检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Yang, Yuting Huang, Pei Ma, Fei Fei Cao, Juan Zhang, Meishan Zhang, Jian Li, Jintao Key Lab Of Intelligent Information Processing Institute Of Computing Technology Chinese Academy Of Sciences Beijing100190 China University Of Chinese Academy Of Sciences Beijing100049 China Beijing100190 China Laboratory Of Parallel Software And Computational Science ISCAS Beijing100190 China China

Deep-learning-based NLP models are found to be vulnerable to word substitution perturbations. Before they are widely adopted, the fundamental issues of robustness need to be addressed. Along this line, we propose a formal framework to evaluate word-level robustness. First, to study safe regions for a model, we introduce robustness radius which is the boundary where the model can resist any perturbation. As calculating the maximum robustness radius is computationally hard, we estimate its upper and lower bound. We repurpose attack methods as ways of seeking upper bound and design a pseudo-dynamic programming algorithm for a tighter upper bound. Then verification method is utilized for a lower bound. Further, for evaluating the robustness of regions outside a safe radius, we reexamine robustness from another view: quantification. A robustness metric with a rigorous statistical guarantee is introduced to measure the quantification of adversarial examples, which indicates the model's susceptibility to perturbations outside the safe radius. The metric helps us figure out why state-of-the-art models like BERT can be easily fooled by a few word substitutions, but generalize well in the presence of real-world noises. Copyright © 2022, The Authors. All rights reserved.

关键词： Dynamic programming

来源：评论

学校读者我要写书评

暂无评论

Reduction algorithm optimization based on the OpenCL

引用

Ruan Jian Xue Bao/Journal of software 2011年第SUPPL. 2期22卷 163-171页

作者： Yan, Shen-Gen Zhang, Yun-Quan Long, Guo-Ping Li, Yan Laboratory of Parallel Software and Computational Science Institute of Software Chinese Academy of Sciences Beijing 100190 China State Key Laboratory of Computing Science Institute of Software Chinese Academy of Sciences Beijing 100190 China Graduate University Chinese Academy of Sciences Beijing 100190 China

Reduction algorithm has a wide range of applications in areas such as scientific computing and image processing. This paper systematically studies the reduction algorithm optimization on the GPU's cross-platform performance optimization based on the OpenCL framework. Previous research has generally focused on a single hardware architecture, however, this paper based on the OpenCL, studies various kinds of optimization methods, such as using vector, on-chip memory bank conflict, threads organization, instruction selection and so on. The research takes the minMax function for example, dilatationed each optimization method for develep the performance, and detailed the reason. The study tests the algorithm both on AMD GPU and NVIDIA GPU platforms. The test results show that the optimized algorithm on both platforms has achieved good performance. In the AMD ATI Radeon HD 5850 platform, Int and Float types of data bandwidth utilization up to 89%. In the NVIDIA GPU Tesla C2050 platform, the performance has reached 1.3 to 1.9 times compare to appropriate function version of CUDA. ©2011 Journal of software.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

A GPU computational model based on latency hidden factor

引用

Ruan Jian Xue Bao/Journal of software 2010年第SUPPL. 1期21卷 251-262页

作者： Yuan, Liang Zhang, Yun-Quan Long, Guo-Ping Wang, Ke Zhang, Xian-Yi Laboratory of Parallel Software and Computational Science Institute of Software The Chinese Academy of Sciences Beijing 100190 China State Key Laboratory of Computer Science Institute of Software The Chinese Academy of Sciences Beijing 100190 China Graduate University The Chinese Academy of Sciences Beijing 100049 China

The general purpose GPU computing technology has been successfully used to accelerator many important applications. Though researches have designed many programming models and performance models, the amount of effort required to optimize the performance of applications on GPUs is still very high. In this paper, we propose a GPU computational model based on the ability that how much an algorithm could hide the latency. The experimental results show that our model could predict three matrix multiplication algorithms well. © by Institute of software, the Chinese Academy of sciences.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

Generalised hermite functions and their applications in spectral approximations

arXiv

引用

arXiv 2020年

作者： Sheng, Changtao Ma, Suna Li, Huiyuan Wang, Li-Lian Jia, Lueling Division of Mathematical Sciences School of Physical and Mathematical Sciences Nanyang Technological University Singapore637371 Singapore School of Mathematical Sciences Peking University Beijing100871 China State Key Laboratory of Computer Science/Laboratory of Parallel Computing Institute of Software Chinese Academy of Sciences Beijing100190 China Beijing Computational Science Research Center Beijing100193 China

In 1939, G. Szegö first introduced a family of generalised Hermite polynomials (GHPs) as a generalisation of usual Hermite polynomials, which are orthogonal with respect to the weight function |x|2µe´x2, µ ą ´ 12 on the whole line. Since then, there have been a few works on the study of their properties, but no any on their applications to numerical solutions of partial differential equations (PDEs). The main purposes of this paper are twofold. The first is to construct the generalised Hermite polynomials and generalised Hermite functions (GHFs) in arbitrary d dimensions, which are orthogonal with respect to |x|2µe´|x|2 and |x|2µ in Rd, respectively. We then define a family of adjoint generalised Hermite functions (A-GHFs) upon GHFs, which has two appealing properties: (i) the Fourier transform maps A-GHF to the corresponding GHF;and (ii) A-GHFs are orthogonal with respect to the inner product ru, vsHspRdq "pṕ∆q 2s u, ṕ∆q 2s vqRd associated with the integral fractional Laplacian. The second purpose is to explore their applications in spectral approximations of PDEs. As a remarkable consequence of the fractional Sobolev-type orthogonality, the spectral-Galerkin method using A-GHFs as basis functions leads to an identity stiffness matrix for the integral fractional Laplacian operator ṕ∆qs, which is known to be notoriously difficult and expensive to discretise. Indeed, the A-GHFs provide an optimal basis for Hermite approximation of ṕ∆qs in Rd. As a by-product, the A-GHFs with integer parameter s can significantly improve the existing Hermite spectral-Galerkin algorithms for PDEs with usual Laplacian. We also conduct rigorous error analysis of approximation by GHFs and of the optimal spectral method for fractional PDEs in Rd. The second application is to solve the eigenvalue problem of the Schrödinger operator involving a general fractional power potential. Following the same spirit, we further introduce a Müntz-type GHFs that are orthogonal with respect to an inner pr

关键词： Stiffness matrix

来源：评论

学校读者我要写书评

暂无评论

Performance testing and analysis of BLAS libraries on multi-core CPUs

引用

Ruan Jian Xue Bao/Journal of software 2010年第SUPPL. 1期21卷 214-223页

作者： Chen, Shao-Hu Zhang, Yun-Quan Zhang, Xian-Yi Cheng, Hao Laboratory of Parallel Software and Computational Science Institute of Software The Chinese Academy of Sciences Beijing 100190 China State Key Laboratory of Computing Science The Chinese Academy of Sciences Beijing 100190 China Graduate University The Chinese Academy of Sciences Beijing 100190 China

BLAS library is the most basic math library in high performance computing. Its performance has a great impact on the performance of supercomputers. With the multi-core technology development, BLAS' multi-core parallel performance has become more important than single-core performance associated with architecture. The experiment takes X86 multi-core processors like Xeon, Opteron series as platform for example, which are popular in HPC. It fully tests GotoBLAS, Atlas, MKL and ACML BLAS libraries of all 1, 2, 3-level functions, and covers different scales and multi-core parallel aspect. BLAS source code, material and papers, test results are used to analyze the way of BASL optimization and parallelism, and which platform they are suitable for. Then we will provide some useful suggestions for the use of BLAS, BLAS optimization method or even the development of high-performance CPUs. It was found that compared with a logically powerful and complex CPU, a CPU which has larger and better caches, wider memory bandwidth, smaller memory latency, higher core frequency can often obtain better performance in HPC applications. At the same time, the condition of X86 platform is also a good example for other architectures. © by Institute of software, the Chinese Academy of sciences.

关键词： Program processors

来源：评论

学校读者我要写书评

暂无评论

OpenBLAS: A high performance BLAS library on Loongson 3A CPU

引用

Ruan Jian Xue Bao/Journal of software 2011年第SUPPL. 2期22卷 208-216页

作者： Zhang, Xian-Yi Wang, Qian Zhang, Yun-Quan Laboratory of Parallel Software and Computational Science Institute of Software Chinese Academy of Sciences Beijing 100190 China State Key Laboratory of Computing Science Chinese Academy of Sciences Beijing 100190 China Graduate University Chinese Academy of Sciences Beijing 100190 China

BLAS is a fundamental math library in scientific computing. Thus, each CPU vendor releases optimized BLAS library for its own CPU. Loongson CPU series are developed by the Institute of Computing Technology, Chinese Academy of sciences. In 2010, it released Loongson 3 CPU series. This paper introduces the open source BLAS library OpenBLAS, which is forked on GotoBLAS 2-1.13 BSD version. BLAS Level 3 functions of OpenBLAS is optimized on Loongson 3A quad cores CPU. In sequential optimizations, blocking, hand coding assembly kernel, Loongson 3A special instructions and reordering instructions are utilized. The performance of BLAS Level 3 subroutines exceeded GotoBLAS and ATLAS by about 75% and 17%. Meanwhile, it exceeded GotoBLAS and ATLAS by about 103% and 36% in double precision functions. In parallel multi-threads optimization, this study used interleaved data buffer layout to avoid shared L2 Cache conflictions among multi-threads. OpenBLAS achieved 3.47 speedups on quad cores. In 4 threads, the performance of OpenBLAS BLAS Level3 functions exceeded GotoBLAS and ATLAS by about 69% and 34%, 89% and 55% in double precision functions. ©2011 Journal of software.

关键词： Open source software

来源：评论

学校读者我要写书评

暂无评论

Intensity model with blur effect on GPUs applied to large-scale star simulators

引用

Ruan Jian Xue Bao/Journal of software 2011年第SUPPL. 2期22卷 172-181页

作者： Li, Chao Zhang, Yun-Quan Zheng, Chang-Wen Hu, Xiao-Hui National Key Laboratory of Integrated Information System Technology Institute of Software Chinese Academy of Sciences Beijing 100190 China Laboratory of Parallel Software and Computational Science Institute of Software Chinese Academy of Sciences Beijing 100190 China Graduate University Chinese Academy of Sciences Beijing 100049 China

Intensity model with blur effect is widely employed to accurately simulate the imaging process of star simulator used for attitude determination and guiding feedback. It imposes great demands of computing power for realistic domains, and modern Graphics Processing Units (GPUs) have demonstrated to be a powerful accelerator for this kind of computationally intensive simulations. This paper presents a parallel design and implementation of the intensity model applied to large-scale star simulators on GPUs using the compute unified device architecture (CUDA) programming model. The study analyzes the double parallel nature inherent in this model and use the CUDA framework to efficiently exploit the potential fine-grain data parallelism. Two versions of simulator are designed and studied: One is sequential simulator used as the baseline simulator, and another is parallel simulator using CUDA. In parallel strategy, model, and GPU implementation level, the study employs specific optimized strategies to efficiently improve the parallel performance. Finally, two benchmarks corresponding with the double parallelism are developed to fully evaluate the performance behavior of our simulators. The result analysis demonstrates the efficiency of the CUDA simulators and also illustrates the restriction and bottlenecks presented in this simulator. ©2011 Journal of software.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

A Complexity-Reduced Block-Selective Algebraic Multigrid Method for Implicitly Coupled Velocity-Pressure System on High Performance Computers

SSRN

引用

SSRN 2023年

作者： Liang, Yuechao Guo, Xiao-Wei Li, Chao Yuan, Fan Song, Min Zhang, Qingyang Chen, Xinhai Liu, Jie Science and Technology on Parallel and Distributed Processing Laboratory National University of Defense Technology Changsha410073 China Laboratory of Digitizing Software for Frontier Equipment National University of Defense Technology Changsha410073 China Institute for Quantum Information State Key Laboratory of High Performance Computing College of Computer Science and Technology National University of Defense Technology Changsha410073 China School of Mathematics and Computational Sciences Xiangtan University Xiangtan411105 China College of Computer Science and Technology National University of Defense Technology Changsha410073 China

The implicitly coupled pressure-based algorithm is widely acknowledged for its superior convergence and robustness in solving incompressible flow problems. However, the increased expansion scale of equations and difficulty in solving the resulting equation system has limited the widespread use of the coupling algorithm. This paper presents an optimized block-SAMG method, referred to as Red-SAMG, to significantly reduce computational complexity and memory consumption. Our approach employs a parallel modified independent set algorithm, allowing each process to perform matrix coarsening individually. An aggressive coarsening strategy further reduces complexity to address larger-scale problems. We conducted numerical experiments on a serial two-dimensional backward-facing step case and a parallel three-dimensional tunnel between rod bundles in the reactor core problem. The results demonstrate that the Red-SAMG method with aggressive coarsening reduces solution time by 12\% on the backward-facing step case. For large-scale three-dimensional problems under the same parallel conditions, the Red-SAMG method shortens the solution time by 14\% compared to the existing SAMG method. The Red-SAMG method with aggressive coarsening further reduces solution time by 49\%. Our optimized Red-SAMG method effectively reduces the complexity of the linear system arising from the implicitly coupled algorithm. The aggressive coarsening strategy also represents a significant advancement in improving the computational efficiency of the SAMG method. © 2023, The Authors. All rights reserved.

关键词： Coarsening

来源：评论

学校读者我要写书评

暂无评论

Ball prolate spheroidal wave functions in arbitrary dimensions

arXiv

引用

arXiv 2018年

作者： Zhang, Jing Li, Huiyuan Wang, Li-Lian Zhang, Zhimin School of Mathematics and Statistics Hubei Key Laboratory of Mathematical Sciences Central China Normal University Wuhan430079 China State Key Laboratory of Computer Science Laboratory of Parallel Computing Institute of Software Chinese Academy of Sciences Beijing100190 China Division of Mathematical Sciences School of Physical and Mathematical Sciences Nanyang Technological University Singapore637371 Beijing Computational Sciences and Research Center Beijing100193 China Department of Mathematics Wayne State University MI48202 United States

In this paper, we introduce the prolate spheroidal wave functions (PSWFs) of real order α > −1 on the unit ball in arbitrary dimension, termed as ball PSWFs. They are eigenfunctions of both a weighted concentration integral operator, and a Sturm-Liouville differential operator. Different from existing works on multi-dimensional PSWFs, the ball PSWFs are defined as a generalisation of orthogonal ball polynomials in primitive variables with a tuning parameter c > 0, through a "perturbation" of the Sturm-Liouville equation of the ball polynomials. From this perspective, we can explore some interesting intrinsic connections between the ball PSWFs and the finite Fourier and Hankel transforms. We provide an efficient and accurate algorithm for computing the ball PSWFs and the associated eigenvalues, and present various numerical results to illustrate the efficiency of the method. Under this uniform framework, we can recover the existing PSWFs by suitable variable *** Codes 42B37, 33E30, 33C47, 42C05, 65D20, 41A10 Copyright © 2018, The Authors. All rights reserved.

关键词： Eigenvalues and eigenfunctions

来源：评论

学校读者我要写书评

暂无评论

基于MPI和CUDA的蛋白质定量软件的设计和分析

基于MPI和CUDA的蛋白质定量软件的设计和分析

引用

2012全国高性能计算学术年会

作者： Zhank Ke 詹科 Wang Jing 王靖 Yuan Liang 袁良 Zhang Yunquan 张云泉 Laboratory of Parallel software and computational science ISCASBeijing 100190 中国科学院软件研究所并行软件与计算科学实验室北京 100190

本文介绍了蛋白质定量软件的设计及测试结果.分析结果表明：P-QuantWiz软件在曙光6000的Intel集群部分,测试规模达到2048核,在256核加速比仍在增加.相对P-QuantWiz软件在单CPU运行的结果,PG-QuantWiz软件在单GPU运行的加速比为***-Quant... 详细信息

本文介绍了蛋白质定量软件的设计及测试结果.分析结果表明：P-QuantWiz软件在曙光6000的Intel集群部分,测试规模达到2048核,在256核加速比仍在增加.相对P-QuantWiz软件在单CPU运行的结果,PG-QuantWiz软件在单GPU运行的加速比为***-QuantWiz软件在16个GPU运行结果相对在单个GPU运行结果,加速比为14.18,效率为89％.

关键词：蛋白质定量软件设计并行算法

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：