Parallel Implementation of the Non-Overlapping Template Matching Test Using CUDA

NIST(National Institute of Standards and Technology) statistical test recognized as the most authoritative is widely used in verifying the randomness of binary sequences. The Non-overlapping Template Matching Test as the 7 th test of the NIST Test Suit is remarkably time consuming and the slow performance is one of the major hurdles in the testing process. In this paper, we present an efficient bit-parallel matching algorithm and segmented scan-based strategy for execution on Graphics Processing Unit(GPU) using NVIDIA Compute Unified Device Architecture(CUDA). Experimental results show the significant performance improvement of the parallelized Non-overlapping Template Matching Test, the running speed is 483 times faster than the original NIST implementation without attenuating the test result accuracy.

Data Mining Algorithm Implementation and Its Application in Parallel Cloud System based on C＋＋

. This paper conducts the analysis on the data mining algorithm implementation and its application in parallel cloud system based on C＋＋. With the increase in the number of the cloud computing platform developers, with the use of cloud computing platform to support the growth of the number of Internet users, the system is also the proportion of log data growth. At present applies in the colony environment many is the news transmission model. In takes in the rest transmission model, between each concurrent execution part exchanges the information, and the coordinated step and the control execution through the transmission news. As for the C＋＋ in the data mining applications, it should ？ rstly hold the following features. Parallel communication and serial communication are two basic ways of general communication. Under this basis, this paper proposes the novel perspective on the data mining algorithm implementation and its application in parallel cloud system based on C＋＋. The later research will be focused on the code based implementation.

Improved Software Implementation for Montgomery Elliptic Curve Cryptosystem

The last decade witnessed rapid increase in multimedia and other applications that require transmitting and protecting huge amount of data streams simultaneously.For such applications,a high-performance cryptosystem is compulsory to provide necessary security services.Elliptic curve cryptosystem(ECC)has been introduced as a considerable option.However,the usual sequential implementation of ECC and the standard elliptic curve(EC)form cannot achieve required performance level.Moreover,the widely used Hardware implementation of ECC is costly option and may be not affordable.This research aims to develop a high-performance parallel software implementation for ECC.To achieve this,many experiments were performed to examine several factors affecting ECC performance including the projective coordinates,the scalar multiplication algorithm,the elliptic curve(EC)form,and the parallel implementation.The ECC performance was analyzed using the different factors to tune-up them and select the best choices to increase the speed of the cryptosystem.Experimental results illustrated that parallel Montgomery ECC implementation using homogenous projection achieves the highest performance level,since it scored the shortest time delay for ECC computations.In addition,results showed thatNAF algorithm consumes less time to perform encryption and scalar multiplication operations in comparison withMontgomery ladder and binarymethods.Java multi-threading technique was adopted to implement ECC computations in parallel.The proposed multithreaded Montgomery ECC implementation significantly improves the performance level compared to previously presented parallel and sequential implementations.

An Improvement Model Presented to Reduce the Required Time for the Implementation of ERP in the Organization

One of the many issues in utilizing ERP systems in organizations are, in fact, the implementation stage. This study by investigating the common and available methods of implementation, as well as their inefficiencies will provide a new more efficient method. This new method, in fact, will first assess the required time of implementation in each of the units of the organization, and then will take advantage of a spherical model with central core instead of a linear model. These units, with regard to the required time of implementation, will surround this core as in the form of some layers. The circuits are ordered in a way that the further we move from the core towards the external layers, the shorter the required time of implementation will become. This way, the priority of implementing ERP will be assigned with a direction from external layers to internal layers. Eventually, all the experiences of the previous stages will be transferred to the central core, which has the most complexity. Through this method, it is expected that we may prevent the fully parallel issue, which was a dominant and apparent issue in previous models, so that the required time of implementation would decrease.

Parallel Spectral Clustering Based on MapReduce

Clustering is one of the most widely used techniques for exploratory data analysis. Spectral clustering algorithm, a popular modern cluslering algorithm, has been shown to be more effective in detecting clusters than many traditional algorithms. It has applications ranging from computer vision and information retrieval to social sienee and biology. With the size of databases soaring, cluostering algorithms bare saling computational time and memory use. In this paper, we propose a parallel spectral elustering implementation based on MapRednee. Both the computation and data storage are dislributed, which solves the sealability problems for most existing algorithms. We empirically analyze the proposed implementation on both benchmark net- works and a real social network dataset of about two million vertices and two billion edges crawled from Sina Weibo. It is shown that the proposed implementation scales well, speeds up the clustering without sacrificing quality, and processes massive datasets efficiently on commodity machine clusters.

交通速度预测时空图卷积网络及其FPGA实现研究

时空图卷积网络(STGCN)通过图卷积和时间卷积捕获交通数据的空间依赖性和时间依赖性,可有效提升交通速度预测的精度。但是硬件实现交通速度预测STGCN具有计算量大难以满足实际应用的实时性要求、资源消耗大导致成本增高等问题,在优化交通速度预测STGCN模型基础上,提出了一种交通速度预测STGCN的FPGA实现结构组合优化的方法。首先,通过轻量化裁剪和预测数据位宽的精确选择,对交通速度预测STGCN进行了模型优化,以降低计算复杂度和资源消耗,并经过Python仿真验证其可行性。其次,通过采用流水线、并行计算和数据交替流水存取等组合优化策略,提出了一种交通速度预测STGCN的FPGA实现结构组合优化的方法,以提升系统计算速度。最后,使用Verilog编程对交通速度预测STGCN进行了FPGA的实现仿真和硬件测试。利用PeMSD7(M)数据集进行实验,结果显示FPGA实现单数据交通速度预测的时间为355.5μs,相比CPU、GPU平台及FPGA设计方案1对比,其处理速度最大分别提高了25.9倍、6.7倍和3.5倍,证明了交通速度预测STGCN的FPGA实现结构组合优化方法,在保持预测准确性的前提下可较大幅度的提升系统处理速度。

分组密码uBlock算法快速软件实现

为对国产分组密码算法uBlock进行软件优化,选择支持256 bit数据位宽的AVX2指令集,提高编译器自动优化等级,优化函数的调用过程,优化数据存储结构,综合使用高位并行、低延迟指令逻辑优化等方法实现单线程并行计算。通过使用这种有效的组合方法,uBlock-128/128算法、uBlock-128/256算法和uBlock-256/256算法单密钥短消息加密的速度较原代码分别提升269%、182%和49%。基于这些优化方法,uBlock-128/128、uBlock-128/256和uBlock-256/256三个算法版本均实现了单密钥场景与多密钥场景。

Parallel Implementation of Linear Algebra Problems on Dawning-1000

In this paper, some parallel algorithms are described for solving numerical linear algebra problems on Dawning-1000. They include matrix multiplication, LU factorization of a dense matrix, Cholesky factorization of a symmetric matrix, and eigendecomposition of symmetric matrix for real and complex data types. These programs are constructed based on fast BLAS library of Dawning-1000 under NX environment.Some comparison results under different parallel environments and implementing methods are also given for Cholesky factorization. The execution time, measured performance and speedup for each problem on Dawning-1000 are shown. For matrix multiplication and LU factorization, 1.86GFLOPS and 1.53GFLOPS are reached.

Linear scaling Coulomb interaction in the multiwavelet basis,a parallel implementation

We present a parallel and linear scaling implementation of the calculation of the electrostatic potential arising from an arbitrary charge distribution.Our approach is making use of the multi-resolution basis of multiwavelets.The potential is obtained as the direct solution of the Poisson equation in its Green’s function integral form.In the multiwavelet basis,the formally non local integral operator decays rapidly to negligible values away from the main diagonal,yielding an effectively banded structure where the bandwidth is only dictated by the requested accuracy.This sparse operator structure has been exploited to achieve linear scaling and parallel algorithms.Parallelization has been achieved both through the shared memory(OpenMP)and the message passing interface(MPI)paradigm.Our implementation has been tested by computing the electrostatic potential of the electronic density of long-chain alkanes and diamond fragments showing(sub)linear scaling with the system size and efficent parallelization.

发动机装配线并线改造项目实施方法

重点介绍了发动机装配线并线改造项目实施步骤及方法,主要从硬件改造方案、防错方案、工艺文件编制方法、工艺相关系统策划与调整、人员规划、物流规划、产品备库、线体调试和质量验证、产能爬坡9个方面进行阐述,并以S公司UCW系列发动机装配线并线改造项目为例,介绍了其实施的主要过程、遇到的问题及为企业带来的经济效益。

并行算法研究方法学

并行算法是计算机科学中重要的研究内容,已有几十年的发展历程.回顾一下其研究历程,既有高潮也有低谷,究其原因是,它没有形成自身的一套研究方法学.为此文中提出并行算法研究要建立起一套完整的"理论-设计-实现-应用"的学科体系,也就是所谓的并行算法研究的生态环境.只有这样才能够保持并行算法研究稳定、可持续发展,并使得并行算法的研究成果更加实用,从而更富有生命力.

一种快速CRC算法的硬件实现方法

介绍了CRC校验算法的硬件电路实现方法。CRC校验广泛应用于通信、存储系统 ,在串行CRC实现的基础上 ,对电路结构提出了改进的方案 ,并实现了CRC的并行计算 ,由此进一步可以适用于任意位数据宽度的数据输入情况。

并行遗传算法分析

在科学计算领域，并行计算越来越成熟，并行遗传算法开始受到关注。分析了遗传算法并行化的动机和实现模型，讨论了遗传算法隐含的并行性，对于灵活应用并行遗传算法有指导意义。

并行测试技术综述

分析了并行测试概念,详细地论述了并行测试的实现方式及各种实现方式的优缺点,提出了并行测试系统的几种基本架构,为并行测试技术应用到我军航空武器装备的维护保障中作了有益的基础性工作。

外辐射源雷达多通道时域杂波抑制算法并行实现

直达波及杂波抑制是外辐射源雷达目标信息提取的关键步骤之一。扩展杂波相消批处理(ECA-B)算法是一种有效的时域杂波抑制算法,但算法空间和时间复杂度极大,在处理多通道(或多波束)数据时尤是如此。结合图形处理器(GPU)数据吞吐量大、浮点运算能力强的优点,该文提出一种适用于GPU处理的多通道ECA-B时域杂波抑制并行算法。首先推导了多通道ECA-B算法的原理,避免了原算法分别对单通道进行单独处理的计算冗余问题。然后针对其中耗时最大的自相关矩阵计算,提出一种迭代计算方法,使时间和空间复杂度均降至常规方法的约1/K(K为杂波自由度)。最后给出了改进算法的GPU并行实现方案。仿真和实测结果验证了算法的准确性和实效性。

数值气象预报中的并行计算研究

介绍了并行计算与数值气象预报的发展概况 ,论述了并行算法的一般设计方法 ,展示了作者在数值气象预报的典型并行算法和并行实现技术方面所做的主要研究工作 ,以及这些研究成果所取得的一系列数值气象预报系统的并行计算结果 ,最后讨论了并行计算与数值气象预报的未来发展。

JPEG2000遥感图像实时压缩系统中基于并行机制的“零时间”数据搬移策略

为了满足JPEG2000遥感图像实时压缩设备的实时性要求,提出了一种基于并行机制的"零时间"数据搬移策略。通过对EBC算法结构进行拆分,并采用EDMA方式进行数据搬移,使得算法运算与数据搬移过程完全并行,搬移时间等效为零。实验结果表明,采用新的数据搬移策略后,编码效率可以提升到100%,EBC算法实现的总时间可缩短46%以上。该搬移机制已成功应用于JPEG2000星载遥感图像实时压缩系统中,系统的实时性达到了设计要求。

线化欧拉方程的高阶间断有限元数值解法研究

采用高阶间断有限元法于非结构网格上针对复杂外形数值求解声学控制方程——线化欧拉方程.背景流场采用有限体积法于结构网格求得,一种高精度数据传递方法将基于有限体积法的背景流场数据传递到声场计算所采用的较为稀疏的非结构网格上,保证了背景流场信息的完整和精确.为提高计算效率,采用了一种更为直接的Quadrature-Free Implementation技术以及网格分区并行技术.数值结果表明采用高阶的情况下即使在稀疏的网格上也可以捕捉到细微的声场结构.

可扩展并行Web服务器集群的实现技术

随着INTERNET用户和流量的不断增长,对Web站点的性能提出了更高的要求,以缩短用户请求的响应时间。该文介绍了高性能的可扩展并行Web服务器集群的工作原理和实现机制,并说明了Web服务器集群的应用前景和发展趋势。

一种求解最大团问题的并行交叉熵算法

为了提高交叉熵算法求解最大团问题(maximum clique problem,MCP)的性能,提出一种领导者-跟随者协作求解的并行策略来实现交叉熵算法,从而达到减少计算时间和保障解的质量这两方面的平衡.算法中领导者活跃在并行处理器之间采集数据,并根据当前获得信息对跟随者作出决策;受控的跟随者则主要根据领导者的决策信息自适应地调整搜索空间,完成各自的集团产生任务.采用了OpenMPI在MIMD平台上实现了该算法,并应用到MCP基准测试问题上.加速比和效率分析结果表明,算法具有很好的加速比和效率.而与其它几种当前最好的启发式算法相比,结果表明算法相对于基于种群的启发式算法有一定的性能改善.