Image Re-Ranking Acceleration on GPUs

Huge image collections are becoming available lately. In this scenario, the use of Content-Based Image Retrieval (CBIR) systems has emerged as a promising approach to support image searches. The objective of CBIR systems is to retrieve the most similar images in a collection, given a query image, by taking into account image visual properties such as texture, color, and shape. In these systems, the effectiveness of the retrieval process depends heavily on the accuracy of ranking approaches. Recently, re-ranking approaches have been proposed to improve the effectiveness of CBIR systems by taking into account the relationships among images. The re-ranking approaches consider the relationships among all images in a given dataset. These approaches typically demands a huge amount of computational power, which hampers its use in practical situations. On the other hand, these methods can be massively parallelized. In this paper, we propose to speedup the computation of the RL-Sim algorithm, a recently proposed image re-ranking approach, by using the computational power of Graphics Processing Units (GPU). GPUs are emerging as relatively inexpensive parallel processors that are becoming available on a wide range of computer systems. We address the image re-ranking performance challenges by proposing a parallel solution designed to fit the computational model of GPUs. We conducted an experimental evaluation considering different implementations and devices. Experimental results demonstrate that significant performance gains can be obtained. Our approach achieves speedups of 7x from serial implementation considering the overall algorithm and up to 36x on its core steps.

Predição de sítios de ligação para a cisplatina e transplatina baseada em ligações de hidrogênio

Pós-graduação em Biofísica Molecular - IBILCE

Heterogeneous Multi-core Architectures for High Performance Computing

This thesis deals with heterogeneous architectures in standard workstations. Heterogeneous architectures represent an appealing alternative to traditional supercomputers because they are based on commodity components fabricated in large quantities. Hence their price-performance ratio is unparalleled in the world of high performance computing (HPC). In particular, different aspects related to the performance and consumption of heterogeneous architectures have been explored. The thesis initially focuses on an efficient implementation of a parallel application, where the execution time is dominated by an high number of floating point instructions. Then the thesis touches the central problem of efficient management of power peaks in heterogeneous computing systems. Finally it discusses a memory-bounded problem, where the execution time is dominated by the memory latency. Specifically, the following main contributions have been carried out: A novel framework for the design and analysis of solar field for Central Receiver Systems (CRS) has been developed. The implementation based on desktop workstation equipped with multiple Graphics Processing Units (GPUs) is motivated by the need to have an accurate and fast simulation environment for studying mirror imperfection and non-planar geometries. Secondly, a power-aware scheduling algorithm on heterogeneous CPU-GPU architectures, based on an efficient distribution of the computing workload to the resources, has been realized. The scheduler manages the resources of several computing nodes with a view to reducing the peak power. The two main contributions of this work follow: the approach reduces the supply cost due to high peak power whilst having negligible impact on the parallelism of computational nodes. from another point of view the developed model allows designer to increase the number of cores without increasing the capacity of the power supply unit. Finally, an implementation for efficient graph exploration on reconfigurable architectures is presented. The purpose is to accelerate graph exploration, reducing the number of random memory accesses.

Numerical and Analytical Methods for Laser-Plasma Acceleration Physics

Theories and numerical modeling are fundamental tools for understanding, optimizing and designing present and future laser-plasma accelerators (LPAs). Laser evolution and plasma wave excitation in a LPA driven by a weakly relativistically intense, short-pulse laser propagating in a preformed parabolic plasma channel, is studied analytically in 3D including the effects of pulse steepening and energy depletion. At higher laser intensities, the process of electron self-injection in the nonlinear bubble wake regime is studied by means of fully self-consistent Particle-in-Cell simulations. Considering a non-evolving laser driver propagating with a prescribed velocity, the geometrical properties of the non-evolving bubble wake are studied. For a range of parameters of interest for laser plasma acceleration, The dependence of the threshold for self-injection in the non-evolving wake on laser intensity and wake velocity is characterized. Due to the nonlinear and complex nature of the Physics involved, computationally challenging numerical simulations are required to model laser-plasma accelerators operating at relativistic laser intensities. The numerical and computational optimizations, that combined in the codes INF&RNO and INF&RNO/quasi-static give the possibility to accurately model multi-GeV laser wakefield acceleration stages with present supercomputing architectures, are discussed. The PIC code jasmine, capable of efficiently running laser-plasma simulations on Graphics Processing Units (GPUs) clusters, is presented. GPUs deliver exceptional performance to PIC codes, but the core algorithms had to be redesigned for satisfying the constraints imposed by the intrinsic parallelism of the architecture. The simulation campaigns, run with the code jasmine for modeling the recent LPA experiments with the INFN-FLAME and CNR-ILIL laser systems, are also presented.

Many-Core Architectures: Hardware-Software Optimization and Modeling Techniques

During the last few decades an unprecedented technological growth has been at the center of the embedded systems design paramount, with Moore’s Law being the leading factor of this trend. Today in fact an ever increasing number of cores can be integrated on the same die, marking the transition from state-of-the-art multi-core chips to the new many-core design paradigm. Despite the extraordinarily high computing power, the complexity of many-core chips opens the door to several challenges. As a result of the increased silicon density of modern Systems-on-a-Chip (SoC), the design space exploration needed to find the best design has exploded and hardware designers are in fact facing the problem of a huge design space. Virtual Platforms have always been used to enable hardware-software co-design, but today they are facing with the huge complexity of both hardware and software systems. In this thesis two different research works on Virtual Platforms are presented: the first one is intended for the hardware developer, to easily allow complex cycle accurate simulations of many-core SoCs. The second work exploits the parallel computing power of off-the-shelf General Purpose Graphics Processing Units (GPGPUs), with the goal of an increased simulation speed. The term Virtualization can be used in the context of many-core systems not only to refer to the aforementioned hardware emulation tools (Virtual Platforms), but also for two other main purposes: 1) to help the programmer to achieve the maximum possible performance of an application, by hiding the complexity of the underlying hardware. 2) to efficiently exploit the high parallel hardware of many-core chips in environments with multiple active Virtual Machines. This thesis is focused on virtualization techniques with the goal to mitigate, and overtake when possible, some of the challenges introduced by the many-core design paradigm.

Particle methods parallel implementations by GP-GPU strategies

This paper outlines the problems found in the parallelization of SPH (Smoothed Particle Hydrodynamics) algorithms using Graphics Processing Units. Different results of some parallel GPU implementations in terms of the speed-up and the scalability compared to the CPU sequential codes are shown. The most problematic stage in the GPU-SPH algorithms is the one responsible for locating neighboring particles and building the vectors where this information is stored, since these specific algorithms raise many dificulties for a data-level parallelization. Because of the fact that the neighbor location using linked lists does not show enough data-level parallelism, two new approaches have been pro- posed to minimize bank conflicts in the writing and subsequent reading of the neighbor lists. The first strategy proposes an efficient coordination between CPU-GPU, using GPU algorithms for those stages that allow a straight forward parallelization, and sequential CPU algorithms for those instructions that involve some kind of vector reduction. This coordination provides a relatively orderly reading of the neighbor lists in the interactions stage, achieving a speed-up factor of x47 in this stage. However, since the construction of the neighbor lists is quite expensive, it is achieved an overall speed-up of x41. The second strategy seeks to maximize the use of the GPU in the neighbor's location process by executing a specific vector sorting algorithm that allows some data-level parallelism. Al- though this strategy has succeeded in improving the speed-up on the stage of neighboring location, the global speed-up on the interactions stage falls, due to inefficient reading of the neighbor vectors. Some changes to these strategies are proposed, aimed at maximizing the computational load of the GPU and using the GPU texture-units, in order to reach the maximum speed-up for such codes. Different practical applications have been added to the mentioned GPU codes. First, the classical dam-break problem is studied. Second, the wave impact of the sloshing fluid contained in LNG vessel tanks is also simulated as a practical example of particle methods

Optimal polygonal L1 linearization and fast interpolation of nonlinear systems

The analysis of complex nonlinear systems is often carried out using simpler piecewise linear representations of them. A principled and practical technique is proposed to linearize and evaluate arbitrary continuous nonlinear functions using polygonal (continuous piecewise linear) models under the L1 norm. A thorough error analysis is developed to guide an optimal design of two kinds of polygonal approximations in the asymptotic case of a large budget of evaluation subintervals N. The method allows the user to obtain the level of linearization (N) for a target approximation error and vice versa. It is suitable for, but not limited to, an efficient implementation in modern Graphics Processing Units (GPUs), allowing real-time performance of computationally demanding applications. The quality and efficiency of the technique has been measured in detail on two nonlinear functions that are widely used in many areas of scientific computing and are expensive to evaluate.

Large scale multiphysics modeling of neurites

Situado en el límite entre Ingeniería, Informática y Biología, la mecánica computacional de las neuronas aparece como un nuevo campo interdisciplinar que potencialmente puede ser capaz de abordar problemas clínicos desde una perspectiva diferente. Este campo es multiescala por naturaleza, yendo desde la nanoescala (como, por ejemplo, los dímeros de tubulina) a la macroescala (como, por ejemplo, el tejido cerebral), y tiene como objetivo abordar problemas que son complejos, y algunas veces imposibles, de estudiar con medios experimentales. La modelización computacional ha sido ampliamente empleada en aplicaciones Neurocientíficas tan diversas como el crecimiento neuronal o la propagación de los potenciales de acción compuestos. Sin embargo, en la mayoría de los enfoques de modelización hechos hasta ahora, la interacción entre la célula y el medio/estímulo que la rodea ha sido muy poco explorada. A pesar de la tremenda importancia de esa relación en algunos desafíos médicos—como, por ejemplo, lesiones traumáticas en el cerebro, cáncer, la enfermedad del Alzheimer—un puente que relacione las propiedades electrofisiológicas-químicas y mecánicas desde la escala molecular al nivel celular todavía no existe. Con ese objetivo, esta investigación propone un marco computacional multiescala particularizado para dos escenarios respresentativos: el crecimiento del axón y el acomplamiento electrofisiológicomecánico de las neuritas. En el primer caso, se explora la relación entre los constituyentes moleculares del axón durante su crecimiento y sus propiedades mecánicas resultantes, mientras que en el último, un estímulo mecánico provoca deficiencias funcionales a nivel celular como consecuencia de sus alteraciones electrofisiológicas-químicas. La modelización computacional empleada en este trabajo es el método de las diferencias finitas, y es implementada en un nuevo programa llamado Neurite. Aunque el método de los elementos finitos es también explorado en parte de esta investigación, el método de las diferencias finitas tiene la flexibilidad y versatilidad necesaria para implementar mode los biológicos, así como la simplicidad matemática para extenderlos a simulaciones a gran escala con un coste computacional bajo. Centrándose primero en el efecto de las propiedades electrofisiológicas-químicas sobre las propiedades mecánicas, una versión adaptada de Neurite es desarrollada para simular la polimerización de los microtúbulos en el crecimiento del axón y proporcionar las propiedades mecánicas como función de la ocupación de los microtúbulos. Después de calibrar el modelo de crecimiento del axón frente a resultados experimentales disponibles en la literatura, las características mecánicas pueden ser evaluadas durante la simulación. Las propiedades mecánicas del axón muestran variaciones dramáticas en la punta de éste, donde el cono de crecimiento soporta las señales químicas y mecánicas. Bansándose en el conocimiento ganado con el modelo de diferencias finitas, y con el objetivo de ir de 1D a 3D, este esquema preliminar pero de una naturaleza innovadora allana el camino a futuros estudios con el método de los elementos finitos. Centrándose finalmente en el efecto de las propiedades mecánicas sobre las propiedades electrofisiológicas- químicas, Neurite es empleado para relacionar las cargas mecánicas macroscópicas con las deformaciones y velocidades de deformación a escala microscópica, y simular la propagación de la señal eléctrica en las neuritas bajo carga mecánica. Las simulaciones fueron calibradas con resultados experimentales publicados en la literatura, proporcionando, por tanto, un modelo capaz de predecir las alteraciones de las funciones electrofisiológicas neuronales bajo cargas externas dañinas, y uniendo lesiones mecánicas con las correspondientes deficiencias funcionales. Para abordar simulaciones a gran escala, aunque otras arquitecturas avanzadas basadas en muchos núcleos integrados (MICs) fueron consideradas, los solvers explícito e implícito se implementaron en unidades de procesamiento central (CPU) y unidades de procesamiento gráfico (GPUs). Estudios de escalabilidad fueron llevados acabo para ambas implementaciones mostrando resultados prometedores para casos de simulaciones extremadamente grandes con GPUs. Esta tesis abre la vía para futuros modelos mecánicos con el objetivo de unir las propiedades electrofisiológicas-químicas con las propiedades mecánicas. El objetivo general es mejorar el conocimiento de las comunidades médicas y de bioingeniería sobre la mecánica de las neuronas y las deficiencias funcionales que aparecen de los daños producidos por traumatismos mecánicos, como lesiones traumáticas en el cerebro, o enfermedades neurodegenerativas como la enfermedad del Alzheimer. ABSTRACT Sitting at the interface between Engineering, Computer Science and Biology, Computational Neuron Mechanics appears as a new interdisciplinary field potentially able to tackle clinical problems from a new perspective. This field is multiscale by nature, ranging from the nanoscale (e.g., tubulin dimers) to the macroscale (e.g., brain tissue), and aims at tackling problems that are complex, and sometime impossible, to study through experimental means. Computational modeling has been widely used in different Neuroscience applications as diverse as neuronal growth or compound action potential propagation. However, in the majority of the modeling approaches done in this field to date, the interactions between the cell and its surrounding media/stimulus have been rarely explored. Despite of the tremendous importance of such relationship in several medical challenges—e.g., traumatic brain injury (TBI), cancer, Alzheimer’s disease (AD)—a bridge between electrophysiological-chemical and mechanical properties of neurons from the molecular scale to the cell level is still lacking. To this end, this research proposes a multiscale computational framework particularized for two representative scenarios: axon growth and electrophysiological-mechanical coupling of neurites. In the former case, the relation between the molecular constituents of the axon during its growth and its resulting mechanical properties is explored, whereas in the latter, a mechanical stimulus provokes functional deficits at cell level as a consequence of its electrophysiological-chemical alterations. The computational modeling approach chosen in this work is the finite difference method (FDM), and was implemented in a new program called Neurite. Although the finite element method (FEM) is also explored as part of this research, the FDM provides the necessary flexibility and versatility to implement biological models, as well as the mathematical simplicity to extend them to large scale simulations with a low computational cost. Focusing first on the effect of electrophysiological-chemical properties on the mechanical proper ties, an adaptation of Neurite was developed to simulate microtubule polymerization in axonal growth and provide the axon mechanical properties as a function of microtubule occupancy. After calibrating the axon growth model against experimental results available in the literature, the mechanical characteristics can be tracked during the simulation. The axon mechanical properties show dramatic variations at the tip of the axon, where the growth cone supports the chemical and mechanical signaling. Based on the knowledge gained from the FDM scheme, and in order to go from 1D to 3D, this preliminary yet novel scheme paves the road for future studies with FEM. Focusing then on the effect of mechanical properties on the electrophysiological-chemical properties, Neurite was used to relate macroscopic mechanical loading to microscopic strains and strain rates, and simulate the electrical signal propagation along neurites under mechanical loading. The simulations were calibrated against experimental results published in the literature, thus providing a model able to predict the alteration of neuronal electrophysiological function under external damaging load, and linking mechanical injuries to subsequent acute functional deficits. To undertake large scale simulations, although other state-of-the-art architectures based on many integrated cores (MICs) were considered, the explicit and implicit solvers were implemented for central processing units (CPUs) and graphics processing units (GPUs). Scalability studies were done for both implementations showing promising results for extremely large scale simulations with GPUs. This thesis opens the avenue for future mechanical modeling approaches aimed at linking electrophysiological- chemical properties to mechanical properties. Its overarching goal is to enhance the bioengineering and medical communities knowledge on neuronal mechanics and functional deficits arising from damages produced by direct mechanical insults, such as TBI, or neurodegenerative evolving illness, such as AD.

Neurite, a finite difference large scale parallel program for the simulation of the electrical signal propagation in neurites under mechanical loading

With the growing body of research on traumatic brain injury and spinal cord injury, computational neuroscience has recently focused its modeling efforts on neuronal functional deficits following mechanical loading. However, in most of these efforts, cell damage is generally only characterized by purely mechanistic criteria, function of quantities such as stress, strain or their corresponding rates. The modeling of functional deficits in neurites as a consequence of macroscopic mechanical insults has been rarely explored. In particular, a quantitative mechanically based model of electrophysiological impairment in neuronal cells has only very recently been proposed (Jerusalem et al., 2013). In this paper, we present the implementation details of Neurite: the finite difference parallel program used in this reference. Following the application of a macroscopic strain at a given strain rate produced by a mechanical insult, Neurite is able to simulate the resulting neuronal electrical signal propagation, and thus the corresponding functional deficits. The simulation of the coupled mechanical and electrophysiological behaviors requires computational expensive calculations that increase in complexity as the network of the simulated cells grows. The solvers implemented in Neurite-explicit and implicit-were therefore parallelized using graphics processing units in order to reduce the burden of the simulation costs of large scale scenarios. Cable Theory and Hodgkin-Huxley models were implemented to account for the electrophysiological passive and active regions of a neurite, respectively, whereas a coupled mechanical model accounting for the neurite mechanical behavior within its surrounding medium was adopted as a link between lectrophysiology and mechanics (Jerusalem et al., 2013). This paper provides the details of the parallel implementation of Neurite, along with three different application examples: a long myelinated axon, a segmented dendritic tree, and a damaged axon. The capabilities of the program to deal with large scale scenarios, segmented neuronal structures, and functional deficits under mechanical loading are specifically highlighted.

Optimal Piecewise Linear Function Approximation for GPU-based Applications

Many computer vision and human-computer interaction applications developed in recent years need evaluating complex and continuous mathematical functions as an essential step toward proper operation. However, rigorous evaluation of this kind of functions often implies a very high computational cost, unacceptable in real-time applications. To alleviate this problem, functions are commonly approximated by simpler piecewise-polynomial representations. Following this idea, we propose a novel, efficient, and practical technique to evaluate complex and continuous functions using a nearly optimal design of two types of piecewise linear approximations in the case of a large budget of evaluation subintervals. To this end, we develop a thorough error analysis that yields asymptotically tight bounds to accurately quantify the approximation performance of both representations. It provides an improvement upon previous error estimates and allows the user to control the trade-off between the approximation error and the number of evaluation subintervals. To guarantee real-time operation, the method is suitable for, but not limited to, an efficient implementation in modern Graphics Processing Units (GPUs), where it outperforms previous alternative approaches by exploiting the fixed-function interpolation routines present in their texture units. The proposed technique is a perfect match for any application requiring the evaluation of continuous functions, we have measured in detail its quality and efficiency on several functions, and, in particular, the Gaussian function because it is extensively used in many areas of computer vision and cybernetics, and it is expensive to evaluate.

3D model reconstruction using neural gas accelerated on GPU

In this work, we propose the use of the neural gas (NG), a neural network that uses an unsupervised Competitive Hebbian Learning (CHL) rule, to develop a reverse engineering process. This is a simple and accurate method to reconstruct objects from point clouds obtained from multiple overlapping views using low-cost sensors. In contrast to other methods that may need several stages that include downsampling, noise filtering and many other tasks, the NG automatically obtains the 3D model of the scanned objects. To demonstrate the validity of our proposal we tested our method with several models and performed a study of the neural network parameterization computing the quality of representation and also comparing results with other neural methods like growing neural gas and Kohonen maps or classical methods like Voxel Grid. We also reconstructed models acquired by low cost sensors that can be used in virtual and augmented reality environments for redesign or manipulation purposes. Since the NG algorithm has a strong computational cost we propose its acceleration. We have redesigned and implemented the NG learning algorithm to fit it onto Graphics Processing Units using CUDA. A speed-up of 180× faster is obtained compared to the sequential CPU version.

Estudio y mejora de métodos de registro 3D: aceleración sobre unidades de procesamiento gráfico y caracterización del espacio de transformaciones iniciales

Durante los últimos años ha sido creciente el uso de las unidades de procesamiento gráfico, más conocidas como GPU (Graphic Processing Unit), en aplicaciones de propósito general, dejando a un lado el objetivo para el que fueron creadas y que no era otro que el renderizado de gráficos por computador. Este crecimiento se debe en parte a la evolución que han experimentado estos dispositivos durante este tiempo y que les ha dotado de gran potencia de cálculo, consiguiendo que su uso se extienda desde ordenadores personales a grandes cluster. Este hecho unido a la proliferación de sensores RGB-D de bajo coste ha hecho que crezca el número de aplicaciones de visión que hacen uso de esta tecnología para la resolución de problemas, así como también para el desarrollo de nuevas aplicaciones. Todas estas mejoras no solamente se han realizado en la parte hardware, es decir en los dispositivos, sino también en la parte software con la aparición de nuevas herramientas de desarrollo que facilitan la programación de estos dispositivos GPU. Este nuevo paradigma se acuñó como Computación de Propósito General sobre Unidades de Proceso Gráfico (General-Purpose computation on Graphics Processing Units, GPGPU). Los dispositivos GPU se clasifican en diferentes familias, en función de las distintas características hardware que poseen. Cada nueva familia que aparece incorpora nuevas mejoras tecnológicas que le permite conseguir mejor rendimiento que las anteriores. No obstante, para sacar un rendimiento óptimo a un dispositivo GPU es necesario configurarlo correctamente antes de usarlo. Esta configuración viene determinada por los valores asignados a una serie de parámetros del dispositivo. Por tanto, muchas de las implementaciones que hoy en día hacen uso de los dispositivos GPU para el registro denso de nubes de puntos 3D, podrían ver mejorado su rendimiento con una configuración óptima de dichos parámetros, en función del dispositivo utilizado. Es por ello que, ante la falta de un estudio detallado del grado de afectación de los parámetros GPU sobre el rendimiento final de una implementación, se consideró muy conveniente la realización de este estudio. Este estudio no sólo se realizó con distintas configuraciones de parámetros GPU, sino también con diferentes arquitecturas de dispositivos GPU. El objetivo de este estudio es proporcionar una herramienta de decisión que ayude a los desarrolladores a la hora implementar aplicaciones para dispositivos GPU. Uno de los campos de investigación en los que más prolifera el uso de estas tecnologías es el campo de la robótica ya que tradicionalmente en robótica, sobre todo en la robótica móvil, se utilizaban combinaciones de sensores de distinta naturaleza con un alto coste económico, como el láser, el sónar o el sensor de contacto, para obtener datos del entorno. Más tarde, estos datos eran utilizados en aplicaciones de visión por computador con un coste computacional muy alto. Todo este coste, tanto el económico de los sensores utilizados como el coste computacional, se ha visto reducido notablemente gracias a estas nuevas tecnologías. Dentro de las aplicaciones de visión por computador más utilizadas está el registro de nubes de puntos. Este proceso es, en general, la transformación de diferentes nubes de puntos a un sistema de coordenadas conocido. Los datos pueden proceder de fotografías, de diferentes sensores, etc. Se utiliza en diferentes campos como son la visión artificial, la imagen médica, el reconocimiento de objetos y el análisis de imágenes y datos de satélites. El registro se utiliza para poder comparar o integrar los datos obtenidos en diferentes mediciones. En este trabajo se realiza un repaso del estado del arte de los métodos de registro 3D. Al mismo tiempo, se presenta un profundo estudio sobre el método de registro 3D más utilizado, Iterative Closest Point (ICP), y una de sus variantes más conocidas, Expectation-Maximization ICP (EMICP). Este estudio contempla tanto su implementación secuencial como su implementación paralela en dispositivos GPU, centrándose en cómo afectan a su rendimiento las distintas configuraciones de parámetros GPU. Como consecuencia de este estudio, también se presenta una propuesta para mejorar el aprovechamiento de la memoria de los dispositivos GPU, permitiendo el trabajo con nubes de puntos más grandes, reduciendo el problema de la limitación de memoria impuesta por el dispositivo. El funcionamiento de los métodos de registro 3D utilizados en este trabajo depende en gran medida de la inicialización del problema. En este caso, esa inicialización del problema consiste en la correcta elección de la matriz de transformación con la que se iniciará el algoritmo. Debido a que este aspecto es muy importante en este tipo de algoritmos, ya que de él depende llegar antes o no a la solución o, incluso, no llegar nunca a la solución, en este trabajo se presenta un estudio sobre el espacio de transformaciones con el objetivo de caracterizarlo y facilitar la elección de la transformación inicial a utilizar en estos algoritmos.

Multi-Tenant Virtual GPUs for Optimising Performance of a Financial Risk Application

Graphics Processing Units (GPUs) are becoming popular accelerators in modern High-Performance Computing (HPC) clusters. Installing GPUs on each node of the cluster is not efficient resulting in high costs and power consumption as well as underutilisation of the accelerator. The research reported in this paper is motivated towards the use of few physical GPUs by providing cluster nodes access to remote GPUs on-demand for a financial risk application. We hypothesise that sharing GPUs between several nodes, referred to as multi-tenancy, reduces the execution time and energy consumed by an application. Two data transfer modes between the CPU and the GPUs, namely concurrent and sequential, are explored. The key result from the experiments is that multi-tenancy with few physical GPUs using sequential data transfers lowers the execution time and the energy consumed, thereby improving the overall performance of the application.

A fast parallel hyperspectral coded aperture algorithm for compressive sensing using OpenCL

In this paper, we develop a fast implementation of an hyperspectral coded aperture (HYCA) algorithm on different platforms using OpenCL, an open standard for parallel programing on heterogeneous systems, which includes a wide variety of devices, from dense multicore systems from major manufactures such as Intel or ARM to new accelerators such as graphics processing units (GPUs), field programmable gate arrays (FPGAs), the Intel Xeon Phi and other custom devices. Our proposed implementation of HYCA significantly reduces its computational cost. Our experiments have been conducted using simulated data and reveal considerable acceleration factors. This kind of implementations with the same descriptive language on different architectures are very important in order to really calibrate the possibility of using heterogeneous platforms for efficient hyperspectral imaging processing in real remote sensing missions.

A three-dimensional representation method for noisy point clouds based on growing self-organizing maps accelerated on GPUs

The research described in this thesis was motivated by the need of a robust model capable of representing 3D data obtained with 3D sensors, which are inherently noisy. In addition, time constraints have to be considered as these sensors are capable of providing a 3D data stream in real time. This thesis proposed the use of Self-Organizing Maps (SOMs) as a 3D representation model. In particular, we proposed the use of the Growing Neural Gas (GNG) network, which has been successfully used for clustering, pattern recognition and topology representation of multi-dimensional data. Until now, Self-Organizing Maps have been primarily computed offline and their application in 3D data has mainly focused on free noise models, without considering time constraints. It is proposed a hardware implementation leveraging the computing power of modern GPUs, which takes advantage of a new paradigm coined as General-Purpose Computing on Graphics Processing Units (GPGPU). The proposed methods were applied to different problem and applications in the area of computer vision such as the recognition and localization of objects, visual surveillance or 3D reconstruction.