A comparison by practical evaluation of several general-purpose distributed simulation algorithms that use different ways of bringing parallelism to the simulation of computer networks is presented. A description is g...
详细信息
ISBN:
(纸本)0818620412
A comparison by practical evaluation of several general-purpose distributed simulation algorithms that use different ways of bringing parallelism to the simulation of computer networks is presented. A description is given of the characteristics of computer networks relevant to distributed simulation, an approach to their modeling, and the characteristics of the parallel computer architecture used. The authors' simulation goals and how these influence the simulator are described. The results and conclusions of the comparison of different strategies for parallelization are also given.
Multicore systems have become standard for desktop computers today. Current operating systems and software development tools provide straightforward means to use the additional computing power. However, a more fundame...
详细信息
ISBN:
(纸本)9780769535739
Multicore systems have become standard for desktop computers today. Current operating systems and software development tools provide straightforward means to use the additional computing power. However, a more fundamental change in the design and development of software is required to fully exploit the power of multicore systems. Furthermore, the fast growing market of embedded systems is currently largely unaffected by the introduction of multicore systems. This will change quickly in the future, which will mean that there will be a demand on efficient development of reliable embedded software that can give real-time guarantees and exploit the available power on multicore systems. The JEOPARD project addresses this demand by developing Java software tools to exploit multicore power while ensuring correctness and predictable timing. This paper gives an overview of the JEOPARD project and focuses on key technical issues such as real-time scheduling and real-time garbage collection on multi-core systems.
This work highlights and takes aim at the most critical security aspects required for two different types of distributedsystems for scientific computation. It covers two open-source systems written in Java: a demand-...
详细信息
ISBN:
(纸本)9780769534725
This work highlights and takes aim at the most critical security aspects required for two different types of distributedsystems for scientific computation. It covers two open-source systems written in Java: a demand-driven system - General Intensional Programming System (GIPSY) and a pipelined system - distributed Modular Audio Recognition Framework (DMARF), which are the distributed scientific computational engines used as case studies with respect to the security aspects. More specific goals include data/demand integrity, data/demand origin authentication, confidentiality, high availability, and malicious code detection. We address some of the goals to a degree, some with the Java Data Security Framework (JDSF) as a work-in-progress.
Matrix computation is considered to be the core of many machine learning and graph algorithm workloads. In traditional single-node age, numerical analysis platforms like R and Matlab provide matrix programming model n...
详细信息
ISBN:
(纸本)9781509036820
Matrix computation is considered to be the core of many machine learning and graph algorithm workloads. In traditional single-node age, numerical analysis platforms like R and Matlab provide matrix programming model natively. As data is increasingly scaled up in the Big Data era, there is an increasing demand to seamlessly integrate large-scale matrix computation into distributed data-parallel computing systems. Therefore a variety of matrix computation libraries have been implemented on these distributed computing platforms such as MPI, Hadoop and Spark. However, a specific matrix-based algorithm has quite different performance over different platforms and it is very challenging for data scientists to specify the platform or combination of platforms for a given algorithm workflow to achieve the best performance. To solve this problem, in this paper, we put forward a time-cost based scheduling framework that can automatically specify the best platforms for the matrix operations and schedule the execution workflow. We have implemented a system prototype which using R as the user language and MPI, R and Spark as the backend computing platforms. The experimental results show that our time-cost based model has good accuracy with less than 10% error rate on average. Moreover, the scheduling framework built on it achieves efficient performance in applications.
A novel bitstream generation algorithm and its software implementation are introduced. Although this tool was developed for the configuration of AMDREL FPGA reconfigurable platform [13], it could be used to program an...
详细信息
ISBN:
(纸本)0769523129
A novel bitstream generation algorithm and its software implementation are introduced. Although this tool was developed for the configuration of AMDREL FPGA reconfigurable platform [13], it could be used to program any other compatible device. This tool is the only one known academic implementation for FPGA configuration with such features. Among them are the run-time-, partial- and dynamic-reconfiguration, the memory management, the bitstream compression and encryption, the read-back technique, the bitstream reallocation, the used low-power techniques as well as the Graphical User Interface.
作者:
You, LingFujian Univ
Yango Univ Informat Engn Coll Engn Res Ctr Spatital Data Min & Applicat Fuzhou 350001 Fujian Peoples R China
During application development, various unknown factors can affect the application and cause failures. According to the traditional softwareengineering techniques, software testing is usually performed later in softw...
详细信息
During application development, various unknown factors can affect the application and cause failures. According to the traditional softwareengineering techniques, software testing is usually performed later in software development, and it is difficult to find and correct these failures at this time. More and more distributedsoftwaresystems are deployed on public cloud computing platforms and use the Internet to provide services outward. The complexity, dynamics, and openness of cloud computing environments make distributedsoftwaresystems more prone to failure, which can result in service failures that can affect the normal use of large numbers of users. Therefore, in this paper, we study the fault detection algorithm of multi-channel parallel data flow software based on cloud computing. The data fusion and clustering models are integrated to detect the data features with the numerical verification, and the theoretical framework is also assigned to a better robustness for the theoretical framework implementation. The final results are also validated through the systematic overview. Future research directions are also included in the conclusion.
In recent years our society has witnessed an unprecedented growth in computing power available to tackle important problems in science, engineering and medicine. For example, the SHARCNET network links large computing...
详细信息
ISBN:
(纸本)0769523439
In recent years our society has witnessed an unprecedented growth in computing power available to tackle important problems in science, engineering and medicine. For example, the SHARCNET network links large computing resources in 11 leading academic institutions in South Central Ontario, thus providing access to thousands of compute processors. It is a continuous challenge to develop efficient and scalable algorithms and methods for solving large scientific and engineering problems on such parallel and distributed computers. If the computing power available in such computational grids can be unleashed effectively in a scalable way, large scientific problems can be solved that would otherwise be hard to solve using the machines available in a stand-alone way. This paper describes techniques and software developed that allow to apply the power of computational grids to large-scale, loosely coupled parallel bioinformatics problems. Our approach is based on decentralization and implemented in Java, leading to a flexible, portable and scalable software solution for parallel bioinformatics. We discuss advantages and disadvantages of this approach, and demonstrate seamless performance on an ad-hoc grid composed of a wide variety of hardware for a real-life parallel bioinformatics problem. The bioinformatics problem described consists of virtual experiments in RNA folding executed on hundreds of compute processors concurrently, which may establish one of the missing links in the chain of events that led to the origin of life.
This paper presents the development of distributed simulation architecture and testbed by combining three different types of tools based on three different approaches to real-time simulation: queuing theory (with SES/...
详细信息
ISBN:
(纸本)0769520367
This paper presents the development of distributed simulation architecture and testbed by combining three different types of tools based on three different approaches to real-time simulation: queuing theory (with SES/workbench as supporting tool), finite state machines (with high-level software design tools, such as Rational Rose Realtime or ObjecTime Developer), and continuous simulation based on solving differential equations (injecting real-time data from a real-time simulator running on a VMEbus platform under VxWorks real-time kernel). Such a combination is very advantageous because it provides the capability of making real-time decisions within the design and simulation tools, which can help significantly improve the quality of simulations.
In a real-rime system, tasks are constrained by global end-to-end (E-T-E) deadlines. In order to cater for high task schedulability, these deadlines must be distributed over component tasks in an intelligent way. In t...
详细信息
ISBN:
(纸本)0769501435
In a real-rime system, tasks are constrained by global end-to-end (E-T-E) deadlines. In order to cater for high task schedulability, these deadlines must be distributed over component tasks in an intelligent way. In this paper, we present an improved version of the slicing technique and extend it to heterogeneous distributed hard real-time systems. The salient feature of the new technique is that it utilizes adaptive metrics for assigning local task deadlines. Using experimental results rue show that the new technique exhibits superior performance with respect to the success ratio of a heuristic scheduling algorithm. For smaller systems, the new adaptive metric outperforms a previously-proposed adaptive metric by 300%, and existing non-adaptive metrics by more than an order of magnitude. In addition, the new technique is shown to be extremely robust for various system configurations.
Complex real-time embedded systems require guarantees regarding the assurance of their timing requirements. Such guarantees can be derived using advanced design and analysis methods. Many design solutions address the ...
详细信息
暂无评论