检索结果-内蒙古大学图书馆

3rd ACM SIGPLAN Workshop on Functional high-performance computing (FHPC)

作者： Hargreaves, Felix P. Merkle, Daniel Schneider-Kamp, Peter Univ Southern Denmark Dept Math & Comp Sci Odense Denmark

ISBN: (纸本)9781450330404

We developed a Functional Object-Oriented PARallel framework (FOOPAR) for high-level high-performance computing in Scala. Central to this framework are Distributed Memory Parallel Data structures (DPDs), i.e., collections of data distributed in a shared nothing system together with parallel operations on these data. In this paper, we first present FOOPAR's architecture and the idea of DPDs and group communications. Then, we show how DPDs can be implemented elegantly and efficiently in Scala based on the Traversable/Builder pattern, unifying Functional and Object-Oriented Programming. We prove the correctness and safety of one communication algorithm and show how specification testing (via ScalaCheck) can be used to bridge the gap between proof and implementation. Furthermore, we show that the group communication operations of FOOPAR outperform those of the MPJ Express open source MPI-bindings for Java, both asymptotically and empirically. FOOPAR has already been shown to be capable of achieving close-to-optimal performance for dense matrix-matrix multiplication via JNI. In this article, we present results on a parallel implementation of the Floyd-Warshall algorithm in FOOPAR, achieving more than 94% efficiency compared to the serial version on a cluster using 100 cores for matrices of dimension 38000 x 38000.

关键词： high-performance-computing Scala performance Design Languages

来源：评论

学校读者我要写书评

暂无评论

high-performance computing Implementation for ADS-IDAC 18

High-Performance Computing Implementation for ADS-IDAC

引用

18th International Probabilistic Safety Assessment and Analysis, PSA 2023

作者： Tezbasaran, Alp Diaconeasa, Mihai A. North Carolina State University RaleighNC United States

ISBN: (纸本)9780894487927

Unlike classical Probabilistic Risk Assessment (PRA) methods, Dynamic Probabilistic Risk Assessment (D-PRA) leverages simulations to estimate the timings of pivotal events and evaluate the system/human interactions during transients and corresponding outcomes of the perturbations made to resolve the success and failure of a system such as nuclear power plants. We must address the uncertainties inherent to simulations to determine the success and failure sequences. These uncertainties stem from the nature of a transient. An operator choosing to (or not to) perform a specific action during an accident might be the deciding factor in preventing a reactor from melting. But before simulating both scenarios, it is unknown how the transient will end. Calculating each possible outcome means creating two branches, one with action performed and one not. A full-scope D-PRA study requires hundreds to thousands of sequences simulated through rigorous branching. ADS-IDAC is the D-PRA tool currently partially developed and maintained by the Probabilistic Risk Assessment Group of the NCSU Nuclear Engineering department. It is capable of advanced branching mechanisms and can create branches dynamically when pivotal events happen without the need to pre-define failures of components. It is currently coupled with RELAP5/MOD3.3 as the thermal-hydraulics simulation tool and can also model human reliability. Additionally, ADS-IDAC can model reactor components not in the simulation tool RELAP5 through fault trees. Many capabilities of ADS-IDAC allow analysts and researchers to perform a comprehensive D-PRA study with advanced branching mechanisms. Still, as mentioned earlier, many thermal-hydraulics runs are required to resolve the success and failure sequences in detail. ADS-IDAC can use the restart information of RELAP5. Restart capability allows ADS-IDAC to simulate the remainder of the newly created branches from the branch point in time rather than executing the entire transient from

关键词： ADS-IDAC Dynamic PRA high-performance-computing MPI

来源：评论

学校读者我要写书评

暂无评论

GPU-DAEMON: GPU algorithm design, data management & optimization template for array based big omics data

引用

COMPUTERS IN BIOLOGY AND MEDICINE 2018年 101卷 163-173页

作者： Awan, Muaaz Gul Eslami, Taban Saeed, Fahad Western Michigan Univ Dept Comp Sci Kalamazoo MI 49008 USA Florida Int Univ Sch Comp & Informat Sci Miami FL 33199 USA

In the age of ever increasing data, faster and more efficient data processing algorithms are needed. Graphics Processing Units (GPU) are emerging as a cost-effective alternative architecture for high-end computing. The optimal design of GPU algorithms is a challenging task which requires thorough understanding of the high performance computing architecture as well as the algorithmic design. The steep learning curve needed for effective GPU-centric algorithm design and implementation requires considerable expertise, time, and resources. In this paper, we present GPU-DAEMON, a GPU Data Management, Algorithm Design and Optimization technique suitable for processing array based big omics data. Our proposed GPU algorithm design template outlines and provides generic methods to tackle critical bottlenecks which can be followed to implement high performance, scalable GPU algorithms for given big data problem. We study the capability of GPU-DAEMON by reviewing the implementation of GPU-DAEMON based algorithms for three different big data problems. Speed up of as large as 386x (over the sequential version) and 50x (over naive GPU design methods) are observed using the proposed GPU-DAEMON. GPU-DAEMON template is available at https://***/pcdslab/GPU-DAEMON and the source codes for GPU-ArraySort, G-MSR and GPU-PCC are available at https://***/pcdslab.

关键词： GPU CUDA high-performance-computing Big-data Omics-data

来源：评论

学校读者我要写书评

暂无评论

MULTIGRID REDUCTION IN TIME WITH RICHARDSON EXTRAPOLATION

引用

ELECTRONIC TRANSACTIONS ON NUMERICAL ANALYSIS 2021年 54卷 210-233页

作者： Falgout, R. D. Manteuffel, T. A. O'Neill, B. Schroder, J. B. Lawrence Livermore Natl Lab Ctr Appl Sci Comp POB 808L-561 Livermore CA 94551 USA Univ Colorado Dept Appl Math Boulder CO 80309 USA Univ New Mexico Dept Math & Stat Albuquerque NM 87131 USA

The advent of exascale computing will leave many users with access to more computational resources than they can simultaneously use, e.g., billion-way parallelism. In particular, this is true for time-dependent simulations that limit parallelism to the spatial domain. One method to add parallelism in time to existing simulation codes and thus take advantage of ever larger compute resources is Multigrid Reduction in Time (MGRIT). The goal is to achieve a smaller time-to-solution through parallelism in time. In this paper, MGRIT is enhanced with Richardson extrapolation in a cost-efficient way to produce a parallel-in-time method with improved accuracy. Overall, this leads to a large improvement in the accuracy per computational cost of MGRIT.

关键词： parallel time integration high-performance-computing multigrid-reduction-in-time extrapolation-methods

来源：评论

学校读者我要写书评

暂无评论

A Case Study: Test-Driven Development in a Microscopy Image-Processing Project 4

A Case Study: Test-Driven Development in a Microscopy Image-...

引用

4th International Workshop on Software Engineering for high performance computing in Computational Science and Engineering (SE-HPCCSE)

作者： Nanthaamornphong, Aziz Prince Songkla Univ Fac Technol & Environm Dept Informat & Commun Technol Phuket Campus Phuket Thailand

ISBN: (纸本)9781509052240

Many scientific software developers have applied software engineering practices to their work in recent years. Agile methods are gaining increased interest from both industry and research fields, including scientific application domains. Test-driven development (TDD) and refactoring practices are critical for the success of agile methods. Although many scientific projects employ agile practices, the effect of TDD on scientific software development remains unknown and should thus be investigated. The objective of this study was to investigate the effects of using TDD to develop scientific software in a high-performance- computing (HPC) environment. The results show that the TDD provides both advantages and disadvantages in a parallel computing environment. In particular, developers face problems when writing tests and there is a lack of software engineering practices.

关键词： Test-driven development scientific software empirical study high-performance-computing

来源：评论

学校读者我要写书评

暂无评论

HPC in a Vacuum: Evaluating Future Space Microprocessors 14

HPC in a Vacuum: Evaluating Future Space Microprocessors

引用

14th IEEE Space computing Conference (SCC)

作者： Gauthier, Thea-Martine CALTECH Jet Prop Lab Pasadena CA 91125 USA

ISBN: (纸本)9798350341430

Advanced algorithms dictated by future mission needs are pushing space-borne computing requirements. Current platforms used in manned and unmanned spaceflight are limited by the intersection of radiation tolerance, power consumption, computing performance and safety critical features. Until recently, advanced instruction set architectures (ISAs), algorithm specific instructions, high speed external interfaces and high performance on-chip networks were eschewed from processor designs and current production spacecraft processors are based on past computing paradigms. Advances in processor manufacturing, emerging ISAs and machine learning techniques will significantly impact future system on-chip (SoC) designs, enabling true high-performance space computing. To better understand the computational requirements of modern spacecraft, a comprehensive set of benchmarks that include basic system characterization, high performance computing, navigation and landing, image recognition, route finding, data mining and machine learning are necessary to characterize candidate architectures. The analysis of these space flight focused algorithms will drive the design of next generation space computing SoCs.

关键词： AI/ML Benchmarking high-performance-computing HPC Space-computing

来源：评论

学校读者我要写书评

暂无评论

A high performance Computational Framework for Dynamic Security Assessment under Uncertainty 3

A High Performance Computational Framework for Dynamic Secur...

引用

3rd IEEE International Workshop on Electronic Power Grid (eGrid)

作者： Chen, Yousu Palmer, Bruce Sharma, Poorva Yuan, Yong Mathew, Bibi Huang, Zhenyu Pacific Northwest Natl Lab Elect Infrasture & Bldg Richland WA 99352 USA Pacific Northwest Natl Lab High Performance Comp Richland WA 99352 USA Pacific Northwest Natl Lab Data Architectures Richland WA 99352 USA Pacific Northwest Natl Lab Hydrol Richland WA 99352 USA

ISBN: (纸本)9781538676653

Dynamic security assessment (DSA) is a critical function to evaluate power grids' capability to survive the transition caused by a set of disturbances to an acceptable steady-state condition. Its computational burden is heavy. With the challenges brought by renewable energy and new smart grid technologies, DSA under uncertainty has to be considered to study the impact of forecast errors on DSA simulation, which further increases the computational burden. To address this challenge, this paper presents a high performance computational framework to support DSA simulation user uncertainty. The computational framework provides a seamless work flow that links data from high performance computing, statistical analysis, to visualization so that a problem can he easily expressed in a way compatible with different functions. It also enables software compatibility such that application development can be more efficient. Case study results of the ESCA60 system and a western U.S. system show the advantages and efficiency of the computational framework.

关键词： dynamic security assessment high-performance-computing uncertainty smart sampling middleware visualization

来源：评论

学校读者我要写书评

暂无评论

Improving the Convergence Rate of Parareal-in-time Power System Simulation using the Krylov Subspace

Improving the Convergence Rate of Parareal-in-time Power Sys...

引用

IEEE-Power-and-Energy-Society General Meeting (PESGM)

作者： Duan, Nan Simunovic, Srdjan Dimitrovski, Aleksandar Sun, Kai Univ Tennessee Elect Engn & Comp Sci Dept Knoxville TN 37996 USA Oak Ridge Natl Lab Computat Sci & Engn Div Oak Ridge TN USA Univ Cent Florida Elect & Comp Engn Dept Orlando FL 32816 USA

ISBN: (纸本)9781538677032

The performance of parareal-in-time algorithms is determined on the number of sequential, coarse step iterations. A common tradeoff in designing an efficient parareal-in-time algorithm is between accuracy of the coarse solver and the number of iterations. Traditional parareal implementation for the power system simulation can also have difficulties handling complex power systems. In this paper, we propose a Krylov subspace enhanced parareal algorithm to reduce the number of coarse iterations. The proposed approach is demonstrated on a single-machine-infinite-bus system and the IEEE 10-machine 39-bus system. Noticeable decrease of number of iterations is observed in both cases.

关键词： power system simulation parareal-in-time numerical method high-performance-computing Krylov subspace

来源：评论

学校读者我要写书评

暂无评论

Benchmark Analysis of AWGR-based Optical Tiled Architectures for Multi-Socket HPC Boards

Benchmark Analysis of AWGR-based Optical Tiled Architectures...

引用

2015 International Conference on Photonics in Switching (PS)

作者： Grani, Paolo Proietti, Roberto Ben Yoo, S. J. Univ Calif Davis Dept Elect & Comp Engn Davis CA 95616 USA

ISBN: (纸本)9781479988211

We analyze the execution time and energy performance of tiled optical multi-socket HPC boards with all-to-all AWGR-based interconnection and with different optimizations techniques under PARSEC benchmarking traffic, and we compare it with a state-of-the-art electronic multi-socket architecture. Benchmark results show significant performance improvements and up to 2x energy saving when using dynamic variable bandwidth communication technique.

关键词： AWGR Chip-Multi-Processor high-performance-computing Multi-Socket Boards Optical Interconnection

来源：评论

学校读者我要写书评

暂无评论

The Productivity, Portability and performance of OpenMP 4.5 for Scientific Applications Targeting Intel CPUs, IBM CPUs, and NVIDIA GPUs 13th

The Productivity, Portability and Performance of OpenMP 4.5 ...

引用

13th International Workshop on OpenMP (IWOMP)

作者： Martineau, Matt McIntosh-Smith, Simon Univ Bristol HPC Grp Merchant Venturers BldgWoodland Rd Bristol BS8 1UB Avon England

ISBN: (纸本)9783319655789;9783319655772

This research considers the productivity, portability, and performance offered by the OpenMP parallel programming model, from the perspective of scientific applications. We discuss important considerations for scientific application developers tackling large software projects with OpenMP, including straightforward code mechanisms to improve productivity and portability. performance results are presented across multiple modern HPC devices, including Intel Xeon, and Xeon Phi CPUs, POWER8 CPUs, and NVIDIA GPUs. The results are collected for three exemplar applications: hydrodynamics, heat conduction and neutral particle transport, using modern compilers with OpenMP support. The results show that while current OpenMP implementations are able to achieve good performance on the breadth of modern hardware for memory bandwidth bound applications, our memory latency bound application performs less consistently.

关键词： OpenMP-4 high-performance-computing Mini-apps

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：