| 1. |
- Aguilar, Xavier
(författare)
-
Performance Monitoring, Analysis, and Real-Time Introspection on Large-Scale Parallel Systems
- 2020
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- High-Performance Computing (HPC) has become an important scientific driver. A wide variety of research ranging for example from drug design to climate modelling is nowadays performed in HPC systems. Furthermore, the tremendous computer power of such HPC systems allows scientists to simulate problems that were unimaginable a few years ago. However, the continuous increase in size and complexity of HPC systems is turning the development of efficient parallel software into a difficult task. Therefore, the use of per- formance monitoring and analysis is a must in order to unveil inefficiencies in parallel software. Nevertheless, performance tools also face challenges as a result of the size of HPC systems, for example, coping with huge amounts of performance data generated.In this thesis, we propose a new model for performance characterisation of MPI applications that tackles the challenge of big performance data sets. Our approach uses Event Flow Graphs to balance the scalability of profiling techniques (generating performance reports with aggregated metrics) with the richness of information of tracing methods (generating files with sequences of time-stamped events). In other words, graphs allow to encode ordered se- quences of events without storing the whole sequence of such events, and therefore, they need much less memory and disk space, and are more scal- able. We demonstrate in this thesis how our Event Flow Graph model can be used as a trace compression method. Furthermore, we propose a method to automatically detect the structure of MPI applications using our Event Flow Graphs. This knowledge can afterwards be used to collect performance data in a smarter way, reducing for example the amount of redundant data collected. Finally, we demonstrate that our graphs can be used beyond trace compression and automatic analysis of performance data. We propose a new methodology to use Event Flow Graphs in the task of visual performance data exploration.In addition to the Event Flow Graph model, we also explore in this thesis the design and use of performance data introspection frameworks. Future HPC systems will be very dynamic environments providing extreme levels of parallelism, but with energy constraints, considerable resource sharing, and heterogeneous hardware. Thus, the use of real-time performance data to or- chestrate program execution in such a complex and dynamic environment will be a necessity. This thesis presents two different performance data introspec- tion frameworks that we have implemented. These introspection frameworks are easy to use, and provide performance data in real time with very low overhead. We demonstrate, among other things, how our approach can be used to reduce in real time the energy consumed by the system.The approaches proposed in this thesis have been validated in different HPC systems using multiple scientific kernels as well as real scientific applica- tions. The experiments show that our approaches in performance character- isation and performance data introspection are not intrusive at all, and can be a valuable contribution to help in the performance monitoring of future HPC systems.
|
|
| 2. |
- Aguilar, Xavier
(författare)
-
Towards Scalable Performance Analysis of MPI Parallel Applications
- 2015
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- A considerably fraction of science discovery is nowadays relying on computer simulations. High Performance Computing (HPC) provides scientists with the means to simulate processes ranging from climate modeling to protein folding. However, achieving good application performance and making an optimal use of HPC resources is a heroic task due to the complexity of parallel software. Therefore, performance tools and runtime systems that help users to execute applications in the most optimal way are of utmost importance in the landscape of HPC. In this thesis, we explore different techniques to tackle the challenges of collecting, storing, and using fine-grained performance data. First, we investigate the automatic use of real-time performance data in order to run applications in an optimal way. To that end, we present a prototype of an adaptive task-based runtime system that uses real-time performance data for task scheduling. This runtime system has a performance monitoring component that provides real-time access to the performance behavior of anapplication while it runs. The implementation of this monitoring component is presented and evaluated within this thesis. Secondly, we explore lossless compression approaches for MPI monitoring. One of the main problems that performance tools face is the huge amount of fine-grained data that can be generated from an instrumented application. Collecting fine-grained data from a program is the best method to uncover the root causes of performance bottlenecks, however, it is unfeasible with extremely parallel applications or applications with long execution times. On the other hand, collecting coarse-grained data is scalable but sometimes not enough to discern the root cause of a performance problem. Thus, we propose a new method for performance monitoring of MPI programs using event flow graphs. Event flow graphs provide very low overhead in terms of execution time and storage size, and can be used to reconstruct fine-grained trace files of application events ordered in time.
|
|
| 3. |
- Ahmed, Laeeq
(författare)
-
Scalable Analysis of Large Datasets in Life Sciences
- 2019
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- We are experiencing a deluge of data in all fields of scientific and business research, particularly in the life sciences, due to the development of better instrumentation and the rapid advancements that have occurred in information technology in recent times. There are major challenges when it comes to handling such large amounts of data. These range from the practicalities of managing these large volumes of data, to understanding the meaning and practical implications of the data.In this thesis, I present parallel methods to efficiently manage, process, analyse and visualize large sets of data from several life sciences fields at a rapid rate, while building and utilizing various machine learning techniques in a novel way. Most of the work is centred on applying the latest Big Data Analytics frameworks for creating efficient virtual screening strategies while working with large datasets. Virtual screening is a method in cheminformatics used for Drug discovery by searching large libraries of molecule structures. I also present a method for the analysis of large Electroencephalography data in real time. Electroencephalography is one of the main techniques used to measure the brain electrical activity.First, I evaluate the suitability of Spark, a parallel framework for large datasets, for performing parallel ligand-based virtual screening. As a case study, I classify molecular library using prebuilt classification models to filter out the active molecules. I also demonstrate a strategy to create cloud-ready pipelines for structure-based virtual screening. The major advantages of this strategy are increased productivity and high throughput. In this work, I show that Spark can be applied to virtual screening, and that it is, in general, an appropriate solution for large-scale parallel pipelining. Moreover, I illustrate how Big Data analytics are valuable in working with life sciences datasets.Secondly, I present a method to further reduce the overall time of the structured-based virtual screening strategy using machine learning and a conformal-prediction-based iterative modelling strategy. The idea is to only dock those molecules that have a better than average chance of being an inhibitor when searching for molecules that could potentially be used as drugs. Using machine learning models from this work, I built a web service to predict the target profile of multiple compounds against ready-made models for a list of targets where 3D structures are available. These target predictions can be used to understand off-target effects, for example in the early stages of drug discovery projects.Thirdly, I present a method to detect seizures in long term Electroencephalography readings - this method works in real time taking the ongoing readings in as live data streams. The method involves tackling the challenges of real-time decision-making, storing large datasets in memory and updating the prediction model with newly produced data at a rapid rate. The resulting algorithm not only classifies seizures in real time, it also learns the threshold in real time. I also present a new feature "top-k amplitude measure" for classifying which parts of the data correspond to seizures. Furthermore, this feature helps to reduce the amount of data that needs to be processed in the subsequent steps.
|
|
| 4. |
- Gebremedhin, Mahder, 1985-
(författare)
-
Automatic and Explicit Parallelization Approaches for Equation Based Mathematical Modeling and Simulation
- 2018
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The move from single-core processor systems to multi-core and manyprocessor systems comes with the requirement of implementing computations in a way that can utilize these multiple computational units efficiently. This task of writing efficient parallel algorithms will not be possible without improving programming languages and compilers to provide the supporting mechanisms. Computer aided mathematical modelling and simulation is one of the most computationally intensive areas of computer science. Even simplified models of physical systems can impose a considerable computational load on the processors at hand. Being able to take advantage of the potential computational power provided by multi-core systems is vital in this area of application. This thesis tries to address how to take advantage of the potential computational power provided by these modern processors in order to improve the performance of simulations, especially for models in the Modelica modelling language compiled and simulated using the OpenModelica compiler and run-time environment.Two approaches of utilizing the computational power provided by modern multi-core architectures for simulation of Mathematical models are presented in this thesis: Automatic and Explicit parallelization respectively. The Automatic approach presents the process of extracting and utilizing potential parallelism from equation systems in an automatic way without any need for extra effort from the modellers/programmers. This thesis explains new and improved methods together with improvements made to the OpenModelica compiler and a new accompanying task systems library for efficient representation, clustering, scheduling, profiling, and executing complex equation/ task systems with heavy dependencies. The Explicit parallelization approach allows utilizing parallelism with the help of the modeller or programmer. New programming constructs have been introduced to the Modelica language in order to enable modellers to express parallelized algorithms to take advantage of the computational capabilities provided by modern multicore CPUs and GPUs. The OpenModelica compiler has been improved accordingly to recognize and utilize the information from these new algorithmic constructs and to generate parallel code for enhanced computational performance, portable to a range of parallel architectures through the OpenCL standard.
|
|
| 5. |
- Gholami, Ali, 1978-
(författare)
-
Security and Privacy of Sensitive Data in Cloud Computing
- 2016
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Cloud computing offers the prospect of on-demand, elastic computing, provided as a utility service, and it is revolutionizing many domains of computing. Compared with earlier methods of processing data, cloud computing environments provide significant benefits, such as the availability of automated tools to assemble, connect, configure and reconfigure virtualized resources on demand. These make it much easier to meet organizational goals as organizations can easily deploy cloud services. However, the shift in paradigm that accompanies the adoption of cloud computing is increasingly giving rise to security and privacy considerations relating to facets of cloud computing such as multi-tenancy, trust, loss of control and accountability. Consequently, cloud platforms that handle sensitive information are required to deploy technical measures and organizational safeguards to avoid data protection breakdowns that might result in enormous and costly damages. Sensitive information in the context of cloud computing encompasses data from a wide range of different areas and domains. Data concerning health is a typical example of the type of sensitive information handled in cloud computing environments, and it is obvious that most individuals will want information related to their health to be secure. Hence, with the growth of cloud computing in recent times, privacy and data protection requirements have been evolving to protect individuals against surveillance and data disclosure. Some examples of such protective legislation are the EU Data Protection Directive (DPD) and the US Health Insurance Portability and Accountability Act (HIPAA), both of which demand privacy preservation for handling personally identifiable information. There have been great efforts to employ a wide range of mechanisms to enhance the privacy of data and to make cloud platforms more secure. Techniques that have been used include: encryption, trusted platform module, secure multi-party computing, homomorphic encryption, anonymization, container and sandboxing technologies. However, it is still an open problem about how to correctly build usable privacy-preserving cloud systems to handle sensitive data securely due to two research challenges. First, existing privacy and data protection legislation demand strong security, transparency and audibility of data usage. Second, lack of familiarity with a broad range of emerging or existing security solutions to build efficient cloud systems. This dissertation focuses on the design and development of several systems and methodologies for handling sensitive data appropriately in cloud computing environments. The key idea behind the proposed solutions is enforcing the privacy requirements mandated by existing legislation that aims to protect the privacy of individuals in cloud-computing platforms. We begin with an overview of the main concepts from cloud computing, followed by identifying the problems that need to be solved for secure data management in cloud environments. It then continues with a description of background material in addition to reviewing existing security and privacy solutions that are being used in the area of cloud computing. Our first main contribution is a new method for modeling threats to privacy in cloud environments which can be used to identify privacy requirements in accordance with data protection legislation. This method is then used to propose a framework that meets the privacy requirements for handling data in the area of genomics. That is, health data concerning the genome (DNA) of individuals. Our second contribution is a system for preserving privacy when publishing sample availability data. This system is noteworthy because it is capable of cross-linking over multiple datasets. The thesis continues by proposing a system called ScaBIA for privacy-preserving brain image analysis in the cloud. The final section of the dissertation describes a new approach for quantifying and minimizing the risk of operating system kernel exploitation, in addition to the development of a system call interposition reference monitor for Lind - a dual sandbox.
|
|
| 6. |
- Peng, Ivy Bo
(författare)
-
Data Movement on Emerging Large-Scale Parallel Systems
- 2017
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Large-scale HPC systems are an important driver for solving computational problems in scientific communities. Next-generation HPC systems will not only grow in scale but also in heterogeneity. This increased system complexity entails more challenges to data movement in HPC applications. Data movement on emerging HPC systems requires asynchronous fine-grained communication and efficient data placement in the main memory. This thesis proposes an innovative programming model and algorithm to prepare HPC applications for the next computing era: (1) a data streaming model that supports emerging data-intensive applications on supercomputers, (2) a decoupling model that improves parallelism and mitigates the impact of imbalance in applications, (3) a new framework and methodology for predicting the impact of largescale heterogeneous memory systems on HPC applications, and (4) a data placement algorithm that uses a set of rules and a decision tree to determine the data-to-memory mapping in heterogeneous main memory.The proposed approaches in this thesis are evaluated on multiple supercomputers with different processors and interconnect networks. The evaluation uses a diverse set of applications that represent conventional scientific applications and emerging data-analytic workloads on HPC systems. The experimental results on the petascale testbed show that the approaches obtain increasing performance improvements as system scale increases and this trend supports the approaches as a valuable contribution towards future HPC systems.
|
|
| 7. |
- Stavåker, Kristian, 1983-
(författare)
-
Contributions to Parallel Simulation of Equation-Based Models on Graphics Processing Units
- 2011
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- In this thesis we investigate techniques and methods for parallel simulation of equation-based, object-oriented (EOO) Modelica models on graphics processing units (GPUs). Modelica is being developed through an international effort via the Modelica Association. With Modelica it is possible to build computationally heavy models; simulating such models however might take a considerable amount of time. Therefor techniques of utilizing parallel multi-core architectures for simulation are desirable. The goal in this work is mainly automatic parallelization of equation-based models, that is, it is up to the compiler and not the end-user modeler to make sure that code is generated that can efficiently utilize parallel multi-core architectures. Not only the code generation process has to be altered but the accompanying run-time system has to be modified as well. Adding explicit parallel language constructs to Modelica is also discussed to some extent. GPUs can be used to do general purpose scientific and engineering computing. The theoretical processing power of GPUs has surpassed that of CPUs due to the highly parallel structure of GPUs. GPUs are, however, only good at solving certain problems of data-parallel nature. In this thesis we relate several contributions, by the author and co-workers, to each other. We conclude that the massively parallel GPU architectures are currently only suitable for a limited set of Modelica models. This might change with future GPU generations. CUDA for instance, the main software platform used in the thesis for general purpose computing on graphics processing units (GPGPU), is changing rapidly and more features are being added such as recursion, function pointers, C++ templates, etc.; however the underlying hardware architecture is still optimized for data-parallelism.
|
|
| 8. |
- Stavåker, Kristian, 1983-
(författare)
-
Contributions to Simulation of Modelica Models on Data-Parallel Multi-Core Architectures
- 2015
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Modelica is an object-oriented, equation-based modeling and simulation language being developed through an international effort by the Modelica Association. With Modelica it is possible to build computationally demanding models; however, simulating such models might take a considerable amount of time. Therefore techniques of utilizing parallel multi-core architectures for faster simulations are desirable. In this thesis the topic of simulation of Modelica on parallel architectures in general and on graphics processing units (GPUs) in particular is explored. GPUs support code that can be executed in a data-parallel fashion. It is also possible to connect and run several GPUs together which opens opportunities for even more parallelism. In this thesis several approaches regarding simulation of Modelica models on GPUs and multi-core architectures are explored.In this thesis the topic of expressing and solving partial differential equations (PDEs) in the context of Modelica is also explored, since such models usually give rise to equation systems with a regular structure, which can be suitable for efficient solution on GPUs. Constructs for PDE-based modeling are currently not part of the standard Modelica language specification. Several approaches on modeling and simulation with PDEs in the context of Modelica have been developed over the years. In this thesis we present selected earlier work, ongoing work and planned work on PDEs in the context of Modelica. Some approaches detailed in this thesis are: extending the language specification with PDE handling; using a software with support for PDEs and automatic discretization of PDEs; and connecting an external C++ PDE library via the functional mockup interface (FMI).Finally the topic of parallel skeletons in the context of Modelica is explored. A skeleton is a predefined, generic component that implements a common specific pattern of computation and data dependence. Skeletons provide a high degree of abstraction and portability and a skeleton can be customized with user code. Using skeletons with Modelica opens up the possibility of executing heavy Modelica-based matrix and vector computations on multi-core architectures. A working Modelica-SkePU library with some minor necessary compiler extensions is presented.
|
|
| 9. |
- Espling, Daniel, 1983-
(författare)
-
Metadata Management in Multi-Grids and Multi-Clouds
- 2011
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Grid computing and cloud computing are two related paradigms used to access and use vast amounts of computational resources. The resources are often owned and managed by a third party, relieving the users from the costs and burdens of acquiring and managing a considerably large infrastructure themselves. Commonly, the resources are either contributed by different stakeholders participating in shared projects (grids), or owned and managed by a single entity and made available to its users with charging based on actual resource consumption (clouds). Individual grid or cloud sites can form collaborations with other sites, giving each site access to more resources that can be used to execute tasks submitted by users. There are several different models of collaborations between sites, each suitable for different scenarios and each posing additional requirements on the underlying technologies.Metadata concerning the status and resource consumption of tasks are created during the execution of the task on the infrastructure. This metadata is used as the primary input in many core management processes, e.g., as a base for accounting and billing, as input when prioritizing and placing incoming task, and as a base for managing the amount of resources allocated to different tasks.Focusing on management and utilization of metadata, this thesis contributes to a better understanding of the requirements and challenges imposed by different collaboration models in both grids and clouds. The underlying design criteria and resulting architectures of several software systems are presented in detail. Each system addresses different challenges imposed by cross-site grid and cloud architectures:The LUTSfed approach provides a lean and optional mechanism for filtering and management of usage data between grid or cloud sites.An accounting and billing system natively designed to support cross-site clouds demonstrates usage data management despite unknown placement and dynamic task resource allocation.The FSGrid system enables fairshare job prioritization across different grid sites, mitigating the problems of heterogeneous scheduling software and local management policies.The results and experiences from these systems are both theoretical and practical, as full scale implementations of each system has been developed and analyzed as a part of this work. Early theoretical work on structure-based service management forms a foundation for future work on structured-aware service placement in cross- site clouds.
|
|
| 10. |
- Jayawardena, Mahen, 1977-
(författare)
-
An e-Science Approach to Genetic Analysis of Quantitative Traits
- 2010
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Many important traits in plants, animals and humans are quantitative, and most such traits are generally believed to be affected by multiple genetic loci. Standard computational tools for mapping of quantitative traits (i.e. for finding Quantitative Trait Loci, QTL, in the genome) use linear regression models for relating the observed phenotypes to the genetic composition of individuals in an experimental population. Using these tools to simultaneously search for multiple QTL is computationally demanding. The main reason for this is the complex optimization landscape for the multidimensional global optimization problems that must be solved. This thesis describes parallel algorithms, implementations and tools for simultaneous mapping of several QTL. These new computational tools enable genetic analysis exploiting new classes of multidimensional statistical models, potentially resulting in interesting results in genetics. We first describe how the standard, brute-force algorithm for global optimization in QTL analysis is parallelized and implemented on a grid system. Then, we also present a parallelized version of the more elaborate global optimization algorithm DIRECT and show how this can be efficiently deployed and used on grid systems and other loosely-coupled architectures. The parallel DIRECT scheme is further developed to exploit both coarse-grained parallelism in grid systems or clusters as well as fine-grained, tightly-coupled parallelism in multi-core nodes. The results show that excellent speedup and performance can be archived on grid systems and clusters, even when using a tightly-coupled algorithm such as DIRECT. Finally, we provide two distinctly different front-ends for our code. One is a grid portal providing a graphical front-end suitable for novice users and standard forms of QTL analysis. The other is a prototype of an R-based grid-enabled problem solving environment. Both of these front-ends can, after some further refinement, be utilized by geneticists for performing multidimensional genetic analysis of quantitative traits on a regular basis.
|
|
