Dynamic optimizers for object-oriented languages collect a variety of profile data to drive optimization decisions. In particular, the dynamic call graph (DCG) informs key structural optimizations such as which methods to optimize and how to optimize them. Unfortunately, current low-overhead call-stack hardware and software sampling methods are subject to sampling bias, which loses accuracy of 40 to 50% when compared with a perfect call graph. This paper introduces DCG correction, a novel approach that uses static and dynamic control-flow graphs (CFGs) to improve DCG accuracy. We introduce the static frequency dominator (FDOM) relation, which extends the dominator relation on the CFG to capture relative execution frequencies and expose static constraints on DCG edges, which we use to correct DCG edge frequencies. Using conservation of flow principles, we further show how to use dynamic CFG basic block profiles to correct DCG edge frequencies intraprocedurally and interprocedurally. We implement and evaluate DCG correction in Jikes RVM on the SPEC JVM98 and DaCapo benchmarks. Default DCG sampling attains an average accuracy of 52-59% compared with perfect, whereas FDOM correction improves average accuracy to 64-68%, while adding 0.2% average overhead. The dynamic correction raises accuracy to 85% on average, while adding 1.2% average overhead. We then provide dynamically corrected DCGs to the inliner with mixed results -1% average degradations and improvements across a variety of configurations. However, prior work shows that increased DCG accuracy in production VMs has benefits. We believe that high-accuracy DCGs will become more important in the future as the complexity and modularity of object-oriented programs increases.

Not available / text

published_or_final_version / Computer Science / Master / Master of Philosophy

No description available.

A new approach is presented for object based interprocess communication in this study. UNIX is a conventional operating system, and this study presents a new approach for representing processes as objects.Object based programming at the system level is discussed. The thesis presents motivation for supporting and implementing process objects with UNIX system calls, message passing and shared memory. Interprocess communication can be simulated using objects to represent processes.Related work in the area of system supported objects in conventional operating systems is also presented. / Department of Computer Science

In many software applications, an object-oriented design (OOD) is generated first, then persistent storage is implemented by mapping the objects to a relational database. This thesis explores the "reverse engineering" of an OOD out of an existing relational data model. Findings from the current literature are presented, and a case study is undertaken using the model and research process published by GENTECH, a nonprofit organization promoting genealogical computing. The model is mapped into an OOD and captured in Unified Modeling Language (UML) class diagrams and object collaboration diagrams. The suitability of the example OOD is evaluated against the GENTECH research process using UML use cases and sequence diagrams. The mapping of relational database designs into OODs is found to be suitable in certain instances. / Department of Computer Science

No description available.

No description available.

The need for a product model which can support the modelling requirements of a broad range of applications leads to the application of a feature-based model. An important requirement in feature-based design and manufacture is that a single feature representation should be capable of supporting a number of different applications. The capability of representing products composed of assemblies is seen to be necessary to serve the information needs of those applications. To achieve this aim it is an essential prerequisite to develop a formal structure for the representation of assembly information in a feature-based design system. This research addresses two basic questions related to the lack of a unified definition for features and the problem of representing assemblies in a feature-based representation. The intention is to extend the concept of designing with features by incorporating assembly information in addition to the geometrical and topological details of component parts. This allows models to be assembled using the assembly information within the feature definitions. Features in this research are defined as machined volumes which are represented in a hierarchical taxonomy. The taxonomy includes several types and profiles of features which cover a general range of machined parts. A hierarchical assembly structure is also defined in which features form basic entities in the assembly. Each feature includes information needed to establish assembly relationships among features in the form of mating relationships. An analysis of typical assemblies shows that assembly interfaces occur at the face level of the mating features and between features themselves. Three mating relationships between pairs of features have been defined (against, fits and align) and are represented in the form of expressions that can be used for evaluations. Various sub-types of these major mating relationships can be identified (e.g. tight fit, clearance fit, etc.) and represented through the use of qualifying attributes. Component Relation Graphs, Feature Relation Graphs and Face Mating Graphs have been developed to represent each level of interaction in an assembly, and assembly relationships are combined with knowledge on process planning into a Component Connectivity Graph. These graphs are used as the basis for deriving an integrated data structure which is used for defining classes for each level in the assembly hierarchy. The implementation of a prototype system has been facilitated by use of an object-oriented programming technique which provides a natural method of adding functionality to the geometric reasoning process of features and the complex relationships between the parts that make up the assembly. The feature-based model is embedded in an object-oriented solid modeller kernel, ACIS®. The research demonstrates the possibilities for a single feature representation to support multiple activities within a computer integrated manufacturing environment. Such a representation can form the basis of design improvement techniques and manufacturing planning as well as be a model to support the life cycle of the product.

This thesis demonstrates the adequacy of an object-oriented approach to the construction of distributed metainformation systems: systems that facilitate information use by maintaining some information about the information. Computer systems are increasingly being used to store information objects and make them accessible via network. This access, however, still relies on an adequate metainformation system: there must be an effective means of specifying relevant information objects. Moreover, distribution requires the metainformation system to cope well with intermittent availability of network resources. Typical metainformation systems developed to date permit information objects to be specified by expressing knowledge about their syntactic properties, such as keywords. Within this approach, however, query results are potentially too large to be transmitted, stored and treated, at reasonable cost and time. Users are therefore finding it difficult to navigate their way through the masses of information available. In contrast, this thesis is based on the principle that a metainformation system IS more effective if it permits information objects to be specified according to their semantic properties, and that this helps managing, filtering and navigating information. Of particular interest is object orientation because it is the stateof- the-art approach to both the representation of information semantics and the Abstract 11 design of reliable systems. The thesis presents the design and implementation of a programming toolkit for the construction of metainformation systems, where information objects can be any entity that contains information, the notion of views permits organising the information space, transactional access is employed to obtain consistency, and replication is employed to obtain high availability and scalability.