scholarly journals Thread-Level CPU Power Measurement for High Performance Parallel Systemss: Impact Analysis of System Control Parameters on HPC Energy Efficiency

2016 ◽  
Vol 5 (3) ◽  
pp. 189-209
Author(s):  
David K. Newsom, et.al.
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.


2006 ◽  
Vol 290 (5) ◽  
pp. H1818-H1825 ◽  
Author(s):  
Lisa M. Smyth ◽  
Leanne T. Breen ◽  
Violeta N. Mutafova-Yambolieva

Using high-performance liquid chromatography techniques with fluorescence and electrochemical detection, we found that β-nicotinamide adenine dinucleotide (β-NAD) is released in response to electrical field stimulation (4–16 Hz, 0.3 ms, 15 V, 120 s) along with ATP and norepinephrine (NE) in the canine isolated mesenteric arteries. The release of β-NAD increases with number of pulses/stimulation frequencies. Immunohistochemistry analysis showed dense distribution of tyrosine hydroxylase-like immunoreactivity (TH-LI) and sparse distribution of TH-LI-negative nerve processes, suggesting that these blood vessels are primarily under sympathetic nervous system control with some contribution of other (e.g., sensory) neurons. Exogenous NE (3 μmol/l), α,β-methylene ATP (1 μmol/l), neuropeptide Y (NPY, 0.1 μmol/l), CGRP (0.1 μmol/l), vasoactive intestinal peptide (VIP, 0.1 μmol/l), and substance P (SP, 0.1 μmol/l) had no effect on the basal release of β-NAD, suggesting that the overflow of β-NAD is evoked by neither the sympathetic neurotransmitters NE, ATP, and NPY, nor the neuropeptides CGRP, VIP, and SP. Botulinum neurotoxin A (BoNTA, 0.1 μmol/l) abolished the evoked release of NE, ATP, and β-NAD at 4 Hz, suggesting that at low levels of neural activity, release of these neurotransmitters results from N-ethylmaleimide-sensitive factor attachment protein receptor/synaptosomal-associated protein of 25 kDa-mediated exocytosis. At 16 Hz, however, the evoked release of NE, ATP, and β-NAD was reduced by BoNTA by ∼90, 60, and 80%, respectively, suggesting that at higher levels of neural activity, β-NAD is likely to be released from different populations of synaptic vesicles or different populations of nerve terminals (i.e., sympathetic and sensory terminals).


Energies ◽  
2018 ◽  
Vol 12 (1) ◽  
pp. 27 ◽  
Author(s):  
Linfei Hou ◽  
Liang Zhang ◽  
Jongwon Kim

To improve the energy efficiency of a mobile robot, a novel energy modeling method for mobile robots is proposed in this paper. The robot can calculate and predict energy consumption through the energy model, which provides a guide to facilitate energy-efficient strategies. The energy consumption of the mobile robot is first modeled by considering three major factors: the sensor system, control system, and motion system. The relationship between the three systems is elaborated by formulas. Then, the model is utilized and experimentally tested in a four-wheeled Mecanum mobile robot. Furthermore, the power measurement methods are discussed. The energy consumption of the sensor system and control system was at the milliwatt level, and a Monsoon power monitor was used to accurately measure the electrical power of the systems. The experimental results showed that the proposed energy model can be used to predict the energy consumption of the robot movement processes in addition to being able to efficiently support the analysis of the energy consumption characteristics of mobile robots.


Sign in / Sign up

Export Citation Format

Share Document