检索结果-内蒙古大学图书馆

massively parallel computing: a statistical application

STATISTICS AND computing 1998年第4期8卷 309-318页

作者： Benn, A Kulperger, R Univ Western Ontario Dept Stat & Actuarial Sci London ON N6A 5B7 Canada

massively parallel computing is a computing environment with thousands of subprocessors. It requires some special programming methods, but is well suited to certain imaging problems. One such statistical example is discussed in this paper. In addition there are other natural statistical problems for which this technology is well suited. This paper describes our experience, as statisticians, with a massively parallel computer in a problem of image correlation spectroscopy. Even with this computing environment some direct computations would still take in the order of a year to finish. It is shown that some of the algorithms of interest can be made parallel.

关键词： massively parallel computing image correlation spectroscopy Fourier transform periodogram inference parallel algorithms

来源：评论

学校读者我要写书评

暂无评论

Brain big data processing with massively parallel computing technology: challenges and opportunities

引用

SOFTWARE-PRACTICE & EXPERIENCE 2017年第3期47卷 405-420页

作者： Chen, Dan Hu, Yangyang Cai, Chang Zeng, Ke Li, Xiaoli Wuhan Univ State Key Lab Software Engn Wuhan 430072 Peoples R China China Univ Geosci Sch Comp Sci Wuhan 430074 Peoples R China Beijing Normal Univ Natl Key Lab Cognit Neurosci & Learning Beijing 100875 Peoples R China Collaborat & Innovat Ctr Educ Technol Beijing Peoples R China

Brain data processing has been embracing the big data era driven by the rapid advances of neuroscience as well as the experimental techniques for recording neuronal activities. Processing of massive brain data has become a constant in neuroscience research and practice, which is vital in revealing the hidden information to better understand the brain functions and malfunctions. Brain data are routinely non-linear and non-stationary in nature, and existing algorithms and approaches to neural data processing are generally complicated in order to characterize the non-linearity and non-stationarity. Brain big data processing has pressing needs for appropriate computing technologies to address three grand challenges: (1) efficiency, (2) scalability and (3) reliability. Recent advances of computing technologies are making non-linear methods viable in sophisticated applications of massive brain data processing. General-purpose computing on the Graphics Processing Unit (GPGPU) technology fosters an ideal environment for this purpose, which benefits from the tremendous computing power of modern graphics processing units in massively parallel architecture that is frequently an order of magnitude larger than the modern multi-core CPUs. This article first recaps significant speed-ups of existing algorithms aided by GPGPU in neuroimaging and processing electroencephalogram (EEG), functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG) and etc. The article then demonstrates a series of successful approaches to processing EEG data in various dimensions and scales in a massively parallel manner: (1) decomposition: a massively parallel Ensemble Local Mean Decomposition (ELMD) algorithm aided by GPGPU can decompose EEG series, which forms the basis of further time-frequency transformation, in real-time without sacrificing the precision of processing;(2) synchronization measurement: a parallelized Nonlinear Interdependence (NLI) method for global synchronization mea

关键词： big data brain data neural signal processing massively parallel computing generalpurpose computing on the graphics processing unit (GPGPU)

来源：评论

学校读者我要写书评

暂无评论

Fault tolerant memory design for HW/SW co-reliability in massively parallel computing systems 2

Fault tolerant memory design for HW/SW co-reliability in mas...

引用

2nd IEEE International Symposium on Network computing and Applications

作者： Choi, M Park, NJ George, KM Jin, B Park, N Kim, YB Lombardi, F Univ Missouri Dept Elect & Comp Engn Rolla MO 65409 USA

ISBN: (纸本)0769519385

A highly dependable embedded fault-tolerant memory architecture for high performance massively parallel computing applications and its dependability assurance techniques are proposed and discussed in this paper. The proposed fault tolerant memory provides two distinctive repair mechanisms: the permanent laser redundancy reconfiguration during the wafer probe stage in the factory to enhance its manufacturing yield and the dynamic BIST/BISD/BISR (built-in-self-test-diagnosis-repair)-based reconfiguration of the redundant resources in field to maintain high field reliability. The system reliability which is mainly determined by hardware configuration demanded by software and field reconfiguration/repair utilizing unused processor and memory modules is referred to as HW/SW Co-reliability. Various system configuration options in terms of parallel processing unit size and processor/memory intensity are also introduced and their HW/SW Co-reliability characteristics are discussed. A modeling and assurance technique for HW/SW Co-reliability with emphasis on the dependability assurance techniques based on combinatorial modeling suitable for the proposed memory design is developed and validated by extensive parametric simulations. Thereby, design and implementation of memory-reliability-optimized and highly reliable fault-tolerant field reconfigurable massively parallel computing systems can be achieved.

关键词： high performance computing massively parallel computing fault-tolerant memory built-in-self-repair (BISR) yield reliability field reconfiguration modular redundancy HW/SW co-reliability reliability assurance

来源：评论

学校读者我要写书评

暂无评论

Development of a Reduced-Order Design/Optimization Tool for Automotive Engines Using massively parallel computing

引用

SAE Technical Papers 2015年 2015卷

作者： Aithal, Shashi Wild, Stefan Argonne National Laboratory United States

Design and optimization of automotive engines present unique challenges on account of the large design space and conflicting constraints. A notable example of such a problem is optimizing the fuel consumption and reducing emissions over the drive cycle of an automotive engine. There are over twenty design variables (including operating conditions and geometry) for the above-mentioned problem. Conducting design, analyses, and optimization studies over such a large parametric space presents a serious computational challenge. The large design parameter space precludes the use of detailed numerical or experimental investigations. Physics-based reduced-order models can be used effectively in the design and optimization of such problems. Since a typical drive cycle is represented by 1500 to 2000 sample data points (engine cycles), it is essential to develop fast and robust computations so that the entire engine cycle computation is done close to real-time speeds (on the order of 100-150 milliseconds). Harnessing the power of high-performance computing, it is possible to perform optimization of automotive drive cycles using massively parallel computations. In this work, we discuss the development of a parallel fast and robust reduced-order modeling tool to compute integrated quantities such as fuel consumption and emissions (NO and CO) over a range of engine drive cycles. As an illustrative example, we perform a massively parallel simulation consisting of 4096 synthetic drive cycles, representative of a fleet of cars. The impact of parameters such as humidity, initial cylinder pressure, inlet air temperature, and residual gas fraction on the performance and emission are presented. Copyright © 2015 SAE International.

关键词： Automobile engines Digital storage Computational challenges Design and optimization Experimental investigations High performance computing Inlet air temperatures massively parallel computing Performance and emissions Reduced order models

来源：评论

学校读者我要写书评

暂无评论

Position: Some Thoughts on the Future of parallel computing and Connectionist massively parallel Models

Position: Some Thoughts on the Future of Parallel Computing ...

引用

Future Technologies Conference (FTC)

作者： Tosic, Predrag T. Washington State Univ Sch EECS Pullman WA 99164 USA

ISBN: (纸本)9781509041718

We discuss the future of massively parallel computing from a fundamental architecture standpoint. Our central thesis is that various versions of Moore's Laws will all unavoidably break down over the next two to three decades, due to fundamental limitations imposed by the laws of physics (especially quantum mechanics). Therefore, the end to scaling-up von Neumann-based architectures by adding more cores and other incremental technology advancements is within sight. How are these challenges to be overcome, in order to continue making computation faster, cheaper, more energy efficient and more reliable? One possible answer lies in connectionist massively parallel models, based on (artificial) neural networks and other paradigms from biology, neuroscience and statistical physics. We discuss what would such massively parallel connectionist computers be like if they were to be based on "implementing neural networks in the hardware", and undertake several thought experiments comparing and contrasting the hypothetical connectionist parallel machines with the "classical" multicore supercomputers and large-scale distributed computing systems.

关键词： massively parallel computing Moore's Law connectionist computing models artificial neural networks fine-grained vs. coarse-grained parallelism

来源：评论

学校读者我要写书评

暂无评论

Position: Some Thoughts on the Future of parallel computing and Connectionist massively parallel Models

Position: Some Thoughts on the Future of Parallel Computing ...

引用

Future Technologies Conference

作者： Predrag T. Tosic School of EECS Washington State University

ISBN: (纸本)9781509041725

关键词： massively parallel computing Moore's Law Connectionist computing models Artificial neural networks Fine-grained vs. coarse-grained parallelism

来源：评论

学校读者我要写书评

暂无评论

Reliable Finite-Difference Time-Domain Simulations of Reverberation Chambers by Using Equivalent Volumetric Losses

引用

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY 2016年第3期58卷 653-660页

作者： Moglie, Franco Bastianelli, Luca Primiani, Valter Mariani Univ Politecn Marche Dipartimento Ingn Informaz I-60131 Ancona Italy

A numerical code based on the finite-difference time-domain technique is adopted to simulate a large reverberation chamber including the stirrers and the actual antenna presence in the frequency range 200-1000 MHz. The power of the code lies in the parallelization that allows the execution on high-performance computers. Chamber losses are accounted for by introducing virtual volumetric losses in the hypothesis of ideal conductor walls. The correction that arises from this choice which allows us to speed up the numerical computation is verified by a comparison with the case of wall losses implemented by assuming a finite conductivity. The analysis of the statistics of the chamber field, stirrer uncorrelated positions, Rician K-factor, backscatter, and field uniformity returns overlapping results for the two techniques. The analysis of the spatial field correlation within a spatial grid shows a correlation greater than 0.5 in the whole frequency range and for all stirrer angles, so confirming the equivalence of the two methods to manage losses.

关键词： Correlation field statistics finite-difference time-domain (FDTD) massively parallel computing reverberation chamber (RC) volumetric losses wall losses

来源：评论

学校读者我要写书评

暂无评论

parallel Lattice Boltzmann computing and Applications in Core Sample Feature Evaluation

引用

TRANSPORT IN POROUS MEDIA 2015年第1期107卷 65-77页

作者： Gao, Jinfang Xing, Huilin Rudolph, Victor Li, Qin Golding, Sue D. Univ Queensland Sch Earth Sci Brisbane Qld 4072 Australia Univ Queensland Sch Chem Engn Brisbane Qld 4072 Australia Griffith Univ Environm Engn Nathan Qld 4111 Australia Griffith Univ Queensland Micro & Nanotechnol Ctr Nathan Qld 4111 Australia

Micro-CT scans with QEMSCAN mapping provide visualization of core samples to quantify heterogeneous physical properties important for subsurface flow including, as examples, pore size distribution and connectivity, mineral compositions, porosity and permeability, amongst many others. 3D high-resolution micro-CT scans can deliver a very high level of microstructures detail, which also implies enormous numerical data sets and associated computational processing load. It is, therefore, important to understand (1) the voxel resolution of micro-CT scans required to retain physical structure fidelity (e.g., mineral compositions, pore size and throats, and porosity and tortuosity), (2) the sensitivity of individual mineral property and voxel resolution on the directional permeability, and (3) the smallest sample size that provides reliable and representative transport calculations (e.g., directional permeability and connectivity). The lattice Boltzmann method is capable of simulating flow in both open pore spaces and porous media and is used here to allow for flow in multiple matrices (quartz aggregate and low-permeable clay matrix). As an application example, a permeability study of the Precipice Sandstone from the Chinchilla 4 well in the Surat Basin has been conducted. Regarding the Chinchilla sample, we established that (1) the composition ratio is relatively sensitive to voxel resolution: higher resolution imaging is required to retain narrow pore throats and excessively coarsened voxel resolutions result in severe loss of internal microstructures information;(2) both voxel resolutions and individual mineral properties affect flow dynamics, and the clay permeability slightly affects the whole permeability at nD scale;(3) it is not feasible to define the accurate ratio of lattice nodes versus pore apertures for meeting the grid independence due to the complex sample tortuosity;and (4) the directional permeability approaches a constant value as the sample size increases

关键词： 3D heterogeneous porous media Fluid transport massively parallel computing Lattice Boltzmann method Representative elementary volume

来源：评论

学校读者我要写书评

暂无评论

Improving the performance scalability of the community atmosphere model

引用

INTERNATIONAL JOURNAL OF HIGH PERFORMANCE computing APPLICATIONS 2012年第1期26卷 17-30页

作者： Mirin, Arthur A. Worley, Patrick H. Lawrence Livermore Natl Lab Ctr Appl Sci Comp Sci Comp Grp Livermore CA 94551 USA Oak Ridge Natl Lab Comp Sci & Math Div Oak Ridge TN USA

The Community Atmosphere Model (CAM), which serves as the atmosphere component of the Community Climate System Model (CCSM), is the most computationally expensive CCSM component in typical configurations. On current and next-generation leadership class computing systems, the performance of CAM is tied to its parallel scalability. Improving performance scalability in CAM has been a challenge, due largely to algorithmic restrictions necessitated by the polar singularities in its latitude-longitude computational grid. Nevertheless, through a combination of exploiting additional parallelism, implementing improved communication protocols, and eliminating scalability bottlenecks, we have been able to more than double the maximum throughput rate of CAM on production platforms. We describe these improvements and present results on the Cray XT5 and IBM BG/P. The approaches taken are not specific to CAM and may inform similar scalability enhancement activities for other codes.

关键词： atmospheric modeling CAM massively parallel computing performance scalability

来源：评论

学校读者我要写书评

暂无评论

Circular Interval-valued Computers and Simulation of (Red-green) Turing Machines

引用

FUNDAMENTA INFORMATICAE 2021年第2-3期181卷 213-238页

作者： Nagy, Benedek Valyi, Sandor Eastern Mediterranean Univ Dept Math Via Mersin 10 Famagusta North Cyprus Turkey Univ Nyiregyhaza Inst Math & Informat Nyiregyhaza Hungary

Interval-valued computing is a kind of massively parallel computing. It operates on specific subsets of the interval [0,1) - unions of subintervals. They serve as basic data units and are called interval-values. It was established in [9], by a rather simple observation, that interval-valued computing, as a digital computing model, has computing power equivalent to Turing machines. However, this equivalence involves an unlimited number of interval-valued variables. In [14], the equivalence with Turing machines is established using a simulation that uses only a fixed number of interval-valued variables and this number depends only on the number of states of the Turing machine - in a logarithmic way. The simulation given there allows us to extend interval-valued computations into infinite length to capture the computing power of redgreen Turing machines. In this extension of [14], based on the quasi-periodic techniques used in the simulations in that paper, a reformulation of the interval-valued computations is given, named circular interval-valued computers. This reformulation enforces the finiteness of the number of used interval-valued variables by building the finiteness into the syntax rules.

关键词： unconventional computing massively parallel computing interval-valued computing red-green Turing machines simulation hypercomputation

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：