Improving energy-efficiency of large-scale workflows in heterogeneous systems

Author(s):  
Peng Xiao ◽  
Zhongxiao Hao
Author(s):  
Mark Endrei ◽  
Chao Jin ◽  
Minh Ngoc Dinh ◽  
David Abramson ◽  
Heidi Poxon ◽  
...  

Rising power costs and constraints are driving a growing focus on the energy efficiency of high performance computing systems. The unique characteristics of a particular system and workload and their effect on performance and energy efficiency are typically difficult for application users to assess and to control. Settings for optimum performance and energy efficiency can also diverge, so we need to identify trade-off options that guide a suitable balance between energy use and performance. We present statistical and machine learning models that only require a small number of runs to make accurate Pareto-optimal trade-off predictions using parameters that users can control. We study model training and validation using several parallel kernels and more complex workloads, including Algebraic Multigrid (AMG), Large-scale Atomic Molecular Massively Parallel Simulator, and Livermore Unstructured Lagrangian Explicit Shock Hydrodynamics. We demonstrate that we can train the models using as few as 12 runs, with prediction error of less than 10%. Our AMG results identify trade-off options that provide up to 45% improvement in energy efficiency for around 10% performance loss. We reduce the sample measurement time required for AMG by 90%, from 13 h to 74 min.


2021 ◽  
Vol 32 (9) ◽  
pp. 2367-2380
Author(s):  
Guangming Tan ◽  
Chaoyang Shui ◽  
Yinshan Wang ◽  
Xianzhi Yu ◽  
Yujin Yan

2018 ◽  
Vol 8 (4) ◽  
pp. 34 ◽  
Author(s):  
Vishal Saxena ◽  
Xinyu Wu ◽  
Ira Srivastava ◽  
Kehan Zhu

The ongoing revolution in Deep Learning is redefining the nature of computing that is driven by the increasing amount of pattern classification and cognitive tasks. Specialized digital hardware for deep learning still holds its predominance due to the flexibility offered by the software implementation and maturity of algorithms. However, it is being increasingly desired that cognitive computing occurs at the edge, i.e., on hand-held devices that are energy constrained, which is energy prohibitive when employing digital von Neumann architectures. Recent explorations in digital neuromorphic hardware have shown promise, but offer low neurosynaptic density needed for scaling to applications such as intelligent cognitive assistants (ICA). Large-scale integration of nanoscale emerging memory devices with Complementary Metal Oxide Semiconductor (CMOS) mixed-signal integrated circuits can herald a new generation of Neuromorphic computers that will transcend the von Neumann bottleneck for cognitive computing tasks. Such hybrid Neuromorphic System-on-a-chip (NeuSoC) architectures promise machine learning capability at chip-scale form factor, and several orders of magnitude improvement in energy efficiency. Practical demonstration of such architectures has been limited as performance of emerging memory devices falls short of the expected behavior from the idealized memristor-based analog synapses, or weights, and novel machine learning algorithms are needed to take advantage of the device behavior. In this article, we review the challenges involved and present a pathway to realize large-scale mixed-signal NeuSoCs, from device arrays and circuits to spike-based deep learning algorithms with ‘brain-like’ energy-efficiency.


Energies ◽  
2019 ◽  
Vol 12 (13) ◽  
pp. 2582 ◽  
Author(s):  
Samuel Lotsu ◽  
Yuichiro Yoshida ◽  
Katsufumi Fukuda ◽  
Bing He

Confronting an energy crisis, the government of Ghana enacted a power factor correction policy in 1995. The policy imposes a penalty on large-scale electricity users, namely, special load tariff (SLT) customers of the Electricity Company of Ghana (ECG), whose power factor is below 90%. This paper investigates the impact of this policy on these firms’ power factor improvement by using panel data from 183 SLT customers from 1994 to 1997 and from 2012. To avoid potential endogeneity, this paper adopts a regression discontinuity design (RDD) with the power factor of the firms in the previous year as a running variable, with its cutoff set at the penalty threshold. The result shows that these large-scale electricity users who face the penalty because their power factor falls just short of the threshold are more likely to improve their power factor in the subsequent year, implying that the power factor correction policy implemented by Ghana’s government is effective.


Author(s):  
К.Н. Привалова ◽  
Р.Р. Каримов

Исследования по определению энергетической эффективности пастбищных систем со злаковыми и бобово-злаковыми травостоями проведены в Федеральном научном центре кормопроизводства и агроэкологии им. В. Р. Вильямса. В статье приведены результаты агроэнергетической оценки многовариантных пастбищных систем со злаковыми травостоями, созданными в 1946 году. Даны количественные показатели по сбору обменной энергии, совокупным затратам на её производство, окупаемости затрат в зависимости от системы ведения пастбищ. Изучена эффективность совокупных затрат в виде овеществлённого труда (на семена, удобрения, сельскохозяйственные машины, средства огораживания загонов и прочее) и живого труда (работы трактористов, пастухов и строителей и др.). Обоснована высокая агроэнергетическая эффективность изучаемых пастбищных систем благодаря мобилизации в продукционный процесс природных факторов, долевое участие которых в структуре производства обменной энергии составило 69–84%. Природные факторы, участвующие в продукционном процессе луговых агроэкосистем, характеризуются большим разнообразием. Это не только использование солнечной энергии и азотфиксация бобовыми травами, но и долголетие травостоев, самовозобновление фитоценозов, дерновообразовательный процесс (повышение плодородия почвы), получение дешёвого корма и улучшение здоровья животных при летнем выпасе. Роль возобновляемых природных факторов выявлена на основе балансового метода, принятого в экономике (по разнице сбора обменной энергии и антропогенных затрат). Благодаря ведущей роли природных факторов в структуре произведённой продукции агроэнергетический коэффициент окупаемости совокупных затрат антропогенной энергии (АК) за счёт сбора обменной энергии достигал 3–6 раз в среднем за 45 лет. Разработанные в результате долголетних исследований многовариантные энергосберегающие пастбищные системы обосновывают возможность рекомендовать их производству с учётом применения различного уровня энергозатрат. Ключевые слова: культурные пастбища, системы ведения, долголетние травостои, сбор обменной энергии, совокупные антропогенные затраты, окупаемость затрат. The investigation was conducted at the Federal Williams Research Center of Fodder Production and Agroecology and was aimed at testing energy efficiency of gramineous and legume-gramineous swards. This article presents the results obtained on pasture ecosystems with gramineous planted in 1946. Exchange energy yield, total production costs and economic effectiveness were analyzed. Total production costs comprised costs for seeds, fertilizers, machinery, construction materials, labor, etc. Introduction of natural factors into the production process resulted in higher energy efficiency. Their share amounted to 69–84% in the final exchange energy yield. There are a lot of natural factors that affect grass productivity such as solar energy, nitrogen-fixation, sward longevity and regeneration, soil fertility, low-cost feed production, and livestock health. The value of natural factors was determined according to the balance method (by the difference between exchange energy yield and anthropogenic costs). Since environmental factors had a leading role in the production process, the return rate raised by 3–6 times for 45 years due to exchange energy increase. Therefore, pasture ecosystems developed can be recommended for a large-scale forage production.


Sign in / Sign up

Export Citation Format

Share Document