Performance analysis and its impact on design

Methods for designing new computer systems have changed rapidly. Consider general purpose microprocessors: gone are the days when one or two expert architects would use hunches, experience, and rules of thumb to determine a processor's features. Marketplace competition has long since forced companies to replace this ad hoc process with a targeted and highly systematic process that focuses new designs on specific workloads. Although the process differs from company to company, there are common elements. The main advantage of a systematic process is that it produces a finely tuned design targeted at a particular market. At its core are models of the processor's performance and its workloads. Developing and verifying these models is the domain now called performance analysis. We cover some of the advances in dealing with modern problems in performance analysis. Our focus is on architectural performance, typically measured in cycles per instruction     

Modular discriminant analysis and its applications

We propose a multi-linear algebra based subspace learning approach for finding linear projection which preserves some implicit structural or locally-spatial information among the original feature space. Our method uses a new tensor data representation model, in which, each group of data points are partitioned into several equal-sized sub-groups with its neighbors affiliated to them, and all sub-groups are concatenated to represent as the tensor space product of the original feature space. Then, a new optimization algorithm called Lagrangian multiplier mode (L-mode) is presented for computing the optimal linear projections. We show that our method has three ways for resolving the Small Sample Size problem: by applying the fuzzy matrix model to avoid the disturbance from non-interested determinant, by a quadratic sample correlation model, and by projecting the samples into a manifold using linear programming. Extensive experimental results conducted on two benchmark face biometrics datasets i.e. Yale-B and CMU-PIE, and a nutrition surveillance dataset demonstrate that our method is effective and robust than the state-of-the-arts such as Principal Component Analysis, Linear Discriminant Analysis, Locality Preserving Projections and their variations on both classification accuracies and computational expenses.     

超混沌Lü系统同步控制与应用

为提高混沌系统在加密算法的不可预测性和同步性能,采用状态反馈控制方法构造了一个新的超混沌Lü系统,通过分析该系统的平衡点性质、李亚普诺夫指数、功率谱和耗散性等指标证明了混沌吸引子的存在性,验证了所提超混沌系统具有更复杂的动力学行为;根据李亚普诺夫稳定原理,从理论上讨论了状态反馈控制方法实现超混沌同步的充分必要条件,并根据其条件进一步构造了超混沌Lü系统状态反馈同步控制器,仿真结果表明所提同步方案具有良好的噪声鲁棒性,并利用硬件电路实验验证了同步结果的正确性;将所提的超混沌Lü系统和对应的同步方案应用于保密通信方案中,展现出了所提方案的工程应用潜力。     

Distance-based discriminant analysis method and its applications

This paper proposes a method of finding a discriminative linear transformation that enhances the data’s degree of conformance to the compactness hypothesis and its inverse. The problem formulation relies on inter-observation distances only, which is shown to improve non-parametric and non-linear classifier performance on benchmark and real-world data sets. The proposed approach is suitable for both binary and multiple-category classification problems, and can be applied as a dimensionality reduction technique. In the latter case, the number of necessary discriminative dimensions can be determined exactly. Also considered is a kernel-based extension of the proposed discriminant analysis method which overcomes the linearity assumption of the sought discriminative transformation imposed by the initial formulation. This enhancement allows the proposed method to be applied to non-linear classification problems and has an additional benefit of being able to accommodate indefinite kernels.     

The theoretical analysis of GLRAM and its applications

Matrix-based methods such as two-dimensional principal component analysis (2DPCA) and generalized low rank approximations of matrices (GLRAM) have gained wide attention from researchers due to their computational efficiency. In this paper, we propose a non-iterative algorithm for GLRAM. Firstly, the optimal property of GLRAM is revealed, which is closely related to PCA. Moreover, it also shows that the reconstruction error of GLRAM is not smaller than that of PCA when considering the same dimensionality. Secondly, a non-iterative algorithm for GLRAM is derived. And the proposed method obtains smaller reconstruction error than 2DPCA or GLRAM. Finally, experimental results on face images and handwritten numeral characters show that the proposed method can achieve competitive results with some existing methods such as 2DPCA and PCA in terms of the classification performance or the reconstruction error.     

Characteristic analysis of Otsu threshold and its applications

This paper proves that Otsu threshold is equal to the average of the mean levels of two classes partitioned by this threshold. Therefore, when the within-class variances of two classes are different, the threshold biases toward the class with larger variance. As a result, partial pixels belonging to this class will be misclassified into the other class with smaller variance. To address this problem and based on the analysis of Otsu threshold, this paper proposes an improved Otsu algorithm that constrains the search range of gray levels. Experimental results demonstrate the superiority of new algorithm compared with Otsu method.     

Positive-realness analysis of sampled-data systems and its applications

In this paper, we first study strong positive-realness of sampled-data systems and introduce a measure called positive-realness gap index. We show that this index can be computed efficiently with a bisection method, and provide state space formulas for its computation. The importance of this index lies in that it is useful for robust stability analysis of sampled-data systems. An iterative procedure for computing an exact robust stability margin is given and illustrated through a numerical example.     

Review on stochastic approach to round-off error analysis and its applications

Because of round-off error propagation in floating-point arithmetic, any result provided by a computer always contains an error. The Permutation-Perturbation method, also known under the CESTAC (Contrôle et Estimation Stochastique des Arrondis de Calcul) method is a precious tool: (i) for evaluating the accuracy of results provided by direct algorithms performed by a computer, (ii) for breaking off correctly any iterative process, and for evaluating the accuracy of the results, and (iii) for determining the optimal step or the optimal mesh in the approximate algorithms. A review of this method and of these applications is presented in this paper.     

Performance analysis of TCP-friendly AIMD algorithms for multimedia applications

In this paper, the performance of TCP-friendly generic AIMD (Additive Increase and Multiplicative Decrease) algorithms for Web-based playback and multirate multimedia applications is investigated. The necessary and sufficient TCP-friendly condition is derived, and the effectiveness and responsiveness of AIMD are studied. Due to practical implications, a Dynamic TCP-friendly AIMD (DTAIMD) algorithm is proposed. Extensive simulation results are given to verify the derived necessary and sufficient condition, and to demonstrate the performance of the proposed DTAIMD algorithm.     

Performance and scalability analysis of image processing applications on the IBM Cell BE simulator

The IBM Cell Broadband Engine （BE） simulator simulates parallel execution of programs on a state-of-the-art 9-core Cell processor model. In this paper, we report our experience with implementing, simulation, and analyzing the performance of image processing applications on the IBM Cell Broadband Engine Simulator measured on the IBM Cell We report the performance simulator for PPE-only and embedded applications, and with various input data file sizes or numbers of SPEs enabled. The simulator results indicate that Cell BE processor can outperform modern single-core RISC processors in orders of magnitude on SIMD compute intensive applications such as edge detection. We also explore different features and development processes available on the simulator. Different techniques for obtaining accurate results （close to real hardware result） are also explored.     

Graph regularized GM-pLSA and its applications to video content analysis

As standard probabilistic latent semantic analysis (pLSA) is oriented to discrete quantity only, pLSA with Gaussian mixtures (GM-pLSA) succeeding in transferring it to continuous feature space is proposed, which uses Gaussian mixture model to describe the feature distribution under each latent aspect. However, inheriting from pLSA, GM-pLSA still overlooks the intrinsic interdependence between terms, which indeed is an important clue for performance improvement. In this paper, we present a graph regularized GM-pLSA (GRGM-pLSA) model as an extension of GM-pLSA to embed this term correlation information into the process of model learning. Specifically, grounded on the manifold regularization principle, a graph regularizer is introduced to characterize the correlation between terms; by imposing it on the objective function of GM-pLSA, model parameters of GRGM-pLSA are derived via corresponding expectation maximization algorithm. Furthermore, two applications to video content analysis are devised. One is video categorization where GRGM-pLSA serves for feature mapping with two kinds of sub-shot correlations, respectively, incorporated, while the other provides a new perspective on video concept detection, which transforms the detection task to a GRGM-pLSA-based visual-to-textual feature conversion problem. Extensive experiments and comparison with GM-pLSA and several state-of-the-art approaches in both applications demonstrate the effectiveness of GRGM-pLSA.     

Time-domain analysis methodology for large-scale RLC circuits and its applications

With soaring work frequency and decreasing feature sizes, VLSI circuits with RLC parasitic components are more like analog circuits and should be carefully analyzed in physical design. However, the number of extracted RLC components is typically too large to be analyzed efficiently by using present analog circuit simulators like SPICE. In order to speedup the simulations without error penalty, this paper proposes a novel methodology to compress the time-descritized circuits resulted from numerical integration approximation at every time step. The main contribution of the methodology is the efficient structure-level compression of DC circuits containing many current sources, which is an important complement to present circuit analysis theory. The methodology consists of the following parts: 1) An approach is proposed to delete all intermediate nodes of RL branches. 2) An efficient approach is proposed to compress and back-solve parallel and serial branches so that it is error-free and of linear complexity to analyze circuits of tree topology. 3) The Y toπtransformation method is used to error-free reduce and back-solve the intermediate nodes of ladder circuits with the linear complexity. Thus, the whole simulation method is very accurate and of linear complexity to analyze circuits of chain topology. Based on the methodology, we propose several novel algorithms for efficiently solving RLC-model transient power/ground (P/G) networks. Among them, EQU-ADI algorithm of linear-complexity is proposed to solve RLC P/G networks with mesh-tree or mesh-chain topologies. Experimental results show that the proposed method is at least two orders of magnitude faster than SPICE while it can scale linearly in both time- and memory-complexity to solve very large P/G networks.     

Power-law relationship and self-similarity in the itemset support distribution: analysis and applications

In this paper, we identify and explore that the power-law relationship and the self-similar phenomenon appear in the itemset support distribution. The itemset support distribution refers to the distribution of the count of itemsets versus their supports. Exploring the characteristics of these natural phenomena is useful to many applications such as providing the direction of tuning the performance of the frequent-itemset mining. However, due to the explosive number of itemsets, it is prohibitively expensive to retrieve lots of itemsets before we identify the characteristics of the itemset support distribution in targeted data. As such, we also propose a valid and cost-effective algorithm, called algorithm PPL, to extract characteristics of the itemset support distribution. Furthermore, to fully explore the advantages of our discovery, we also propose novel mechanisms with the help of PPL to solve two important problems: (1) determining a subtle parameter for mining approximate frequent itemsets over data streams; and (2) determining the sufficient sample size for mining frequent patterns. As validated in our experimental results, PPL can efficiently and precisely identify the characteristics of the itemset support distribution in various real data. In addition, empirical studies also demonstrate that our mechanisms for those two challenging problems are in orders of magnitude better than previous works, showing the prominent advantage of PPL to be an important pre-processing means for mining applications.     

Performance and energy effects on task-based parallelized applications

Heterogeneity, parallelization and vectorization are key techniques to improve the performance and energy efficiency of modern computing systems. However, programming and maintaining code for these architectures poses a huge challenge due to the ever-increasing architecture complexity. Task-based environments hide most of this complexity, improving scalability and usage of the available resources. In these environments, while there has been a lot of effort to ease parallelization and improve the usage of heterogeneous resources, vectorization has been considered a secondary objective. Furthermore, there has been a swift and unstoppable burst of vector architectures at all market segments, from embedded to HPC. Vectorization can no longer be ignored, but manual vectorization is tedious, error-prone and not practical for the average programmer. This work evaluates the feasibility of user-directed vectorization in task-based applications. Our evaluation is based on the OmpSs programming model, extended to support user-directed vectorization for different SIMD architectures (i.e., SSE, AVX2, AVX512). Results show that user-directed codes achieve manually optimized code performance and energy efficiency with minimal code modifications, favoring portability across different SIMD architectures.     

Performance analysis of Nitzberg's clutter map for Weibull distribution

The clutter map analyzed by Nitzberg, which averages the detector outputs of the same resolution cell over several past scans to set the detection threshold, plays an important role in radar detection fields. In this work, the average detection threshold (ADT) of Nitzberg's clutter map is derived and its steady-state property is analyzed, moreover, the detection performance of Nitzberg's clutter map for Weibull background is analyzed by the importance sampling technique.     

Performance analysis of server sharing collectives for content distribution

Demand for content served by a provider can fluctuate with time, complicating the task of provisioning serving resources so that requests for its content are not rejected. One way to address this problem is to have providers form a collective in which they pool together their serving resources to assist in servicing requests for one another's content. In this paper, we determine the conditions under which a provider's participation in a collective reduces the rejection rate of requests for its content - a property that is necessary for such a provider to justify its participation within the collective. We show that all request rejection rates are reduced when the collective is formed from a homogeneous set of providers, but that some rates can increase within heterogeneous sets. We also show that, asymptotically, growing the size of the collective will sometimes, but not always, resolve this problem. We explore the use of thresholding techniques, where each collective participant sets aside a portion of its serving resources to serve only requests for its own content. We show that thresholding allows a more diverse set of providers to benefit from the collective model, making collectives a more viable option for content delivery services.     

Object density-based image segmentation and its applications in biomedical image analysis

In many applications of medical image analysis, the density of an object is the most important feature for isolating an area of interest (image segmentation). In this research, an object density-based image segmentation methodology is developed, which incorporates intensity-based, edge-based and texture-based segmentation techniques. The proposed method consists of three main stages: preprocessing, object segmentation and final segmentation. Image enhancement, noise reduction and layer-of-interest extraction are several subtasks of preprocessing. Object segmentation utilizes a marker-controlled watershed technique to identify each object of interest (OI) from the background. A marker estimation method is proposed to minimize over-segmentation resulting from the watershed algorithm. Object segmentation provides an accurate density estimation of OI which is used to guide the subsequent segmentation steps. The final stage converts the distribution of OI into textural energy by using fractal dimension analysis. An energy-driven active contour procedure is designed to delineate the area with desired object density. Experimental results show that the proposed method is 98% accurate in segmenting synthetic images. Segmentation of microscopic images and ultrasound images shows the potential utility of the proposed method in different applications of medical image processing.     

Cyclic routing algorithms in graphs: Performance analysis and applications to robot scheduling

In this paper we revisit and extend the algorithm for the cyclic project scheduling problem which was originally proposed by Romanovskii (1967). While the algorithm has been derived for fixed numerical data, we show how it can be extended to handle the problems with interval data. We also propose a new algorithm for the cyclic scheduling problem with interval data that extends the parametric method developed by Megiddo (1979) and runs in strongly polynomial time.     

Performance issues and performance analysis tools for HPC cloud applications: a survey

Cloud Computing is an eminent emerging technology that surpasses Grids from their IT resource administrations and arduous Grid middleware solutions. At present, users could access an abundant number of pre-defined cloud services or run their programs on demand as a pay-as-you-go processing model without much distribution problems. In addition, the IT business market has pumped enough revenue for establishing salient common-use cloud solutions. Despite adequate researchers have been involved in the cloud development, scientific application developers are still reluctant to execute their applications in the cloud due to the performance concerns, such as, scalability, availability, and service level agreement violations of the cloud providers. In this paper, a survey of various High Performance Computing (HPC) applications and possible performance concerns while executing applications in cloud is presented. Pointing out the need for Performance Analysis (PA) tools, this paper focuses on the study of cloud-based PA tools in detail. This paper could leverage HPC application developers to cope with the performance issues and to best utilize the available performance analysis tools of clouds.