检索结果-内蒙古大学图书馆

18th International Conference on Computational Science and Its Applications (ICCSA)

作者： Shichkina, Yulia Awadh, Al-Mardi Mohammed Haidar Storublevtcev, Nikita Degtyarev, Alexander St Petersburg Electrotech Univ LETI Ul Prof Popova 5 St Petersburg 197376 Russia St Petersburg State Univ Univ Skaia Nab 7-9 St Petersburg 199034 Russia

ISBN: (纸本)9783319951713;9783319951706

The method of constructing a schedule for parallel algorithm execution is considered in the article. This algorithm takes into account the execution time of each operation of the algorithm and the relationship of operations on the data. The method is based on an information graph in which the nodes are the operations of the algorithm, and the edges are the directions of the data transfer. As a result of the interchange of operations between computing nodes, it is possible to achieve a reduction in the execution time of the algorithm by reducing the time spent on data transfer between computing nodes and reducing the downtime of computational nodes. The algorithm can be applied both in parallel programming and in adjacent areas, for example, when scheduling tasks in distributed systems.

关键词： parallel algorithm algorithm optimization Information graph Operation execution time algorithm execution schedule Process Processor Interprocessor data transmission

来源：评论

学校读者我要写书评

暂无评论

Design and Implementation of Word2Vec parallel algorithm Based on HPC

Design and Implementation of Word2Vec Parallel Algorithm Bas...

引用

Chinese Automation Congress (CAC)

作者： Yi, Xianyong Zheng, Rongge Wang, Aoyu Qin, Hao Chen, Yufeng Shandong Univ Sch Mech Elect & Informat Engn Weihai Weihai Peoples R China

ISBN: (纸本)9781538635247

Word2Vec, (Word to Vector) processes natural language by calculating the cosine similarity. However, the serial algorithm of original Word2Vec fails to satisfy the demands of training of corpus text because of the explosive growth of information. It has become the bottleneck owing to its comparatively low processing efficiency. The High Performance Computing (HPC) specializes in improving the calculation efficiency;therefore, the training efficiency of corpus texts can be greatly improved by parallelizing Word2Vec algorithm. After analyzing the characteristics of the Word2Vec algorithm in detail, we design and implement a parallel Word2Vec algorithm and use it to train corpus text on HPC. Furthermore, the corpus texts of different sizes are collected and trained, and the speed-up ratio is calculated by using the serial algorithm and parallel algorithm of Word2Vec, respectively. The experimental results show that there is a higher speed-up ratio when using the Word2Vec parallel algorithm running on HPC.

关键词： HPC Word2Vec parallel algorithm Natural Language Processing

来源：评论

学校读者我要写书评

暂无评论

Scalable parallel algorithm for Solving Non-stationary Systems of Linear Inequalities

引用

LOBACHEVSKII JOURNAL OF MATHEMATICS 2020年第8期41卷 1571-1580页

作者： Sokolinsky, L. B. Sokolinskaya, I. M. Natl Res Univ South Ural State Univ Chelyabinsk 454080 Russia

In this paper, a scalable iterative projection-type algorithm for solving non-stationary systems of linear inequalities is considered. A non-stationary system is understood as a large-scale system of inequalities in which coefficients and constant terms can change during the calculation process. The proposed parallel algorithm uses the concept of pseudo-projection which generalizes the notion of orthogonal projection. The parallel pseudo-projection algorithm is implemented using the parallel BSF-skeleton. An analytical estimation of the algorithm scalability boundary is obtained on the base of the BSF cost metric. The large-scale computational experiments were performed on a cluster computing system. The obtained results confirm the efficiency of the proposed approach.

关键词： non-stationary system of linear inequalities feasibility problem iterative method parallel algorithm approximate solution cluster computing system

来源：评论

学校读者我要写书评

暂无评论

A parallel algorithm for Finding All Elementary Circuits of a Directed Graph 37

A Parallel Algorithm for Finding All Elementary Circuits of ...

引用

37th Chinese Control Conference (CCC)

作者： Lu, Wenbo Zhao, Qianchuan Zhou, Cangqi Tsinghua Univ Dept Automat Ctr Intelligent & Networked Syst Beijing 100084 Peoples R China Nanjing Univ Sci & Technol Sch Comp Sci & Engn Nanjing 210094 Jiangsu Peoples R China

ISBN: (纸本)9789881563958

Graph can be used as a significant model to simulate the real world. We can find this kind of models in many different fields, such as transportation, navigation, logistics, management and social networks. In such graphs, a circuit is often regarded as a closed-loop way by which information propagates. There are several algorithms for finding all elementary circuits. Among these algorithms, Jonson's algorithm suffers the lowest time complexity so far. In this paper, we introduce a parallel algorithm based on Johnson's algorithm. We demonstrate that it is capable to handle large-scale graphs with a computing cluster. Experiments on both synthetic and real-world data sets demonstrate the effectiveness of the proposed algorithm.

关键词： Graph Theory Elementary Circuits MapReduce parallel algorithm

来源：评论

学校读者我要写书评

暂无评论

A parallel algorithm to Construct Node-Independent Spanning Trees on the Line Graph of Locally Twisted Cube 12

A Parallel Algorithm to Construct Node-Independent Spanning ...

引用

12th International Symposium on parallel Architectures, algorithms and Programming (PAAP)

作者： Pan, Zhiyong Cheng, Baolei Fan, Jianxi Zhang, Huanwen Soochow Univ Sch Comp Sci & Technol Suzhou Peoples R China

ISBN: (纸本)9781665496391

An interconnection network can be abstracted into a graph, and the basic mathematical research in the graph can provide a good reference for the research in the practical application. The study of node-independent spanning trees (node-ISTs) in a graph has received extensive attention because of their application in reliable communication, fault-tolerant broadcasting and secure message distribution, and has achieved remarkable results on many special networks. But there are few results in the line graph of them. As one of the typical variations of hypercube, locally twisted cube has many excellent properties, whose line graph has all the advantages of locally twisted cube. So it makes sense to do some research on the line graph of locally twisted cube. In this paper, we propose a parallel algorithm to construct 2n -2 node-ISTs rooted at node [u,N(u, 2)], where u is an arbitrary node on locally twisted cube and n >= 1. And the correctness of our algorithm is proved.

关键词： Locally twisted cube Line graph parallel algorithm Node-independent spanning trees

来源：评论

学校读者我要写书评

暂无评论

Efficient parallel algorithm for Optimal DAG Structure Search on parallel Computer with Torus Network 16th

Efficient Parallel Algorithm for Optimal DAG Structure Searc...

引用

16th International Conference on algorithms and Architectures for parallel Processing (ICA3PP)

作者： Honda, Hirokazu Tamada, Yoshinori Suda, Reiji Univ Tokyo Grad Sch Informat Sci & Technol Tokyo 1138656 Japan

ISBN: (纸本)9783319495835;9783319495828

The optimal directed acyclic graph search problem constitutes searching for a DAG with a minimum score, where the score of a DAG is defined on its structure. This problem is known to be NP-hard, and the state-of-the-art algorithm requires exponential time and space. It is thus not feasible to solve large instances using a single processor. Some parallel algorithms have therefore been developed to solve larger instances. A recently proposed parallel algorithm can solve an instance of 33 vertices, and this is the largest solved size reported thus far. In the study presented in this paper, we developed a novel parallel algorithm designed specifically to operate on a parallel computer with a torus network. Our algorithm crucially exploits the torus network structure, thereby obtaining good scalability. Through computational experiments, we confirmed that a run of our proposed method using up to 20,736 cores showed a parallelization efficiency of 0.94 as compared to a 1296-core run. Finally, we successfully computed an optimal DAG structure for an instance of 36 vertices, which is the largest solved size reported in the literature.

关键词： Optimal DAG structure Optimal bayesian network structure parallel algorithm Distributed algorithm Torus network

来源：评论

学校读者我要写书评

暂无评论

A parallel algorithm for modeling of dynamical processes on large stochastic Kronecker graphs

A parallel algorithm for modeling of dynamical processes on ...

引用

16th Annual International Conference on Computational Science (ICCS)

作者： Bochenina, Klavdiya Kesarev, Sergey ITMO Univ St Petersburg Russia

A stochastic Kronecker graph (SKG) is one of the most widely used approaches to model complex networks due to the simplicity of generative procedure and the ability to reproduce the properties of real graphs. In this paper, we present a novel parallel algorithm for modeling dynamical processes on large Poisson SKGs (PSKGs). Proposed algorithm combines the modeling of spreading of a process with the generation of vertices and edges which are involved in it. An experimental study of an algorithm was carried out for different initiator matrices and graphs with the widely varying (from several million to one billion) number of vertices using the supercomputer from Lobachevsky State University of Nizhni Novgorod, Russia. The results confirmed the applicability of an algorithm for mid-size HPC clusters with the efficiency varying on average from 0.2 to 0.7.

关键词： complex networks stochastic Kronecker graphs parallel algorithm

来源：评论

学校读者我要写书评

暂无评论

A parallel algorithm for LZW Decompression, with GPU Implementation 11th

A Parallel Algorithm for LZW Decompression, with GPU Impleme...

引用

11th International Conference on parallel Processing and Applied Mathematics (PPAM)

作者： Funasaka, Shunji Nakano, Koji Ito, Yasuaki Hiroshima Univ Dept Informat Engn Kagamiyama 1-4-1 Higashihiroshima 7398527 Japan

ISBN: (纸本)9783319321493;9783319321486

The main contribution of this paper is to present a parallel algorithm for LZW decompression and to implement it in a CUDA-enabled GPU. Since sequential LZW decompression creates a dictionary table by reading codes in a compressed file one by one, its parallelization is not an easy task. We first present a parallel LZW decompression algorithm on the CREW-PRAM. We then go on to present an efficient implementation of this parallel algorithm on a GPU. The experimental results show that our parallel LZW decompression on GeForce GTX 980 runs up to 69.4 times faster than sequential LZW decompression on a single CPU. We also show a scenario that parallel LZW decompression on a GPU can be used for accelerating big data applications.

关键词： Data compression Big data parallel algorithm GPU CUDA

来源：评论

学校读者我要写书评

暂无评论

PaSTiLa: Scalable parallel algorithm for Unsupervised Labeling of Long Time Series

引用

LOBACHEVSKII JOURNAL OF MATHEMATICS 2024年第3期45卷 1333-1347页

作者： Zymbler, M. L. Goglachev, A. I. South Ural State Univ Chelyabinsk 454080 Russia

Summarization aims at discovering a small set of typical subsequences (patterns) in the given long time series that represent the whole series. Further, one can implement unsupervised labeling of the given time series by assigning each subsequence a tag that corresponds to its most similar pattern. In the previous research, we developed the PSF (parallel Snippet-Finder) algorithm for the time series summarization on GPU, where a snippet is the given-length subsequence, which is similar to many other subsequences w.r.t. the bespoke distance measure MPdist. However, PSF is limited by the demand that the snippet length be predefined by a domain expert. In this article, we introduce the novel parallel algorithm PaSTiLa (parallel Snippet-based Time series Labeling) that discovers snippets and produces the labeling of the given time series on an HPC cluster with GPU nodes. As opposed to its predecessor, PaSTiLa employs the automatic selection of the snippet length from the specified range through our proposed heuristic criterion. In the experiments on labeling quality over time series from the TSSB (Time Series Segmentation Benchmark) dataset, PaSTiLa outperforms state-of-the-art segmentation-based competitors in average F-1 score. In the case of long-length time series (typically more than 8-10 K points), PaSTiLa outruns the rivals. Finally, over the million-length time series, our algorithm demonstrates a close-to-linear speedup.

关键词： time series summarization labeling snippet MPdist Snippet-Finder parallel algorithm GPU CUDA high-performance cluster

来源：评论

学校读者我要写书评

暂无评论

A load balancing parallel algorithm for solving large-scale tridiagonal linear systems

A load balancing parallel algorithm for solving large-scale ...

引用

International Conference on algorithms, High Performance Computing, and Artificial Intelligence (AHPCAI)

作者： Tian, Min Qiao, Shan Wang, Junjie Du, Wei Qilu Univ Technol Shandong Acad Sci Natl Supercomp Ctr Jinan Shandong Comp Sci Ctr Jinan Peoples R China Qilu Univ Technol Shandong Acad Sci Sch Math & Stat Jinan Peoples R China

ISBN: (数字)9781510651890

ISBN: (纸本)9781510651890;9781510651883

Solving large-scale sparse linear systems is a critical problem in scientific and engineering computing. Partial differential equations can solve problems in many fields. They can be transformed into large-scale linear systems with a series of methods, and the parallel solution of tridiagonal linear systems is one of them. The solution of linear systems is very time-consuming in most of the problems, accounting for more than half of the total time. Load balancing can reduce process time for waiting and improves computational efficiency, and it is the focus of many algorithms. The article is based on Stone's proposed recursive doubling algorithm, an improved algorithm for solving tridiagonal linear systems using the full-recursive-doubling communication model and the Mobiu transform. The improved algorithm can calculate the million-dimensional linear systems. Numerical experiments show that compared with ordinary parallel algorithms, the improved algorithm shows up to 2x improvement than the original version, and some results even show up to 3x. In addition, the load-balancing performance has been greatly improved, and the time difference of the processes is 1/7 of the original version. The improved algorithm has a good load balancing, and the running time of each process is not much different, avoiding process waiting and resource wastage.

关键词： tridiagonal linear systems parallel algorithm load balancing recursive doubling Mobius transformation

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：