一个支持软件并行工程的过程模型

实施软件并行工程是缩短软件开发周期、加快软件开发速度的有效途径。分析了存在于软件生存周期中的两类不同层次的并行性能提出了一个基于Ｐｅｔｒｉ网的支持软件并行工程的软件过程模型ＳＤＤＭ，给出了将软件过程内部的局部并行性延拓到该过程全局的方法。     

软件过程中的并行性挖掘

挖掘软件过程中的并行性,使其中的活动尽量并行进行,是提高软件生产率的重要手段。本文提出了一种通过活动间相关性分析,寻找软件过程中可并行化的因素,挖掘出可并行进行的活动,进而构造出Petri网表示的并行经的软件过程模型的技术,获得了较理想的并行性挖掘效果。     

并行测试任务可并行性分析研究

针对并行测试系统中不同测试序列测试效率提高存在差异的现象,分析、研究测试序列可并行性的概念,以描述测试序列在并行测试中的固有属性。建立的可并行度指标解决了测试序列测试效率提高不同无法解释的难题,对并行测试系统开发过程中仪器资源的选取与确定具有指导意义。通过测试序列实例,验证了可并行性是测试序列的固有属性,说明了可并行度概念的合理性。     

并行测试任务可并行性分析与应用

针对并行测试系统中不同UUT测试任务组的并行测试效率提高存在差异的现象,提出了测试任务的可并行性概念,以描述测试任务在并行测试过程中的固有属性;首先根据测试任务组的并行测试效率不同的现象,提出了测试任务可并行性的概念;然后给出了衡量可并行性的可并行度指标,用以定量地指导并行测试系统开发过程中仪器资源的配置;最后通过应用实例验证了可并行性概念的合理性、可并行度指标的实用性;建立的可并行性概念不仅丰富了测试领域的理论基础,对并行测试系统的仪器资源的配置也具有指导价值。     

基于谓词代码的编译优化技术研究

程序中大量分支指令的存在,严重制约了体系结构和编译器开发并行性的能力。有效发掘指令级并行性的一个主要挑战是要克服分支指令带来的限制。利用谓词执行可有效地删除分支,将分支指令转换为谓词代码,从而扩大了指令调度的范围并且删除了分支误测带来的性能损失。阐述了基于谓词代码的指令调度、软件流水、寄存器分配、指令归并等编译优化技术。设计并实现了一个基于谓词代码的指令调度算法。实验表明,对谓词代码进行编译优化,能有效提高指令并行度,缩短代码执行时间,提高程序性能。     

主—从分布式实时控制系统中的通信技术——PC机与单板机的通信技术

为了提高计算机控制系统的使用效率,计算机从单处理机系统和多计算机系统两个方面向着具有高并行度的分布处理系统发展。单处理机内部千方百计改进各种的功能部件以及采用时间重叠、资源共享等途径提高并行性,并发展了多终端,远程终端等技术,但要进一步提高系统的并行性,必然向     

程序潜在最大并行程度分析方法及其实现

现有的并行编译器不能充分地识别和开发程序潜在并行性。此文提出了一种新的程序最大并行度分析方法，这一方法可开发程序循环一级的并行性。     

向量并行度指导的循环SIMD向量化方法

SIMD扩展部件是集成到通用处理器中的加速部件,旨在发掘多媒体和科学计算等领域程序的数据级并行.当前两种基本的向量发掘方法分别是发掘迭代间并行的Loop-based方法和发掘迭代内并行的SLP方法.Loop-aware方法是对SLP方法的改进,其思想是首先通过循环展开将迭代间并行转换为迭代内并行,使循环体内的同构语句条数足够多,再利用SLP方法进行向量发掘.但当循环展开不合法或者并行度低于向量化因子时,Loop-aware方法无法实现程序向量并行性的发掘.因此提出了向量并行度指导的循环向量化方法,依据迭代间并行度、迭代内并行度和向量化因子,构建循环向量化方法选择方案,同时提出不充分向量化方法发掘并行度低于向量化因子的循环向量并行性,最后依据向量并行度对生成的向量循环进行展开.经过标准测试集测试,向量并行度指导的循环SIMD向量化方法比Loop-aware方法识别率提升107.5%,性能提升12.1%.     

一种基于非正规域的区域依赖关系分析法

在自动并行编译中，并行性的识别主要集中在循环及语句级，而许多程序实际上可通过挖掘子程序级这种“任务“并行性来提高性能。本文提出了基于非正规域的区域依赖分析方法，旨在发掘这类并行性，它能精确地刻划程序中的数据访问区域。克服了现有区域分析技术中趋于保守的弱点，从而提出了并行度，依赖关系的测试算法简单而有效。     

一种面向同步交互的软件演化过程建模方法

全球化软件开发导致以交互方式协作开发的频率和复杂性越来越高。为了控制和规范软件演化的开发行为进而提高软件质量,文献[10]设计了软件演化过程元模型EPMM,由EPMM定义的软件演化过程模型未能形式描述其交互的特点。基于EPMM[10]定义的软件演化过程包括全局层、过程层、活动层和任务层4个抽象层的思想,设计了软件演化过程元模型CEPMM。CEPMM定义的软件演化过程模型可 形式描述同步交互的特点是在其活动层,由此提出一种基于通信系统演算(CCS)的软件演化过程活动层的建模方法,然后在Visual Studio平台下实现软件演化过程活动层可视化的建模工具CAmodel。CEPMM构建的软件演化过程活动层模型不仅可以描述并发、迭代等特点,还可以形式描述同步交互的特性,为模型进行严格的数学方法分析、推理奠定了基础。     

基于COM原理的进程结构图呈现技术

对于具有分布、并发特征的面向对象系统,抽取和呈现进程间的创建和通信关系,可以从系统的高层体系结构方面帮助用户理解系统。在充分分析Rose扩充技术原理的基础上,给出了反应进程间创建和通信关系的进程结构图的一种呈现技术,通过扩充Rose的版型,将进程结构图在Rose中直观地呈现出来。最后给出了一个分布、并发的软件系统,作为测试案例,对该呈现技术的有效性进行了验证。     

A transformation‐based approach to testing concurrent programs using UML activity diagrams

Unified Modeling Language (UML) activity diagrams are widely used to model concurrent interaction among multiple objects. In this paper, we propose a transformation‐based approach to generating scenario‐oriented test cases for applications modeled by UML activity diagrams. Using a set of transformation rules, the proposed approach first transforms a UML activity diagram specification into an intermediate representation, from which it then constructs test scenarios with respect to the given concurrency coverage criteria. The approach then finally derives a set of test cases for the constructed test scenarios. The approach resolves the difficulties associated with fork and join concurrency in the UML activity diagram and enables control over the number of the resulting test cases. We further implemented a tool to automate the proposed approach and studied its feasibility and effectiveness using a case study. Experimental results show that the approach can generate test cases on demand to satisfy a given concurrency coverage criterion and can detect up to 76.5% of seeded faults when a weak coverage criterion is used. With the approach, testers can not only schedule the software test process earlier, but can also better allocate the testing resources for testing concurrent applications. Copyright © 2015 John Wiley & Sons, Ltd.     

Metrics for assessment of designs and implementations during reengineering of computer-based systems

Modern computer-based systems have many required characteristics, including performance, concurrency, timeliness, availability, dependability, safety and security. Aging computer-based systems are being reengineered to exploit distributed and highly parallel hardware platforms, in order to accommodate increased functional requirements and to achieve dependability. Simultaneously, reengineering is introducing modern software engineering principles such as component layering, encapsulation and information hiding, in order to reduce maintenance burdens and to simplify future enhancement. This paper summarizes a process for reengineering of computer-based systems to achieve these goals. The process incorporates experience gained through the reengineering of the Weapons Selection module and the Surface Operations module of the AEGIS Weapon System. A key component of the reengineering process is themission critical software architecture (MCSA), a hierarchical view of systems that includes the abstraction levels of programs, tasks, packages, procedures, and statements. To guide the reengineering processes of software transformation, concurrency metrics are defined at the procedure, package and task levels of the MCSA. The paper defines the intermediate representation (IR) needed to compute the metrics, and shows how to use the IR to compute several metrics: the percentage of concurrency within procedures and packages, and the amount of potential concurrency among tasks. The metrics are important in any reengineering approach that iteratively transforms design and code, or that deals with system configuration (the integration of software, hardware and humanware).This work is supported in part by the US NSWC (N00178-95-R-2007, N60921-94-M-6096, N60921-94-M-1960, N60921-94-M-2555, N60921-95-M-0311), by the State of New Jersey (SBR-421290), and by the US ONR(N00014-92-J-1367).     

Interactive State-Space Analysis of Concurrent Systems

The introduction of concurrency into programs has added to the complexity of the software design process. This is most evident in the design of communications protocols where concurrency is inherent to the behavior of the system. The complexity exhibited by such software systems makes more evident the need for computer-aided tools for automatically analyzing behavior.     

Tailored generation of concurrent benchmarks

In this paper, we show how to extend our approach to property-driven benchmark generation (PDBG) to support concurrency in the benchmarks: we systematically produce multi-process PROMELA code of known and tailored complexity that can then serve as benchmark for the evaluation of analysis and verification tools for concurrent software systems. Key to this extension is the way in which we control the corresponding property profiles, i.e., the statements about the validity of a predefined set of LTL properties, along the PDBG tool chain. It is based on the idea to introduce intricate structural elements like synchronization and interference at a local and therefore controllable level, only concerning a few parallel components, and subsequently obfuscating the locality of these structures at the global level via property-preserving transformations of both the system components and the properties. This leads to a scalable approach which we illustrate along some concrete examples.     

线程池中最大并发数的研究与设计

线程池中最大并发数影响线程效率,在实际的软件设计中,软件设计者往往依赖经验来设置线程池中的最大并发数,这就导致了软件设计的主观性和盲目性。将目标文件大小作为研究影响线程池最大并发数的一个影响因子,对目标文件与线程池最大并发数建模并分析,分析的结果表明目标文件大小在一定范围内与线程池的最大并发数成反比例关系,并对模型进行优化。提出根据目标文件的大小动态地设置线程池中的最大并发数的方法,同时对研究结果通过IOS平台的多线程下载模型进行了检验。     

Structural testing of concurrent programs

Although structural testing techniques are among the weakest available with regard to developing confidence in sequential programs, they are not without merit. The authors extend the notion of structural testing criteria to concurrent programs and propose a hierarchy of supporting structural testing techniques. Coverage criteria described include concurrency state coverage, state transition coverage and synchronization coverage. Requisite support tools include a static concurrency analyzer and either a program transformation system or a powerful run-time monitor. Also helpful is a controllable run-time scheduler. The techniques proposed are suitable for Ada or CSP-like languages. Best results are obtained for programs having only static naming of tasking objects