عنوان فارسی مقاله:
الگوریتم تعیین اندازه نمونه برای سیستم های تعیین موقعیت داخلی مبتنی بر اثر انگشت
عنوان انگلیسی مقاله:
Sample Size Determination Algorithm for fingerprint-based indoor localization systems
سال انتشار : 2016
عنوان فارسی مقاله:
برنامه ریزی لینک های بی سیم توسط راس مولتی کالرینگ در مدل تداخل فیزیکی
عنوان انگلیسی مقاله:
Scheduling wireless links by vertex multicoloring in the physical interference model
سال انتشار : 2016
عنوان فارسی مقاله:
توزیع قدرت بهینه در شبکه های بی سیم غیر باینری مبتنی بر کد LDPC
عنوان انگلیسی مقاله:
Optimal power distribution in non-binary LDPC code-based cooperative wireless networks
سال انتشار : 2016
عنوان فارسی مقاله:
طرح تخصیص منابع مبتنی بر ارزش شپلی مجانبی برای خدمات اینترنت اشیا
عنوان انگلیسی مقاله:
Asymptotic shapley value based resource allocation scheme for IoT services
سال انتشار : 2016
عنوان فارسی مقاله:
هیبرید اکتشافی برای مساله مسیریابی تحت قیمت گذاری منطقه ای پویا
عنوان انگلیسی مقاله:
A hybrid heuristic for the inventory routing problem under dynamic regional pricing
سال انتشار : 2016
عنوان فارسی مقاله:
یک الگوریتم موازی سازگار برای شبیه سازی جریان عددی مستقیم آشفته تراکم ناپذیر با استفاده از روش تبدیل فوریه طیفی / اسب بخار عنصر و توپولوژی مجازی MPI
عنوان انگلیسی مقاله:
An adaptable parallel algorithm for the direct numerical simulation of incompressible turbulent flows using a Fourier spectral/hp element method and MPI virtual topologies
سال انتشار : 2016
عنوان فارسی مقاله: |
بهره وری از انرژی در شبکه های سنسوری بی سیم |
| عنوان انگلیسی مقاله: |
Energy-Efficiency in Wireless Sensor Networks: a top-down review approach |
عنوان فارسی مقاله:
پالایش مش تطبیقی بر اساس سلول واحد پردازش گرافیکی پرشتاب در گرید چهار ضلعی بدون ساختار
عنوان انگلیسی مقاله:
GPU accelerated cell-based adaptive mesh refinement on unstructured quadrilateral grid
سال انتشار : 2016
برای دانلود رایگان مقاله مش تطبیقی بر اساس سلول پردازش گرافیکی در گرید چهار ضلعی اینجا کلیک نمایید.
مقدمه انگلیسی مقاله:
1. Introduction
In recent years, the GPU (Graphics Processing Unit), once usedonly for graphics processing, has been extended to general purposecomputing for its high computing power and bandwidth. Someearly researches adopted graphics programming languages such asCg, OpenGL to accelerate particle algorithms [1–3]. These worksshowed great potential of using GPU for scientific computing. Butcoding for scientific computing with these languages was difficultand the application fields were also limited. However, the developmentof general purpose computing on GPU has never stopped.NVIDIA Corporation released their parallel computing model calledCUDA (Compute Unified Device Architecture) for general purposecomputing in 2007 which provides an easy-to-use tool for scientificcomputing and is now widely used in many fields. Manyresearchers have used the tool in Computational Fluid Dynamics(CFD) and obtained remarkable results of performance increasing.Thibault et al. developed a Navier–Stokes solver for incompressibleflow on multi-GPU with a 2nd order accurate centraldifference scheme and achieved 100× speedup [4]. Bailey and his co-workers used CUDA for accelerating Lattice Boltzmann Methodon GPU and obtained remarkable performance enhancement [5].Frezzotti’s group adopted semi-regular methods to solve the Boltzmannequation on GPUs with high efficiency [6]. Ran et al. realizedthe GPU accelerated CESE method for 1D shock tube problems andachieved high acceleration ratios [7]. Brodtkorb et al. implementedshallow water simulations on GPUs and performed a detailed analysisof it [8]. Lutsyshyn presented a scheme for the parallelizationof quantum Monte Carlo method on GPU and the program wasbenchmarked on several models of NVIDIA GPUs [9].Implementing CFD method on GPU greatly depends on themesh type used. Compared with the structured counterpart, methodsbased on unstructured mesh cannot be efficiently acceleratedby GPU because the unstructured configuration leads to thenon-coalescent memory accessing on GPU. Some researchers madetheir efforts to overcome this difficulty. Corrigan et al. implementedan unstructured grid based Euler solver on GPU and obtaineda speedup of 33×’s [10]. Kampolis et al. accomplisheda GPU accelerated Navier–Stokes solver on unstructured grid inthe same year [11] and achieved a remarkable computing performanceincreasing. Waltz described the performance of CHICOMA,a 3D unstructured mesh compressible flow solver, on GPU and observedspeedup of 4–5× over single-CPU performance [12]. Laniet al. provided a GPU-enabled finite volume solver for ideal magnetohydrodynamicson unstructured grids within the COOLFluiD platform [13]. Almost all authors employed the renumberingtechnique to cope the problem of non-coalescent memory accessing,which has been discussed in detail in [14]. As demonstrated intheir works, with the renumbering technique, shared memory canbe introduced and therefore their codes’ performance is efficientlyimproved.
برای دانلود رایگان مقاله مش تطبیقی بر اساس سلول پردازش گرافیکی در گرید چهار ضلعی اینجا کلیک نمایید.
کلمات کلیدی:
GAMER: a GPU-Accelerated Adaptive Mesh Refinement Code for ...https://arxiv.org/pdf/0907.3390by HY Schive - 2009 - Cited by 92 - Related articlesDec 24, 2009 - GAMER: a GPU-Accelerated Adaptive Mesh Refinement Code for ... The AMR implementation is based on a hierarchy of grid patches with an oct-tree ... the mass density in each grid cell is estimated by the cloud-in-cell (CIC) ...Adaptive Kinetic-Fluid Solvers for Heterogeneous Computing ...https://arxiv.org/pdf/1503.00707by S Zabelok - 2015 - Cited by 13 - Related articlesAdaptive Mesh Refinement (AMR) with automatic cell-by-cell selection of kinetic or fluid ..... (GPU-accelerated Adaptive MEsh Refinement), which is based on a ...GPU accelerated cell-based adaptive mesh refinement on ... - Trovetrove.nla.gov.au/work/217096909?2016-10-01, English, Article, Journal or magazine article edition: GPU accelerated cell-based adaptive mesh refinement on unstructured quadrilateral grid.GAMER: A GRAPHIC PROCESSING UNIT ACCELERATED ...iopscience.iop.org/article/10.1088/0067-0049/186/2/457/metaby HY Schive - 2010 - Cited by 92 - Related articlesGAMER is a parallel code that can be run in a multi-GPU cluster system. .... GPU accelerated cell-based adaptive mesh refinement on unstructured quadrilateral ...GPU accelerated cell-based adaptive mesh refinement on ... - خانهzoodyab.ir/.../122880-gpu-accelerated-cell-based-adaptive-mesh-re... - Translate this pageFor the first time, the cell-based adaptive mesh refinement (AMR) is fully implemented on GPU for the unstructured quadrilateral grid, which greatly reduces the ...Improving Parallel IO Performance of Cell-based AMR Cosmology ...ieeexplore.ieee.org/document/6267900/by Y Yu - 2012 - Cited by 14 - Related articlesIn this study, we present a parallel IO design for cell-based AMR cosmology applications, ... aggregate small IO accesses per process to accelerate IO performance. ... various regions with different resolutions, adaptive mesh refinement (AMR) is .... GPU-based simulation of cellular neural networks for image processing.Data-centric GPU-based adaptive mesh refinement - ACM Digital Librarydl.acm.org/citation.cfm?id=2833179.2833181by M Wahib - 2015 - Cited by 2 - Related articlesNov 15, 2015 - The performance of two GPU-based AMR applications is enhanced by .... GAMER: a GPU-Accelerated Adaptive Mesh Refinement Code for ...
عنوان فارسی مقاله:
MaMR: مدل برنامه نویسی نگاشت کاهش با عملکرد بالا برای برنامه های کاربردی ابری مواد
عنوان انگلیسی مقاله:
MaMR: High-performance MapReduce programming model for material cloud applications
سال انتشار : 2016
برای دانلود رایگان مقاله برنامه نویسی نگاشت کاهش برای برنامه کاربردی ابری مواد اینجا کلیک نمایید.
مقدمه انگلیسی مقاله:
1. Introduction
In recent years, a large number of known and hypotheticalmaterials have been studied, such as batteries, catalysts, and thestable structures of solid materials, so the amount of calculateddata increases exponentially with time [1]. Thus, big data presentsa huge challenge to the computing disciplines.To improve computing speed, a large number of computingtasks in materials science must be moved from traditional HighPerformanceComputing (HPC) to High-Throughput Computing(HTC) and Many-Task Computing (MTC) platforms based on bigdata, such as the widely used cloud computing systems [2] and gridcomputing [3].As is known, cloud computing, which should provide differentlayers of service to users, is constituted by large-scale distributedcomputers and various resources such as CPUs and storage. Meanwhile,different types of services are also offered, such as software[4]. In cloud computing systems, multiple virtual machines (VMs) are run on a single physical computer, for a homogeneousresult [5].It is more convenient to develop and deploy applicationsthrough the cloud computing platform, Therefore, in this paper,we adopt this approach to process large amounts of materialdata. We use cloud computing’s powerful computation and storagecapacity to effectively solve the problem of large-scale materialdata in the process of analytical calculations. To fit the materialhigh-performance computing needs of materials science, differentmodels have been proposed to maximize the performance.Meanwhile, to provide greater flexibility and higher parallelefficiency, the challenges to the programming model should befaced [6]. The MapReduce programming model has been widelyused in large-scale and data-intensive applications, such as Googleand Amazon [7,8]. The other most successful programming modelis Microsoft’s Dryad [9]. Yahoo also has similar infrastructures.Hadoop is an open-source implementation of MapReduce, and ithas already been applied to various fields because of the reliabilityand scalability of the parallel programming framework in theMapReduce model [10]. In Hadoop and MapReduce, the input dataare split into chunks of size 64 M, and each task is allocated to a VM.Therefore, we can take the computation nodes as the local modes.However, traditional programming models cannot adapt tomaterial data calculations, and Hadoop’s frequent reading and writing become the bottleneck of the cloud system. First, in a cloudcomputing system, many computing jobs are running in a singlephysical computer because the data nodes in Hadoop are deployedin virtual machines. How to avoid I/O resource competition andreduce I/O overhead is a big problem in the programming modelsfor cloud computing.
برای دانلود رایگان مقاله برنامه نویسی نگاشت کاهش برای برنامه کاربردی ابری مواد اینجا کلیک نمایید.
کلمات کلیدی:
MaMR: High-performance MapReduce programming model for ... - خانهzoodyab.ir/.../157299-mamr-high-performance-mapreduce-progra...Translate this pageTo enhance the capability of workflow applications in material data processing, we defined a programming model for material cloud applications that supports ...[PDF]Large-Scale Image Classification using High Performance Clusteringdsc.soic.indiana.edu/.../Large-Scale_Image_Classification_using_High_Performance_...by B Zhang - Cited by 1 - Related articlesLarge-Scale Image Classification using High Performance Clustering ... MapReduce computation, leading to the iterative MapReduce programming .... histogram over this vocabulary: given an image, we densely sample patches, compute HOG ...... “PortLand: A Scalable Fault-Tolerant Layer 2 Data Center Network Fabric.”.[PDF]MapReduce for Scientist using FutureGrid - Digital Science Centerdsc.soic.indiana.edu/publications/CCGrid%20tutorial.pdfMapReduce programming model has simplified the implementations of many ... demonstrate Map reduce environments in FG utilizing traditional high-performance ... The tutorial will be available online as part of the FG educational material.Exploring performance models of Hadoop applications on cloud ...ieeexplore.ieee.org/document/7450806/by X Wu - 2015 - Cited by 3 - Related articlesAbstract: Hadoop is an open source implementation of the MapReduce programming model, and provides the runtime infrastructure for map and reduce ...A Coarse-Grained Reconfigurable Architecture for Compute-Intensive ...ieeexplore.ieee.org/document/7163277/by S Liang - 2016 - Related articlesAlthough processors such as GPGPUs and FPGAs show good performance of speedup, there is still vacancy for a low power, high efficiency and ... a dynamically reconfigurable acceleration to MapReduce-based (MR-based) applications. ... board, and a programming model with compilation flow for CGRA is presented.MapReduce - Wikipediahttps://en.wikipedia.org/wiki/MapReduceMapReduce is a programming model and an associated implementation for processing and generating big data sets with a parallel, distributed algorithm on a ...Big Data Tutorial 1: MapReduce - High Performance Computing at ...https://wikis.nyu.edu/display/NYUHPC/Big+Data+Tutorial+1%3A+MapReduceApr 12, 2017 - HDFS provides high throughput access to application data and is suitable for ... MapReduce is a programming model and an associated ...
عنوان فارسی مقاله:
اجرای GPU روش قطعه سه بعدی مقیاس پذیری خطی برای محاسبات ساختار الکترونی در مقیاس بزرگ
عنوان انگلیسی مقاله:
GPU implementation of the linear scaling three dimensional fragment method for large scale electronic structure calculations
سال انتشار : 2016
برای دانلود رایگان مقاله محاسبه ساختار الکترونی اینجا کلیک نمایید.
مقدمه انگلیسی مقاله:
1. Introduction
To calculate the physical properties of many real materials,large systems containing thousands or tens of thousands of atomsare often necessary. For example, a 5 nm quantum dot cancontain 5 thousand atoms. To describe the fluctuation of thedipole moment in (CH3NH3) PbI3, a 20,000 atom system have beenused [1]. An even larger system might be necessary to study themechanical properties and the electronic structure consequence ofdislocations. Although density functional theory (DFT) calculationsare necessary for many of such problems, conventional DFTcalculations cannot be used due to their O(N3) scaling of thesystem size N [2]. To solve this problem, various types of linearscaling methods have been developed. One particular approachof the linear scaling method is based on the divide-and-conquerstrategy. In this approach, a global system is divided into manysmall (fragment) systems, and each fragment is solved quantummechanically. After all the small fragment systems have beensolved, their results are combined together to yield the resultof the global system. Iteration loops can be used to ensure selfconsistencybetween the global charge density and the fragment potentials. A major advantage of this divide-and-conquer approachis the possibility to use different computer process groups to solvedifferent fragments. Since no communication is needed betweendifferent process groups for the computationally most expensivestep (the quantum mechanical fragment calculation), the weakscaling of this algorithm can be extremely good. Thus, a dual linearscaling, one to the system size, one to the number of computerprocesses can be achieved.Linear scaling three dimensional fragment (LS3DF) code isa Gordon Bell winning code based on a divide-and-conquermethod [3]. It can be scaled efficiently to hundreds of thousandsof processes. It has been used to calculate many nanostructureproblems, including the dipole moment of nanorod [4]; thelocalized state in random alloy; the effects of MoSe2/MoS2 Moire’spattern [5]; the ferroelectric vortex structure [6]; the electronicstructure of disordered (CH3NH3) PbI3 [1]. Nevertheless, basedon CPU, even though tens of thousands of processes (e.g., 60,000processes) have been used, such calculations can often take severalhours. This makes such calculations computationally expensive.
برای دانلود رایگان مقاله محاسبه ساختار الکترونی اینجا کلیک نمایید.
کلمات کلیدی:
PDF]GPU implementation of the linear scaling three dimensional fragment ...iranarze.ir/wp-content/uploads/2016/12/E994.pdfJul 9, 2016 - GPU implementation of the linear scaling three dimensional fragment method for large scale electronic structure calculations. Weile Jiaa,b, Jue ...GPU implementation of the linear scaling three dimensional ... - INFONAhttps://www.infona.pl/.../bwmeta1.element.elsevier-594c6841-4ebf-39a3-a00e-dff13c...by W Jia - 2017 - Related articlesLS3DF, namely linear scaling three-dimensional fragment method, is an efficient linear scaling ab initio total energy electronic structure calculation code based ...GPU implementation of the linear scaling three dimensional fragment ...https://www.bibsonomy.org/bibtex/fe1401f47803004ca10747793868be3e@article{journals/cphysics/JiaWCW17, added-at = {2016-12-09T00:00:00.000+0100}, author = {Jia, Weile and Wang, Jue and Chi, Xuebin and Wang, ...GPU implementation of the linear scaling three dimensional fragment ...https://www.semanticscholar.org/.../GPU-implementation-of-the-linear-scaling-three...Semantic Scholar extracted view of "GPU implementation of the linear scaling three dimensional fragment method for large scale electronic structure ...Electronic Structure Calculations on Graphics Processing Units: From ...https://books.google.com/books?isbn=1118661788Ross C. Walker, Andreas W. Goetz - 2016 - ScienceHere we illustrate the practical application of the GPU algorithms discussed previously in real-world calculations. ... scaling of our GPU implementation with increasing system size, we consider two types of systems: First, linear alkene chains ... provide a dense three-dimensional test system, which is more representative of ...GPU Solutions to Multi-scale Problems in Science and Engineeringhttps://books.google.com/books?isbn=3642164056David A. Yuen, Long Wang, Xuebin Chi - 2013 - ComputersWith periodic boundary condition imposed on all three directions, Eq. 18.7 can be reduced to a set of ... for each pair of ( κx ,κy ) , where the linear equations are of size Nz (number of grid points along z direction. ... The three-dimensional computational domain is set as a cubic box with each side as 2π. ... Implementation. of.
