WWW大规模cache技术

本文分析Ｉｎｔｅｒｎｅｔ网络交通模式，ＷＷＷ ｃａｃｈｅ对端用户、ＩＳＰ和干线供应商的利益驱动，分析了构造大规模ｃａｃｈｅ的技术需求，并提出相应的解决方案：机群计算，缓冲优化的数据库，多线程数据流。     

一种MPP机Cache一致性协议的设计与实现

Ｃａｃｈｅ一致性技术是大规模并行处理计算机系统中的关键技术之一。本文构造了一个典型的ＭＰＰ模型，在此基础上阐述了系统对Ｃａｃｈｅ一致性协议的支持，并给出了具体的实现方案。     

Cache性能与程序优化

超高速缓存是目前用来提高计算机性能的关键技术之一，而算法和程序优化对ｃａｃｈｅ性能的发挥起着重要作用。文中，我们提出一个提高ｃａｃｈｅ性能的优化方法，基本思想是用“加边”法改变ｃａｃｈｅ映射图象而减少冲突以提高ｃａｃｈｅ利用率。     

多机Cache中存储访问不命中率的理论分析

本文主要研究了多处理机系统中访问ｃａｃｈｅ不命中的平均情况。从多机间相互干扰的角度，分析了访问ｃａｃｈｅ的平均不命中次数，本文还从算法设计出发，提出了分析ｃａｃｈｅ伪共享的直观方法——访问模式图     

Cache连续性操作问题研究

Ｃａｃｈｅ调度策略在磁盘操作中占有举足轻重的地位，调度策略设计的好坏直接影响着磁盘Ｉ／Ｏ性能。磁盘上非连续的扇区的存取需要多个Ｉ／Ｏ来完成。当数据读入Ｃａｃｈｅ或Ｃａｃｈｅ腾空时，磁盘旋转等待时间是不可避免的。本文分析了磁盘中常用的Ｃａｃｈｅ调度算法存在的不足，提出了一种新的调度策略，一次Ｉ／Ｏ即可完成非连续扇区的存取操作把旋转延迟缩至最小程度，在某些情况下可达到零延迟。     

PCI协议对Cache的支持

论述了Ｃａｃｈｅ在高性能计算机系统中的作用和访问Ｃａｃｈｅ的过程，以及Ｃａｃｈｅ数据一致性问题和解决的方法，介绍和分析了ＰＣＩ协议对Ｃａｃｈｅ的支持。     

一种 新的Cache优化方法—部分Cache局部性方法

Ｃａｃｈｅ的性能优化在高性能计算中起着非常重要的作用。传统的Ｃａｃｈｅ优化方法存在着一些缺陷。本文分析ＲＩＳＣ处理器的特点的基础 上，提出了“部分ｃａｃｈｅ方法。实践表明，该方法有很好的优化效果，且易实于实现。     

Cache一致性协议的研究与评价

Ｃａｃｈｅ一致性是紧耦合多处理机系统设计中的一项重要课题．为提高访存效率，每台处理机通常带有高速缓冲存储器Ｃａｃｈｅ。这便产生了Ｃａｃｈｅ一致性问题，要求共享数据在各Ｃａｃｈｅ间以及Ｃａｃｈｅ与主存间保持一致。为此出现了多种Ｃａｃｈｅ一致性协议。本文分析了几种类型的一致性协议，并对其进行了软件模拟和性能评价．     

并行文件系统中Disk Cache一致性协议及其性能分析

本文在并行文件系统中引入ｄｉｓｋｃａｃｈｅ多复本技术，从而为并行计算机提供高性能的文件系统．对于ｄｉｓｋｃａｃｈｅ多复本间数据一致性维护，本文提出了“主从式”和“对称式”两类方法，并从其应用的通用性角度，基于等概率模型，对各类方法以及ｄｉｓｋｃａｃｈｅ单复本系统进行了性能分析和比较．     

廉价冗余磁盘阵列（RAID）Cache浅析

廉介冗余磁盘陈列技术已掀起研究开发热潮，磁盘Ｃａｃｈｅ技术的研究早在七十年代就已广泛展开，但是关于磁盘阵列Ｃａｃｈｅ技术的专门性研究文献在国内外并不多见。本文论述了磁盘阵列中引入高速缓存Ｃａｃｈｅ的必要性，综述了磁盘阵列Ｃａｃｈｅ技术的国内外技术动态，提出了磁盘阵列Ｃａｃｈｅ研究中的几个关键问题，并阐述了作者的观点。     

基于信号重构的阵列失效校准方法

针对失效阵元给阵列性能带来的影响,提出一种基于信号重构的阵列失效校准方法。对于理想的均匀分布线列阵,2个相邻阵元相对同一信号源的接收信号之间仅相差一个固定的相移。利用这一点,原本已经失真的失效阵元的输出可以通过合成其他正常阵元的输出信号得到恢复,在一定程度上抑制了阵列失效带来的旁瓣增长。仿真结果表明该方法是可行、有效的。     

A new noniterative method for pattern synthesis of unequally spaced linear arrays

In this article, a new noniterative beam shaping method is introduced to synthesise array factor (AF) of an unequally spaced linear array (UESLA). The proposed method is based on eigenvector decomposition of sampled data matrix of a given pattern. Using matrix analysis, the eigenvalues and their corresponding eigenvectors of the sampled matrix are determined. It is shown that the eigenvalues and eigenvectors of the sampled matrix are related to the locations and complex excitation coefficients of the array elements. According to the concept of generalized eigenvalue concept, the solution of locations and excitation coefficients is derived using least square method. In order to reduce the number of array elements, singular value decomposition is applied to obtain a low ranked matrix using rejection of nonzero eigenvalues. Using the approximated sampled matrix, the excitation and locations of the optimized array elements are calculated. A few comprehensive examples are investigated to verify the accuracy of the proposed method and the obtained results are compared with those of an equally spaced linear array (ESLA). It is shown that the total number of array elements in an UESLA is less than that of ESLA, which is the most advantage of the introduced method in AF synthesis.     

Integration of Trench-Isolated Through-Wafer Interconnects with 2D Capacitive Micromachined Ultrasonic Transducer Arrays

This paper presents a method to provide electrical connection to a 2D capacitive micromachined ultrasonic transducer (CMUT) array. The interconnects are processed after the CMUTs are fabricated on the front side of a silicon wafer. Connections to array elements are made from the back side of the substrate via highly conductive silicon pillars that result from a deep reactive ion etching (DRIE) process. Flip-chip bonding is used to integrate the CMUT array with an integrated circuit (IC) that comprises the front-end circuits for the transducer and provides mechanical support for the trench-isolated array elements. Design, fabrication process and characterization results are presented. The advantages when compared to other through-wafer interconnect techniques are discussed.     

A fast and compact elimination method of empty elements from a double‐array structure

A double‐array is a well‐known data structure to implement the trie. However, the space efficiency of the double‐array degrades with the number of key deletions because the double‐array keeps empty elements produced by the key deletion. This paper presents a fast and compact elimination method of empty elements using properties of the trie nodes that have no siblings. The present elimination method is implemented by C language. From simulation results for large sets of keys, the present elimination method is about 30–330 times faster than the conventional elimination method and maintains high space efficiency. Copyright © 2003 John Wiley & Sons, Ltd.     

混合MIMO相控阵雷达的交错稀疏阵列设计

针对常规的发射子阵分割会使混合MIMO相控阵雷达的孔径减小和馈电网络复杂度增加的问题,提出一种交错稀疏的发射子阵分割方法,通过寻求最大输出信干噪比的方式获得最佳的阵列结构.首先构建交错稀疏结构下的混合MIMO相控阵雷达模型,运用序列凸近似方法将非凸的目标函数转化为凸函数;然后通过凸优化分别求解出一维和二维混合MIMO相控阵雷达在阵元数目固定和阵元数目作为变量的情况下的最佳阵列结构;最后,通过仿真表明所提出方法不仅可以获得较大的信干噪比和较低的旁瓣电平值,且较常规的子阵分割方式能获得更高的波达方向估计精度.     

Matrix pencil method for simultaneously estimating azimuth and elevation angles of arrival along with the frequency of the incoming signals

In this paper we describe a method for simultaneously estimating the direction of arrival (DOA) of the signal along with its unknown frequency. In a typical DOA estimation problem it is often assumed that all the signals are arriving at the antenna array at the same frequency which is assumed to be known. The antenna elements in the array are then placed half wavelength apart at the frequency of operation. However, in practice seldom all the signals arrive at the antenna array at a single pre-specified frequency, but at different frequencies. The question then is what to do when there are signals at multiple frequencies, which are unknown. This paper presents an extension of the matrix pencil method to simultaneously estimate the DOA along with the operating frequency of each of the signals. This novel approach involves approximating the voltages that are induced in a three-dimensional antenna array, by a sum of complex exponentials by jointly estimating the direction of arrival (both azimuth and elevation angles) along with the carrier frequencies of multiple far-field sources impinging on the array by using the three-dimensional matrix pencil method. The matrix pencil method is a direct data domain method for approximating a function by a sum of complex exponentials in the presence of noise. The variances of the estimates computed by the matrix pencil method are quite close to the Cramer–Rao bound. Finally, we illustrate how to carry out the broadband DOA estimation procedure using realistic antenna elements located in a conformal array. Some numerical examples are presented to illustrate the applicability of this methodology in the presence of noise. It is shown that the variance decreases as the SNR increases. The Cramer–Rao bound for the estimators are also provided to illustrate the accuracy and the computational efficiency of this new methodology.     

Recognition of dependent objects based on acyclic Markov models

The problem to recognize objects that form an array of interrelated data is investigated. In the problem of machine learning, the array components belong to some class of a finite set. In this paper the interrelationship of array elements is presented by its adjacency graph. An efficient noniterative recognition algorithm for restoring an a posteriori marginal distributions of hidden classes for array elements is developed for a treelike adjacency graph. This algorithm modifies for each array element the hidden class distribution obtained as a result of learning for independent objects. Usually arbitrary graphs for real data contain cycles, for example, the rectangular adjacency lattice of points for 2D raster images or 3D seismic data. The treelike approximation of such graphs inevitably strongly distorts the interrelations between array elements. In the present paper, the reduced set of interrelations between array elements is balanced by an extended set of acyclic graphs themselves. By the example of the segmentation problem for texture raster images, we investigate a set of acyclic graphs and present the experimental results.     

Response‐correction‐based fault detection in small linear microstrip patch arrays using magnitude‐only far‐field pattern samples

We present a computationally efficient method for detecting faulty elements in a small linear microstrip patch array from samples of the array's far‐field magnitude radiation pattern (here represented by realistic EM simulations). Regardless of the array size, our method requires only one expensive full‐wave entire‐array simulation—compared to, e.g., the 696 required by the previous best method (Patnaik et al., IEEE Trans Antennas Propag 55 (2007), 775–777) for a 16‐element array. This one simulation gives the accurate far‐field magnitude pattern of the original defect‐free array, and is used in conjunction with the defect‐free array's analytical array factor to formulate a response correction function, which can then be used to construct an accurate approximation of the EM‐simulated pattern of any arbitrary faulty array at very low cost. The low cost and high accuracy of these approximations make possible an enumeration strategy for identifying the faulty elements, which would have been computationally prohibitive were EM‐simulated patterns to be used. Our method was robust in handling arrays of double the size considered in Patnaik et al., IEEE Trans Antennas Propag 55 (2007), 775–777, while expanding on (Patnaik et al., IEEE Trans Antennas Propag 55 (2007), 775–777) by also addressing partial faults and measurement noise. Accuracies in detecting up to three faults (including partial ones) in arrays of 16 and 32 elements exceeded 97% under noise‐free conditions, and were above 93% in the presence of 2 dB measurement noise. © 2016 Wiley Periodicals, Inc. Int J RF and Microwave CAE 26:683–689, 2016.     

Generating Communication for Array Statements: Design,Implementation, and Evaluation

Array statements as included in Fortran 90 or High Performance Fortran (HPF) are a well-accepted way to specify data parallelism in programs. When generating code for such a data parallel program for a private memory parallel system, the compiler must determine when array elements must be moved from one processor to another. This paper describes a practical method to compute the set of array elements that are to be moved; it covers all the distributions that are included in HPF: block, cyclic, and block-cyclic. This method is the foundation for an efficient protocol for modern private memory parallel systems: for each block of data to be sent, the sender processor computes the local address in the receiver′s address space, and the address is then transmitted together with the data. This strategy increases the communication load but reduces the overhead on the receiving processor. We implemented this optimization in an experimental Fortran compiler, and this paper reports an empirical evaluation on a 64-node private memory iWarp system, using a number of different distributions.