A maior parte dos mecanismos em processadores superescalares atuais usam granularidade de instrução para criar ou caracterizar especulações, tais como predição de desvios ou prefetchers. No entanto, muitas das características das instruções podem ser obtidas ao analisar uma granularidade mais grossa, o bloco básico de código, aumentando a quantidade de código coberta em um espaço similar de armazenamento. Adicionalmente, códigos podem ser analisados mais precisamente e prover uma variedade maior de informação ao observar diferentes tipos de instruções e suas relações. Devido a estas vantagens, a análise no nível de blocos pode fornecer mais oportunidades para mecanismos que necessitam desta informação. Por exemplo, é possível integrar informações de desvios mal previstos e acessos a memória para gerar informações mais precisas de quais acessos a memória oferecem melhor desempenho ao serem priorizados. Nesta tese propomos o Block-Level Architecture Profiler (BLAP) (Block Level Architecture Profiler), um mecanismo em hardware que caracteriza gargalos no nível microarquitetural, tal como loads delinquentes, desvios de difícil previsão e contenção nas unidades funcionais. O BLAP trabalha no nível de bloco básico, apenas detectando e fornecendo informações que podem ser usada para otimizar tais gargalos. Um mecanismo para a remoção de prefetches e uma política de controlador de memória DRAM foram criados para usar a informação criada pelo BLAP e demonstrar seu potencial. Juntos, estes mecanismos são capazes de melhorar o desempenho do sistema em até 17.39% (3.9% em média). Nosso método mostrou também ganhos médios de 13.14% quando avaliado com uma pressão na memória mais alta devido a prefetchers mais agressivos. / Most mechanisms in current superscalar processors use instruction granularity information for speculation, such as branch predictors or prefetchers. However, many of these characteristics can be obtained at the basic block level, increasing the amount of code that can be covered while requiring less space to store the data. Moreover, the code can be profiled more accurately and provide a higher variety of information by analyzing different instruction types inside a block. Because of these advantages, block-level analysis can offer more opportunities for mechanisms that use this information. For example, it is possible to integrate information about branch prediction and memory accesses to provide precise information for speculative mechanisms, increasing accuracy and performance. We propose a BLAP, an online mechanism that profiles bottlenecks at the microarchitectural level, such as delinquent memory loads, hard-to-predict branches and contention for functional units. BLAP works at the basic block level, providing information that can be used to reduce the impact of these bottlenecks. A prefetch dropping mechanism and a memory controller policy were developed to use the profiled information provided by BLAP. Together, these mechanisms are able to improve performance by up to 17.39% (3.90% on average). Our technique showed average gains of 13.14% when evaluated under high memory pressure due to highly aggressive prefetch.

Ever-increasing demands for high transaction rates, limitations of high-end processors, high availability, and modular growth considerations are all driving forces toward distributed architectures for transaction processing. However, a prerequisite to taking advantage of the capacity of a distributed transaction processing system is an effective strategy for workload allocation. The distribution of the workload should not only achieve load balancing, but also support an efficient transaction processing with a minimum of intersystem communication. To this end, adaptive schemes for transaction routing have to be employed that are highly responsive to workload fluctuations and configuration changes. Adaptive allocation schemes are also important for simplifying system administration, which is a major problem in distributed transaction processing systems. In this article we develop a taxonomic framework for workload allocation, in particular, transaction routing, in distributed transaction processing systems. This framework considers the influence of the underlying system architecture (e.g., shared nothing, shared disk) and transaction execution model as well as the major dependencies between workload, program, and data allocation. The main part of the framework covers structural (or architectural) and implementational alternatives for transaction routing to help identify key factors and basic tradeoffs in the design of appropriate allocation schemes. Finally, we show how existing schemes fit our taxonomy. The framework substantially facilitates a comparison of the different schemes and can guide the development of new, more effective protocols.

<p>GPGPU is a highly multithreaded throughput architecture that can deliver high speed-up for regular applications while remaining energy efficient. In recent years, there has been much focus on tuning irregular applications and/or the GPU architecture to achieve similar benefits for irregular applications as well as efforts to extract data parallelism from task parallel applications. In this work we tackle both problems.</p><p>The first part of this work tackles the problem of Control divergence in GPUs. GPGPUs’ SIMT execution model is ineffective for workloads with irregular control-flow because GPGPUs serialize the execution of divergent paths which lead to thread-level parallelism (TLP) loss. Previous works focused on creating new warps based on the control path threads follow, or created different warps for the different paths, or ran multiple narrower warps in parallel. While all previous solutions showed speedup for irregular workloads, they imposed some performance loss on regular workloads. In this work we propose a more fine-grained approach to exploit <i>intra-warp</i>convergence: rather than threads executing the same code path, <i>opcode-convergent threads</i>execute the same instruction, but with potentially different operands. Based on this new definition we find that divergent control blocks within a warp exhibit substantial opcode convergence. We build a compiler that analyzes divergent blocks and identifies the common streams of opcodes. We modify the GPU architecture so that these common instructions are executed as convergent instructions. Using software simulation, we achieve a 17% speedup over baseline GPGPU for irregular workloads and do not incur any performance loss on regular workloads.</p><p>In the second part we suggest techniques for extracting data parallelism from irregular, task parallel applications in order to take advantage of the massive parallelism provided by the GPU. Our technique involves dividing each task into multiple sub-tasks each performing less work and touching a smaller memory footprint. Our framework performs a locality-aware scheduling that works on minimizing the memory footprint of each warp (a set of threads performing in lock-step). We evaluate our framework with 3 task-parallel benchmarks and show that we can achieve significant speedups over optimized GPU code.</p>

Indiana University-Purdue University Indianapolis (IUPUI) / Systems Engineering is catching pace in many segments of product manufacturing industries. Model-Based Systems Engineering (MBSE) is the formalized application of modeling to perform systems engineering activities. In order to effectively utilize the complete potential of MBSE, a methodology consisting of appropriate processes, methods and tools is a key necessity. In the last decade, several MBSE projects have been implemented in industries varying from aerospace and defense to automotive, healthcare and transportation. The Systems Modeling Language (SysML) standard has been a key enabler of these projects at many companies. Although SysML is capable of providing a rich representation of any system through various viewpoints, the journey towards adopting SysML to realize the true potential of MBSE has been a challenge. Among all, one of the common roadblocks faced by systems engineers across industries has been the software engineering-based nature of SysML which leads to difficulties in grasping the modeling concepts for people that do not possess a software engineering background. As a consequence, developing a system (or a system of systems) architecture model using SysML has been a challenging task for many engineers even after a decade of its inception and multiple successive iterations of the language specification. Being a modeling language, SysML is method-agnostic, but its associated limitations outweigh the advantages. ARCADIA (Architecture Analysis and Design Integrated Approach) is a systems and software architecture engineering method based on architecture-centric and model-based engineering activities. If applied properly, ARCADIA allows for a very effective way to model the architecture of multi-domain systems, and overcome many of the limitations faced in traditional SysML implementation. This thesis evaluates the architecture development capabilities of ARCADIA/Capella versus SysML following the Object-Oriented Systems Engineering Method (OOSEM). The study focuses on the key equivalences and differences between the two MBSE solutions from a model development perspective and provides several criteria to evaluate their effectiveness for architecture development using a conceptual case of Adaptive Cruise Control (ACC). The evaluation is based on three perspectives namely, architecture quality, ability to support key process deliverables, and the overall methodology. Towards this end, an industry-wide survey of MBSE practitioners and thought leaders was conducted to identify several concerns in using models but also to validate the results of the study. The case study demonstrates how the ARCADIA/Capella approach addresses several challenges that are currently faced in SysML implementation. From a process point of view, ARCADIA/Capella and SysML equally support the provision of the key deliverable artifacts required in the systems engineering process. However, the candidate architectures developed using the two approaches show a considerable difference in various aspects such as the mapping of the form to function, creating functional architectures, etc. The ARCADIA/Capella approach allows to develop a ‘good’ system architecture representation efficiently and intuitively. The study also provides answers to several useful criteria pertaining to the overall candidate methodologies while serving as a practitioner’s reference in selecting the most suitable approach.

no / As part of a defense-in-depth security solution for domain-controlled enterprise networks, a proposed self-healing system architecture is designed to increase resiliency against botnets with minimal disruption to network services.

No description available.

Digital Libraries (DLs) are software systems specifically designed to assist users in information seeking activities. Stemming from the intersection of library sciences and computer networking, traditional DL systems impose library philosophies of structure and management on the sprawling collections of data that are made possible through the Internet. DLs evolve to keep pace with innovation on the Internet so there is little standardization in the architecture of such systems. However, in attempting to provide users with the highest possible levels of service with the minimum possible effort, many systems work collaboratively with others, e.g., meta-search engines. This type of system interoperability is encouraged by the emergence of simple data transfer protocols such as the Open Archives Initiative?s Protocol for Metadata Harvesting (OAI-PMH). Open Digital Libraries are an extension of the work of the OAI. It is proposed in this dissertation that the philosophy and approach adopted by the OAI can easily be extended to support inter-component interaction within a componentized DL. In particular, DLs can be built by connecting small components that communicate through a family of lightweight protocols, using XML as the data interchange mechanism. In order to test the feasibility of this, a set of protocols was designed based on a generalization of the work of the OAI. Components adhering to these protocols were implemented and integrated into production and research DLs. These systems were then evaluated for simplicity, reusability, and performance. On the whole, this study has shown promise in the approach of applying the fundamental concepts of the OAI protocol to the task of DL component design and implementation. Further, it has shown the feasibility of building componentized DL systems using techniques that are a precursor to the Web Services approach to system design. / Ph. D.

Electronic power conversion systems with large number of power converters have a variety of applications, such as data center, electric vehicles and future smart "nanogrid" in residential home. Those systems could have very different architectures. For example, one system could be based on ac, dc or hybrid power distribution bus, and the bus voltage could be different. Also those systems have great need to develop low-cost architectures which reduce weight, increase efficiency and improve reliability of the system. However, how to evaluate different architectures and select a better one is still not clear. This thesis presents a procedure to estimate weight of electronic power conversion systems, which provides an angle to evaluate different system architectures. This procedure has three steps. Step I, according to application of the system and system structure, determines the electrical and environmental specifications for each converter in the system. Step II studies the design procedures for each converter in the system and determines parameters such as the wire gauge and length of cable; the parameters of the passive components, such as inductance and capacitance; the parameters of the power switch, such as the voltage rating, current rating and loss; and parameters of the cooling system, such as the thermal resistance of the heat sink. Step III, according to the converters' parameters, carry out the physical design and selection of sub-components such as the inductor and heat sink to get the components' weight; the sum of those components' weight is the estimated system weight. This procedure has also been implemented in the form of software – system weight estimation tool. Using this software, weight of sample systems with ac dc bus and two different bus voltages have been estimated and compared. / Master of Science

<div>The internal and external interactions between the complex structural and behavioral characteristics of the system of interest and the surrounding environment result in unpredictable emergent behaviors. These emergent behaviors are not well understood, especially when modeled using the traditional top-down systems engineering approach. The intrinsic nature of current complex systems has called for an elegant solution that provides an integrated framework in Model-Based Systems Engineering. A considerable gap exists to integrate system engineering activities and engineering analysis, which results in high risk and cost. This thesis presents a framework that incorporates indefinite and definite modeling aspects that are developed to determine the complexity that arises during the development phases of the system. This framework provides a workflow for modeling complex systems using Systems Modeling Language (SysML) that captures the system’s requirements, behavior, structure, and analytical aspects at both problem definition and solution levels. This research introduces a new level/dimension to the framework to support engineering analysis integrated with the system architecture model using FMI standards. A workflow is provided that provides the enabling methodological capabilities. It starts with a statement of need and ends with system requirement verification. Detailed traceability is established that glues system engineering and engineering analysis together. Besides, a method is proposed for predicting the system’s complexity by calculating the complexity index that can be used to assess the complexity of the existing system and guide the design and development of a new system. To test and demonstrate this framework, a case study consisting of a complex district cooling system is implemented. The case study shows the framework’s capabilities in enabling the successful modeling of a complex district cooling system. The system architecture model was developed using SysML and the engineering analysis model using Modelica. The proposed framework supports system requirements verification activity. The analysis results show that the district chiller model developed using Modelica produces chilled water below 6.6 degrees Celsius, which satisfies the system requirement for the district chiller system captured in the SysML tool. Similarly, many such requirement verification capabilities using dynamic simulation integration with the high-level model provides the ability to perform continuous analysis and simulation during the system development process. The systems architecture complexity index is measured for the district cooling case study from the black-box and white box-perspective. The measured complexity index showed that the system architecture’s behavioral aspect increases exponentially compared to the structural aspect. The systems architecture’s complexity index at black-box and white-box was 4.998 and 67.3927, respectively.</div>

Indiana University-Purdue University Indianapolis (IUPUI) / The internal and external interactions between the complex structural and behavioral characteristics of the system of interest and the surrounding environment result in unpredictable emergent behaviors. These emergent behaviors are not well understood, especially when modeled using the traditional top-down systems engineering approach. The intrinsic nature of current complex systems has called for an elegant solution that provides an integrated framework in Model-Based Systems Engineering. A considerable gap exists to integrate system engineering activities and engineering analysis, which results in high risk and cost. This thesis presents a framework that incorporates indefinite and definite modeling aspects that are developed to determine the complexity that arises during the development phases of the system. This framework provides a workflow for modeling complex systems using Systems Modeling Language (SysML) that captures the system’s requirements, behavior, structure, and analytical aspects at both problem definition and solution levels. This research introduces a new level/dimension to the framework to support engineering analysis integrated with the system architecture model using FMI standards. A workflow is provided that provides the enabling methodological capabilities. It starts with a statement of need and ends with system requirement verification. Detailed traceability is established that glues system engineering and engineering analysis together. Besides, a method is proposed for predicting the system’s complexity by calculating the complexity index that can be used to assess the complexity of the existing system and guide the design and development of a new system. To test and demonstrate this framework, a case study consisting of a complex district cooling system is implemented. The case study shows the framework’s capabilities in enabling the successful modeling of a complex district cooling system. The system architecture model was developed using SysML and the engineering analysis model using Modelica. The proposed framework supports system requirements verification activity. The analysis results show that the district chiller model developed using Modelica produces chilled water below 6.6 degrees Celsius, which satisfies the system requirement for the district chiller system captured in the SysML tool. Similarly, many such requirement verification capabilities using dynamic simulation integration with the high-level model provides the ability to perform continuous analysis and simulation during the system development process. The systems architecture complexity index is measured for the district cooling case study from the black-box and white box-perspective. The measured complexity index showed that the system architecture’s behavioral aspect increases exponentially compared to the structural aspect. The systems architecture’s complexity index at black-box and white-box was 4.998 and 67.3927, respectively.