| 1. |
- Gebrewahid, Essayas, 1984-
(author)
-
Compiling Concurrent Programs for Manycores
- 2015
-
Licentiate thesis (other academic/artistic)abstract
- The arrival of manycore systems enforces new approaches for developing applications in order to exploit the available hardware resources. Developing applications for manycores requires programmers to partition the application into subtasks, consider the dependence between the subtasks, understand the underlying hardware and select an appropriate programming model. This is complex, time-consuming and prone to error.In this thesis, we identify and implement abstraction layers in compilation tools to decrease the burden of the programmer, increase programming productivity and program portability for manycores and to analyze their impact on performance and efficiency. We present compilation frameworks for two concurrent programming languages, occam-pi and CAL Actor Language, and demonstrate the applicability of the approach with application case-studies targeting these different manycore architectures: STHorm, Epiphany and Ambric.For occam-pi, we have extended the Tock compiler and added a backend for STHorm. We evaluate the approach using a fault tolerance model for a four stage 1D-DCT algorithm implemented by using occam-pi’s constructs for dynamic reconfiguration, and the FAST corner detection algorithm which demonstrates the suitability of occam-pi and the compilation framework for data-intensive applications. We also present a new CAL compilation framework which has a front end, two intermediate representations and three backends: for a uniprocessor, Epiphany, and Ambric. We show the feasibility of our approach by compiling a CAL implementation of the 2D-IDCT for the three backends. We also present an evaluation and optimization of code generation for Epiphany by comparing the code generated from CAL with a hand-written C code implementation of 2D-IDCT.
|
|
| 2. |
- Kindberg, Elin, 1981-
(author)
-
Host genetic risk factors to viral diseases - a double-edged sword : Studies of norovirus and tick-borne encephalitis virus
- 2010
-
Doctoral thesis (other academic/artistic)abstract
- It is today well known that the outcome of a certain infection depends on factors of both the host and the pathogen. Studies of host genetic susceptibility to infectious diseases aim to increase the understanding of why some individuals are more susceptible than others, to a certain infection. Knowledge of genetic susceptibility to a viral disease may be used in development of new therapeutic means, and also to recognize individuals who are at increased risk of severe symptoms if infected with a pathogen. It seems however that a risk factor for one disease may play a protective role in another situation; like a double-edged sword.In this thesis I have studied genetic factors affecting susceptibility to norovirus (NoV) and factors affecting the risk of developing tick-borne encephalitis (TBE) after infection with TBE virus (TBEV). NoV is the cause of the “winter vomiting disease”, affecting millions of people every year, and causing up to 200,000 fatalities among children in developing countries, each year. It is today recognized that the secretor status of an individual, i.e. the ability to express ABO blood groups and related antigens, in secretions and on mucosa, affect the risk of being infected by NoV. By studying authentic NoV outbreaks in Denmark, Spain and Sweden and by comparing the secretor status of affected and unaffected individuals we were able to confirm that secretor status have indeed great impact on susceptibility to some NoV strains, but also that there are strains circulating, which infect individuals regardless of secretor status.TBEV is endemic in many parts of Europe and Asia but studies have shown that 70-95% of all infections are asymptomatic or sub-clinical. Some individuals do however develop TBE, a severe disease including meningitis or encephalitis with or without myelitis. Also, many patients suffer from long-time sequelae and TBEV infections may in worst case be fatal. The reason for difference in disease outcome is not known and we have chosen to study if genetic factors affecting the immune response may play a role in disease outcome. To do this we used a prospectively collected Lithuanian material with samples from patients with TBE, AME (aseptic meningoencephalitis) and matched healthy controls. So far we have found that a deletion in chemokine receptor 5 (CCR5), a gene encoding a receptor involved in cell migration, is a risk factor for developing disease. We have also data showing that toll-like receptor 3 (TLR3), a receptor recognizing double stranded RNA (dsRNA), which is a product of TBEV replication, may instead of being protective increase the risk of TBE.
|
|
| 3. |
- Ul-Abdin, Zain, 1975-
(author)
-
Programming of coarse-grained reconfigurable architectures
- 2011
-
Doctoral thesis (other academic/artistic)abstract
- Coarse-grained reconfigurable architectures, which offer massive parallelism coupled with the capability of undergoing run-time reconfiguration, are gaining attention in order to meet not only the increased computational demands of high-performance embedded systems, but also to fulfill the need of adaptability to functional requirements of the application. This thesis focuses on the programming aspects of such coarse-grained reconfigurable computing devices, including the relevant computation models that are capable of exposing different kinds of parallelism inherent in the application and the ability of these models to capture the adaptability requirements of the application. The thesis suggests the occam-pi language for programming of a broad class of coarse-grained reconfigurable architectures as an intermediate language; we call it intermediate, since we believe that the applicationprogramming is best done in a high-level domain-specific language. The salient properties of the occam-pi language are explicit concurrency with built-in mechanisms for interprocessorcommunication, provision for expressing dynamic parallelism, support for the expression of dynamic reconfigurations, and placement attributes. To evaluate the programming approach, a compiler framework was extended to support the language extensions in the occam-pi language, and backends were developed to target two different coarse-grained reconfigurable architectures. XPP and Ambric. The results on XPP reveal that the occam-pi based implementations produce comparable throughput to those of NML programs, while programming at a much higher level of abstraction than that of NML. Similarly the two occam-pi implementations of autofocus criterion calculation targeted to the Ambric platform outperform the CPU implementation by factors of 11-23. Thus, the results of the implemented case-studies suggest that the occam-pi language based approach simplifies the development of applications employing run-time reconfigurable devices without compromising the performance benefits.
|
|
| 4. |
- Gebrewahid, Essayas, 1984-
(author)
-
Tools to Compile Dataflow Programs for Manycores
- 2017
-
Doctoral thesis (other academic/artistic)abstract
- The arrival of manycore systems enforces new approaches for developing applications in order to exploit the available hardware resources. Developing applications for manycores requires programmers to partition the application into subtasks, consider the dependence between the subtasks, understand the underlying hardware and select an appropriate programming model. This is complex, time-consuming and prone to error. In this thesis, we identify and implement abstraction layers in compilation tools to decrease the burden of the programmer, increase program portability and scalability, and increase retargetability of the compilation framework. We present compilation frameworks for two concurrent programming languages, occam-pi and CAL Actor Language, and demonstrate the applicability of the approach with application case-studies targeting these different manycore architectures: STHorm, Epiphany, Ambric, EIT, and ePUMA. For occam-pi, we have extended the Tock compiler and added a backend for STHorm. We evaluate the approach using a fault tolerance model for a four stage 1D-DCT algorithm implemented by using occam-pi's constructs for dynamic reconguration, and the FAST corner detection algorithm which demonstrates the suitability of occam-pi and the compilation framework for data-intensive applications. For CAL, we have developed a new compilation framework, namely Cal2Many. The Cal2Many framework has a front end, two intermediate representations and four backends: for a uniprocessor, Epiphany, Ambric, and a backend for SIMD based architectures. Also, we have identied and implemented of CAL actor fusion and fission methodologies for efficient mapping CAL applications. We have used QRD, FAST corner detection, 2D-IDCT, and MPEG applications to evaluate our compilation process and to analyze the limitations of the hardware.
|
|
| 5. |
- Hoang, Hoai, 1976-
(author)
-
Enhancing the Performance of Distributed Real-time Systems
- 2007
-
Doctoral thesis (other academic/artistic)abstract
- Advanced embedded systems can consist of many sensors, actuators and processors that are deployed on one or several boards, while having a demand of interacting with each other and sharing resources. Communication between different components usually has strict timing constraints. There is thus a strong need to provide solutions for time critical communication. This thesis focuses on both the support of real-time services over standard switched Ethernet networks and the improvement of systems' real-time characteristics, such as reducing delay and jitter in processors and on communication links.Switched Ethernet has been chosen in this work because of its major advantages in industry; it supports higher bit-rates than most other current LAN (Local Area Network) technologies, including field buses, still at a low cost. We propose using a star network topology with a single Ethernet switch. Each node is connected to a separate port of the switch via a full-duplex link, thereby eliminating collisions. A solid real-time communication protocol for switched Ethernet networks is proposed in the thesis, including a real-time layer between the Ethernet layer and the TCP/IP suite. The network has the capability of supporting both real-time and non real-time traffic and assuring adaptation to the surrounding protocol standards.Most embedded systems work in a dynamic environment, where the precise behavior of the network traffic can usually not be predicted. To support real-time services, we have chosen the Earliest Deadline scheduling algorithm (EDF) because of its optimality, high efficiency and suitability for being used in adaptive schemes. To be able to increase the amount of guaranteed real-time traffic, the notion of Asymmetric Deadline Partitioning Scheme (ADPS) is introduced. ADPS allows distribution of the end-to-end deadline of a message, sent from any source node in the network to any destination node via the switch, into two sub-deadlines, one for each hop according to the load of the physical link that it must traverse.For the EDF scheduling algorithm, the feasibility test is one of the most important techniques that provides us with information about whether or not the real-time traffic can be guaranteed by the network. With the same computational complexity as the feasibility test, a method has been developed to compute the minimum EDF-feasible deadline for a real-time task. The importance of this method in real-time applications lies in that it can be effectively used to reduce the response times of specific control activities or limit their input-output jitter. To allow more flexibility in the control of delay and jitter in real-time systems, a general approach for reducing task deadlines according to the requirements of individual tasks has been developed. The method allows the user to specify a deadline reduction factor for each task in order to better exploit the available slack according to the tasks' actual requirements.
|
|
| 6. |
- Ourique de Morais, Wagner, 1979-
(author)
-
Architecting Smart Home Environments for Healthcare : A Database-Centric Approach
- 2015
-
Doctoral thesis (other academic/artistic)abstract
- The development of system architectures and applications for smart homes and ambient assisted living has been the main activity of a number of academic and industrial research projects around the world. Existing system architectures for smart environments usually employ different architectural styles in a multi-layer logical architecture to support the integration and interoperation of heterogeneous hardware and software technologies, which are subsequently used to provide two major functionalities: monitoring and assistance. It is also usual among existing architectures that the database management system is the most common but the least exploited architectural component, existing in the periphery of the system and devoted exclusively for data storage and retrieval. However, database technology has advanced and matured considerably over the years, and, as a result, current database management systems can be and do more.This thesis considers the hypothesis of several features of modern database management systems being employed to address functional (e.g. well-being and security monitoring, automated control, data processing) and non-functional (e.g. interoperability, extensibility, data security and privacy) requirements of smart environments, i.e. the database management system serves as a platform for smart environments. The scope of this thesis is therefore to investigate the possibility of using different features supported by database management systems to create a database-centric system architecture for the development of smart home environments and ambient assisted living. The thesis also investigates the development of applications for health monitoring and assistance: 1) a serious game for fall prevention that assists people in practicing Tai Chi at home, and 2) a non-intrusive home-based method for sleep assessment.These features are explored in this thesis to address general functional aspects of smart environments, such as monitoring, processing, coordination and control of various types of events in a given environment. Extensibility and security features and cross-platform capabilities of database management systems are employed to accommodate non-functional, but still technical, properties of smart environments, including interoperability, extensibility, portability, scalability, security and privacy. Heterogeneous technologies are integrated into the system using programming language and platform independent software resource adapters. Interoperation among integrated technologies is mediated in an active database.The feasibility of the proposed database-centric system architecture was pragmatically investigated with the development of a "smart bedroom'' demonstrator and with the implementation of a number of short-term and long-term types of services to support active aging, aging in place and ambient assisted living. In the proposed architecture, active in-database processing maintains sensitive data within the database. This increases data security and independence from external software applications for data analysis. Changes in the system are managed during runtime, which improves flexibility and avoids system downtime. The proposed system architecture was evaluated taking into account different application scenarios and heterogeneous computing platforms.As a conclusion, modern database management systems support features that can be successfully employed in a database-centric system architecture to effectively and efficiently address functional and non-functional requirements of smart environments.
|
|
| 7. |
- Bengtsson, Jerker
(author)
-
Models and Methods for Development of DSP Applications on Manycore Processors
- 2009
-
Doctoral thesis (other academic/artistic)abstract
- Advanced digital signal processing systems require specialized high-performance embedded computer architectures. The term high-performance translates to large amounts of data and computations per time unit. The term embedded further implies requirements on physical size and power efficiency. Thus the requirements are of both functional and non-functional nature. This thesis addresses the development of high-performance digital signal processing systems relying on manycore technology. We propose building two-level hierarchical computer architectures for this domain of applications. Further, we outline a tool flow based on methods and analysis techniques for automated, multi-objective mapping of such applications on distributed memory manycore processors. In particular, the focus is put on how to provide a means for tunable strategies for mapping of task graphs on array structured distributed memory manycores, with respect to given application constraints. We argue for code mapping strategies based on predicted execution performance, which can be used in an auto-tuning feedback loop or to guide manual tuning directed by the programmer. Automated parallelization, optimisation and mapping to a manycore processor benefits from the use of a concurrent programming model as the starting point. Such a model allows the programmer to express different types and granularities of parallelism as well as computation characteristics of importance in the addressed class of applications. The programming model should also abstract away machine dependent hardware details. The analytical study of WCDMA baseband processing in radio base stations, presented in this thesis, suggests dataflow models as a good match to the characteristics of the application and as execution model abstracting computations on a manycore. Construction of portable tools further requires a manycore machine model and an intermediate representation. The models are needed in order to decouple algorithms, used to transform and map application software, from hardware. We propose a manycore machine model that captures common hardware resources, as well as resource dependent performance metrics for parallel computation and communication. Further, we have developed a multifunctional intermediate representation, which can be used as source for code generation and for dynamic execution analysis. Finally, we demonstrate how we can dynamically analyse execution using abstract interpretation on the intermediate representation. It is shown that the performance predictions can be used to accurately rank different mappings by best throughput or shortest end-to-end computation latency.
|
|
| 8. |
- Bengtsson, Niklas, 1980-
(author)
-
Essays in Development and Labor Economics
- 2009
-
Doctoral thesis (other academic/artistic)abstract
- This thesis consists of three self-contained essays. Essay I. This essay uses time-series of rainfall to estimate the response of body weight to transitory changes in household income in rural Tanzania. The response is positive on average, but is markedly higher among female children. For female children, a ten-percent increase in household income implies an increase in body weight by about 0.4 kilos. In contrast, the body weight of adolescents and young adults is virtually invariant to transitory income changes. Essay II. In this essay, the impact of a village-level assistance program run by the Evangelical Lutheran Church of Tanzania is estimated. We find that the program increased literacy by 15-20 percentage points and educational attainment by 10-15 percentage points, but only among Protestant children. Catholic children living in the same targeted villages were unaffected. The results support the concern that faith organizations might overstate their ability to aid households of different faith. Essay III. This essay exploits a rapid introduction of an unconditional cash grant (child support) in South Africa in order to estimate the marginal propensity to consume and earn out of a permanent change in unearned income. We find that the marginal propensity to earn is about -0.25 to -0.4. A very small fraction of the grant is saved. All in all, the marginal propensities estimated here are all similar to those reported in comparable papers using US data. However, they stand in contrast to some results on conditional cash transfers in other developing countries.
|
|
| 9. |
- Hertz, Erik, 1956-
(author)
-
Methodologies for Approximation of Unary Functions and Their Implementation in Hardware
- 2016
-
Doctoral thesis (other academic/artistic)abstract
- Applications in computer graphics, digital signal processing, communication systems, robotics, astrophysics, fluid physics and many other areas have evolved to become very computation intensive. Algorithms are becoming increasingly complex and require higher accuracy in the computations. In addition, software solutions for these applications are in many cases not sufficient in terms of performance. A hardware implementation is therefore needed. A recurring bottleneck in the algorithms is the performance of the approximations of unary functions, such as trigonometric functions, logarithms and the square root, as well as binary functions such as division. The challenge is therefore to develop a methodology for the implementation of approximations of unary functions in hardware that can cope with the growing requirements. The methodology is required to result in fast execution time, low complexity basic operations that are simple to implement in hardware, and – sincemany applications are battery powered – low power consumption. To ensure appropriate performance of the entire computation in which the approximation is a part, the characteristics and distribution of the approximation error are also things that must be possible to manage. The new approximation methodologies presented in this thesis are of the type that aims to reduce the sizes of the look-up tables by the use of auxiliary functions. They are founded on a synthesis of parabolic functions by multiplication – instead of addition, which is the most common. Three approximation methodologies have been developed; the two last being further developments of the first. For some functions, such as roots, inverse and inverse roots, a straightforward solution with an approximation is not manageable. Since these functions are frequent in many computation intensive algorithms, it is necessary to find very efficient implementations of these functions. New methods for this are also presented in this thesis. They are all founded on working in a floating-point format, and, for the roots functions, a change of number base is also used. The transformations not only enable simpler solutions but also increased accuracy, since the approximation algorithm is performed on a mantissa of limited range. Tools for error analysis have been developed as well. The characteristics and distribution of the approximation error in the new methodologies are presented and compared with existing state-of-the-art methods such as CORDIC. The verification and evaluation of the solutions have to a large extent been made as comparative ASIC implementations with other approximation methods, separately or embedded in algorithms. As an example, an implementation of the logarithm made using the third methodology developed, Harmonized Parabolic Synthesis (HPS), is compared with an implementation using the CORDIC algorithm. Both implementations are designed to provide 15-bit resolution. The design implemented using HPS performs 12 times better than the CORDIC implementation in terms of throughput. In terms of energy consumption, the new methodology consumes 96% less. The chip area is 60% smaller than for the CORDIC algorithm. In summary, the new approximation methodologies presented are found to well meet the demanding requirements that exist in this area.
|
|
| 10. |
- Parsapoor, Mahboobeh, 1978-
(author)
-
Towards Emotion inspired Computational Intelligence (EiCI)
- 2015
-
Doctoral thesis (other academic/artistic)abstract
- One of the main challenges in the computational intelligence (CI) community is to develop nature-inspired algorithms that can efficiently solve real-world problems such as the prediction of space weather phenomena. An early example in this context is taking inspiration from the biological neurons in the mammal’s nervous system and developing an artificial neuron. This work laid the foundation for artificial neural networks (ANNs) that aim to mimic the connections between neurons in the mammal’s nervous system and to develop an artificial model of the brain. ANNs are well-known CI models that have shown high generalization capability when solving real-world problems, e.g., chaotic time-series prediction problems. However, ANNs mostly tend to suffer from long computation time and high model complexity. This thesis presents a new category of CI paradigms by taking inspiration from emotions, and these CI models are referred to as emotion-inspired computational intelligence models (EiCIs). In the thesis, I have outlined the preliminary steps that have been taken to develop EiCIs. These steps include studying different emotional theories and hypotheses, designing and implementing CI models for two specific applications in artificial intelligence (prediction and optimization), evaluating the performance of the new CI models, and comparing the obtained results with the results of well-known CI models (e.g., ANNs) and discussing the potential improvement that can be achieved. The first step, and a significant contribution of this thesis, is to review the various definitions of emotions and to investigate which emotional theories that are the most relevant for developing a CI model. Amongst different theories and hypotheses of emotions, the fear conditioning hypothesis as well as affect theory have been two main sources of inspiration in the development of the EiCIs proposed in this thesis. The fear conditioning hypothesis that was first proposed by LeDoux reveals some important characteristics of the underlying neural structure of fear conditioning behavior in biological systems. Based on the features of such networks, it could be an applicable hypothesis to be the basis of the development of a subgroup of EiCIs that could be used for prediction applications, e.g. BELIMs (Brain Emotional Learning Inspired Models), and as emotion-inspired engines for decision-making applications.The second emotional theory of the thesis is the affect theory (which was first suggested by Silvan Tomkins) that describes what the basic emotions are and how they can be associated with facial expressions. A mechanism to express the basic emotional feelings is also useful in designing another category of EiCIs that are referred to as emotion-inspired optimization methods. The fundamental hypotheses of the thesis, have led to developing EiCIs, can be presented as follows. The first hypothesis is that the neural structure of fear conditioning can be considered to be a nature-based system with the capability to show intelligent behavior through its functionality. This hypothesis is stated on the basis of the three main characteristics of the neural structure of fear conditioning behavior.The first characteristic is that the amygdala is the main center for processing fear-induced stimuli and that it provides the fear reaction through its interaction with other regions of the brain such as the sensory cortex, the thalamus, and the hippocampus. The second characteristic is that the procedure of processing of fearful stimuli and the provision of emotional reactions is simple and quick. The third aspect is that the amygdala not only provides fear responses but also learns to predict aversive events by interacting with other regions of the brain, which means that an intelligent behavior emerges.The second hypothesis is that the system in which the three monoamines neurotransmitters serotonin, dopamine, and noradrenalin and thus produces emotional behaviors, can be viewed as a biological system associated with the emergence of intelligent behavior.The above hypotheses state that a suitable way to develop new CI models is to take inspiration from the neural structure of fear conditioning and the natural system of three monoamine neurotransmitters. A significant contribution of this thesis is the evaluation of the ability of EiCIs by examining them to solve real-world problems such as the prediction of space weather phenomena (e.g., predicting real time-series such as sunspot number, auroral electrojet index, and disturbance time index) and the optimization of some central procedures in network communications. These evaluations have led to that comparable results have been obtained, which in turn supports the conclusion that EiCIs have acceptable and reasonable performance regarding computation time and model complexity. However, to achieve the final goal of the research study (i.e., to develop a CI model with low computation time and low model complexity), some enhancements of EiCIs are necessary. Moreover, new designs and implementations of these models can be developed by taking inspiration from other theories.
|
|
