de dados confiÃveis, alÃm de fornecer seguranÃa a quem o manuseia. Dada a sua importÃncia, este trabalho tem como objetivo o desenvolvimento, a prototipagem, e conhecer o comportamento de um rotor de turbina eÃlica em escala sujeito a testes em tÃnel de vento e comparar os resultados obtidos com os presentes na literatura para um protÃtipo em escala real. Para tanto, faz necessÃrio o uso de tÃcnicas de correÃÃo de efeitos de bloqueio de tÃnel de vento. Foi projetado e fabricado 4 conjuntos de rotores com valores de velocidade especÃfica de ponta de 6 atà 9 (λp=6, 7, 8 e 9). Os testes foram conduzidos em um tÃnel de vento onde foram coletados dados de velocidade de escoamento livre, velocidade de escoamento com a turbina em operaÃÃo, alÃm de medidas de velocidade angular e torque gerado pelas pÃs, com a finalidade de conhecer a curva de potÃncia de cada rotor. Foi observado que em situaÃÃes de escoamento em que as rotaÃÃes nÃo sejam representativas (o suficiente para atingir valores superiores ao intervalo de λ entre 3 e 5,6), indica-se um projeto com λp=6. Jà em situaÃÃes nas quais os valores de λ oscilam entre 4,7 e 7,3, λp=7 mostrou-se mais eficiente. Jà λp=9 mostrou-se nÃo vantajoso em comparaÃÃo aos demais projetos. Ao comparar os dados obtidos neste trabalho com os da literatura e do BEM, pode-se afirmar que o estudo de turbinas eÃlicas em tÃnel de vento à bastante confiÃvel. / The constancy of Brazilian winds and the need to mitigate the demand of large cities have found in wind energy a strong partner. As an alternative for the production of electrical power, modern buildings and condominiums, plus some rural applications, have resorted to small wind turbines as an alternative to solve your needs. However, most of the technology found in the market is imported and has not been developed exclusively for applications in Brazil. The most important tool in experimental aerodynamics is the wind tunnel. Scale controlled experiments provide a large number of reliable data, besides providing security to those who handle. Given its importance, this paper aims at the development, prototyping, and understands the behavior of a wind turbine rotor scale subjected to wind tunnel tests and compares the results with those in the literature for a prototype scale real. Therefore, it required the use of correction techniques blockage effects of the wind tunnel. It was designed and manufactured 4 sets of rotors with values specific tip speed of 6 to 9 (λp=6, 7, 8 e 9). The tests were conducted in a wind tunnel where velocity data free stream, stream velocity with the turbine in operation, and angular speed and torque generated by the blades, was collected in order to know the curve of each rotor. It was observed that in situations where the flow speeds are not representative enough to reach the higher values of λ range between 3 and 5,6, indicates a design with λp=6. Already in situations where the values of λ ranging between 4,7 and 7,3, λp=7 proved to be more efficient. Have λp=9 proved to be no advantage in comparison to other projects. By comparing the data obtained in this work with the literature and the BEM, it can be stated that the study of wind turbines in a wind tunnel is quite reliable.

Wind power production is of growing importance to many countries around the world. To improve reliability and power grid stability related to wind power, forecasting of wind power is becoming an important commercial and research area. Machine learning methods are considered to be highly valuable when making predictions on time series data and as such have become prominent within wind forecasting as well. This thesis extends an existing neural network prediction system with new input data series, in particular the observed wind speed from the wind farm itself. The goal was to investigate the effect this new data series has, and whether or not it could be used to improve predictions as compared to the baseline prediction system defined within this thesis. To do this multiple methods of including the observed wind speed are developed, including a multi-stage network concept. These results are statistically tested to give more evidence for their comparison to baseline. The results show that the multi-stage network concept can use the observed wind speed to improve performance over the baseline case for specific prediction horizons. / Betydelsen för vindkraftsproduktion växer i länder runt om i världen. För att förbättratillförlitligheten och elnätstabiliteten i vindkraften blir dess prognoser viktiga kommersielltoch ett forskningsområde. Maskininlärningsmetoder anses vara mycket värdefullanär man gör förutsägelser om tidsseriedata och har därmed framträdat inom vindprognoser. Detta arbete utökar ett existerande prediktionssystem av neurala nätverk med ny indata,med särskilt den observerade vindhastigheten från själva vindkraftparken. Måletvar att undersöka effekten av denna nya dataserie, och huruvida den skulle kunna användasför att förbättra förutsägelserna jämfört med det befintliga referensprognossystemetdefinierat i denna uppsats. För att kunna göra detta utvecklas flera metoder för att inkludera den observeradevindhastigheten, inklusive ett flerstegs nätverkskoncept. Dessa resultat är statistiskt testadeför att ge mer grund i deras jämförelse med referensmodellen. Resultaten visar att detflerstega nätverkskonceptet kan använda den observerade vindhastigheten för att förbättraprestanda över referensmodellen för specifika prediktionshorisonter.

The expansion of wind power for electrical energy production has increased in recent years and shows no signs of slowing down. This unpredictable source of energy has contributed to destabilization of the electrical grid causing the energy market prices to vary significantly on a daily basis. For energy producers and consumers to make good investments, methods have been developed to make predictions of wind power production. These methods are often based on machine learning were historical weather prognosis and wind power production data is used. However, the data often contain outliers, causing the machine learning methods to create inaccurate predictions. The goal of this Master’s Thesis was to identify and remove these outliers from the data so that the accuracy of machine learning predictions can improve. To do this an outlier detection method using unsupervised clustering has been developed and research has been made on the subject of using machine learning for outlier detection and wind power production prediction. / Vindkraftsproduktion som källa för hållbar elektrisk energi har på senare år ökat och visar inga tecken på att sakta in. Den här oförutsägbara källan till energi har bidragit till att destabilisera elnätet vilket orsakat dagliga kraftiga svängningar i priser på elmarknaden. För att elproducenter och konsumenter ska kunna göra bra investeringar har metoder för att prediktera vindkraftsproduktionen utvecklats. Dessa metoder är ofta baserade på maskininlärning där historiska data från väderleksprognoser och vindkraftsproduktion använts. Denna data kan innehålla så kallade outliers, vilket resulterar i försämrade prediktioner från maskininlärningsmetoderna. Målet med det här examensarbetet var att identifiera och ta bort outliers från data så att prediktionerna från dessa metoder kan förbättras. För att göra det har en metod för outlier-identifikation utveklats baserad på oövervakad maskininlärning och forskning har genomförts på områdena inom maskininlärning för att identifiera outliers samt prediktion för vindkraftsproduktion.

In modern computer systems, memory accesses and power management are the two major performance limiting factors. Accesses to main memory are very slow when compared to operations within a processor chip. Hardware write buffers, caches, out-of-order execution, and prefetch logic, are commonly used to reduce the time spent waiting for main memory accesses. Compiler loop interchange and data layout transformations also can help. Unfortunately, large data structures often have access patterns for which none of the standard approaches are useful. Using smaller data structures can significantly improve performance by allowing the data to reside in higher levels of the memory hierarchy. This dissertation proposes using lossy data compression technology called ’Compressive Hashing’ to create “surrogates”, that can augment original large data structures to yield faster typical data access. One way to optimize system performance for power consumption is to provide a predictive control of system-level energy use. This dissertation creates a novel instruction-level cost model called the variable-n-gram model, which is closely related to N-Gram analysis commonly used in computational linguistics. This model does not require direct knowledge of complex architectural details, and is capable of determining performance relationships between instructions from an execution trace. Experimental measurements are used to derive a context-sensitive model for performance of each type of instruction in the context of an N-instruction sequence. Dynamic runtime power prediction mechanisms often suffer from high overhead costs. To reduce the overhead, this dissertation encodes the static instruction-level predictions into a data structure and uses compressive hashing to provide on-demand runtime access to those predictions. Genetic programming is used to evolve compressive hash functions and performance analysis of applications shows that, runtime access overhead can be reduced by a factor of ~3x-9x.

<p>Prediction of the rapidly fading envelope of a mobile radio channel enables a number of capacity improving techniques like fast resource allocation and fast link adaptation. This thesis deals with linear prediction of the complex impulse response of a channel and unbiased quadratic prediction of the power. The design and performance of these predictors depend heavily on the correlation properties of the channel. Models for a channelwhere the multipath is caused by clusters of scatterers are studied. The correlation for the contribution from a cluster can be approximated as a damped complex sinusoid. A suitable model for the dynamics of the channel is an ARMA-process. This motivates the use of linear predictors.</p><p>A limiting factor in the prediction are the estimation errors on the observed channels. This estimation error, caused by measurement noise and time variation, is analyzed for a block based least squares algorithm which operates on a Jakes channel model. Efficient noise reduction on the estimated channel impulse responses can be obtained with Wienersmoothers that are based on simple models for the dynamics of the channel combined with estimates of the variance of the estimation error.</p><p>Power prediction that is based on the squared magnitude of linear prediction of the taps will be biased. Hence, a bias compensated power predictor is proposed and the optimal prediction coefficients are derived for the Rayleigh fading channel. The corresponding probability density functions for the predicted power are also derived. A performance evaluation of the prediction algorithm is carried out on measured broadband mobile radio channels. The performance is highly dependent on the variance of the estimation error and the dynamics of the individual taps.</p>

Prediction of the rapidly fading envelope of a mobile radio channel enables a number of capacity improving techniques like fast resource allocation and fast link adaptation. This thesis deals with linear prediction of the complex impulse response of a channel and unbiased quadratic prediction of the power. The design and performance of these predictors depend heavily on the correlation properties of the channel. Models for a channelwhere the multipath is caused by clusters of scatterers are studied. The correlation for the contribution from a cluster can be approximated as a damped complex sinusoid. A suitable model for the dynamics of the channel is an ARMA-process. This motivates the use of linear predictors. A limiting factor in the prediction are the estimation errors on the observed channels. This estimation error, caused by measurement noise and time variation, is analyzed for a block based least squares algorithm which operates on a Jakes channel model. Efficient noise reduction on the estimated channel impulse responses can be obtained with Wienersmoothers that are based on simple models for the dynamics of the channel combined with estimates of the variance of the estimation error. Power prediction that is based on the squared magnitude of linear prediction of the taps will be biased. Hence, a bias compensated power predictor is proposed and the optimal prediction coefficients are derived for the Rayleigh fading channel. The corresponding probability density functions for the predicted power are also derived. A performance evaluation of the prediction algorithm is carried out on measured broadband mobile radio channels. The performance is highly dependent on the variance of the estimation error and the dynamics of the individual taps.