Martin Peniak · Ghent University Academic Bibliography (Ghent University) · 2014
This thesis presents the first investigation of the impact of GPU computing on cognitive robotics by providing a series of novel experiments in the area of action and language acquisition in humanoid robots and computer vision. Cognitive robotics is concerned with endowing robots with high-level cognitive capabilities to enable the achievement of complex goals in complex environments. Reaching the ultimate goal of developing cognitive robots will require tremendous amounts of computational power, which was until recently provided mostly by standard CPU processors. CPU cores are optimised for serial code execution at the expense of parallel execution, which renders them relatively inefficient when it comes to high-performance computing applications. The ever-increasing market demand for high-performance, real-time 3D graphics has evolved the GPU into a highly parallel, multithreaded, many-core processor extraordinary computational power and very high memory bandwidth. These vast computational resources of modern GPUs can now be used by the most of the cognitive robotics models as they tend to be inherently parallel. Various interesting and insightful cognitive models were developed and addressed important scientific questions concerning action-language acquisition and computer vision. While they have provided us with important scientific insights, their complexity and application has not improved much over the last years. The experimental tasks as well as the scale of these models are often minimised to avoid excessive training times that grow exponentially with the number of neurons and the training data. This impedes further progress and development of complex neurocontrollers that would be able to take the cognitive robotics research a step closer to reaching the ultimate goal of creating intelligent machines. This thesis presents several cases where the application of the GPU computing on cognitive robotics algorithms resulted in the development of large-scale neurocontrollers of previously unseen complexity enabling the conducting of the novel experiments described herein.
Cited by 0Open access
Sengupta, Shubhabrata, Harris, Mark, Zhang, Yao et al. · Eurographics · 2007
The scan primitives are powerful, general-purpose data-parallel primitives that are building blocks for a broad range of applications. We describe GPU implementations of these primitives, specifically an efficient formulation and implementation of segmented scan, on NVIDIA GPUs using the CUDA API. Using the scan primitives, we show novel GPU implementations of quicksort and sparse matrix-vector multiply, and analyze the performance of the scan primitives, several sort algorithms that use the scan primitives, and a graphical shallow-water fluid simulation using the scan framework for a tridiagonal matrix solver.
Cited by 4Open access
Pleiter, Dirk · JuSER (Forschungszentrum Jülich) · 2018
Cited by 0
Jonathan Jung · SPIRE - Sciences Po Institutional REpository · 2012
International audience
Cited by 0
El‐Ghazali Talbi · HAL (Le Centre pour la Communication Scientifique Directe) · 2011
International audience
Cited by 0
Gu, Zhongshu, Valdez, Enriquillo, Ahmed, Salman et al. · ArXiv.org · 2025
GPU Confidential Computing (GPU-CC) was introduced as part of the NVIDIA Hopper Architecture, extending the trust boundary beyond traditional CPU-based confidential computing. This innovation enables GPUs to securely process AI workloads, providing a robust and efficient solution for handling sensitive data. For end users, transitioning to GPU-CC mode is seamless, requiring no modifications to existing AI applications. However, this ease of adoption contrasts sharply with the complexity of the underlying proprietary systems. The lack of transparency presents significant challenges for security researchers seeking a deeper understanding of GPU-CC's architecture and operational mechanisms. The challenges of analyzing the NVIDIA GPU-CC system arise from a scarcity of detailed specifications, the proprietary nature of the ecosystem, and the complexity of product design. In this paper, we aim to demystify the implementation of NVIDIA GPU-CC system by piecing together the fragmented and incomplete information disclosed from various sources. Our investigation begins with a high-level discussion of the threat model and security principles before delving into the low-level details of each system component. We instrument the GPU kernel module -- the only open-source component of the system -- and conduct a series of experiments to identify the security weaknesses and potential exploits. For certain components that are out of reach through experiments, we propose well-reasoned speculations about their inner working mechanisms. We have responsibly reported all security findings presented in this paper to the NVIDIA PSIRT Team.
Cited by 0Open access
Piccoli, María Fabiana · El Servicio de Difusión de la Creación Intelectual (National University of La Plata) · 2012
Through this textbook (written in Spanish), the author introduces the GPU as a parallel computer that is able to solve general-propose problems. The book has a didactic approach, it includes the first steps to CUDA programming, some tips to take account to develop good applications performance, many examples and exercises.
Cited by 0
Jaewook Kim, Tae Joon Jun, Daeyoun Kang et al. · 2018
A new form of cloud computing, serverless computing, is drawing attention as a new way to design micro-services architectures. In a serverless computing environment, services are developed as service functional units. The function development environment of all serverless computing framework at present is CPU based. In this paper, we propose a GPU-supported serverless computing framework that can deploy services faster than existing serverless computing framework using CPU. Our core approach is to integrate the open source serverless computing framework with NVIDIA-Docker and deploy services based on the GPU support container. We have developed an API that connects the open source framework to the NVIDIA-Docker and commands that enable GPU programming. In our experiments, we measured the performance of the framework in various environments. As a result, developers who want to develop services through the framework can deploy high-performance micro services and developers who want to run deep learning programs without a GPU environment can run code on remote GPUs with little performance degradation.
Cited by 34
Lorenzo Dematté, Davide Prandi · Briefings in Bioinformatics · 2010
The development of detailed, coherent, models of complex biological systems is recognized as a key requirement for integrating the increasing amount of experimental data. In addition, in-silico simulation of bio-chemical models provides an easy way to test different experimental conditions, helping in the discovery of the dynamics that regulate biological systems. However, the computational power required by these simulations often exceeds that available on common desktop computers and thus expensive high performance computing solutions are required. An emerging alternative is represented by general-purpose scientific computing on graphics processing units (GPGPU), which offers the power of a small computer cluster at a cost of approximately $400. Computing with a GPU requires the development of specific algorithms, since the programming paradigm substantially differs from traditional CPU-based computing. In this paper, we review some recent efforts in exploiting the processing power of GPUs for the simulation of biological systems.
Cited by 134Open access
Jack Choquette, Wish Gandhi · 2020
This article consists only of a collection of slides from the author's conference presentation.
Cited by 92
Wilson Wai Lun Fung · Open Collections · 2015
Many applications with regular parallelism have been shown to benefit from using Graphics Processing Units (GPUs). However, employing GPUs for applications with irregular parallelism tends to be a risky process, involving significant effort from the programmer and an uncertain amount of performance/efficiency benefit. One known challenge in developing GPU applications with irregular parallelism is the underutilization of SIMD hardware in GPUs due to the application’s irregular control flow behavior, known as branch divergence. Another major development effort is to expose the available parallelism in the application as 1000s of concurrent threads without introducing data races or deadlocks. The GPU software developers may need to spend significant effort verifying the data synchronization mechanisms used in their applications. Despite various research studies indicating the potential benefits, the risks involved may discourage software developers from employing GPUs for this class of applications. This dissertation aims to reduce the burden on GPU software developers with two major enhancements to GPU architectures. First, thread block compaction (TBC) is a microarchitecture innovation that reduces the performance penalty caused by branch divergence in GPU applications. Our evaluations show that TBC provides an average speedup of 22% over a baseline per-warp, stack-based reconvergence mechanism on a set of GPU applications that suffer significantly from branch divergence. Second, Kilo TM is a cost effective, energy efficient solution for supporting transactional memory (TM) on GPUs. With TM, programmers can uses transactions instead of fine-grained locks to create deadlock-free, maintainable, yet aggressively-parallelized code. In our evaluations, Kilo TM achieves 192X speedup over coarse-grained locking and captures 66% of the performance of fine-grained locking with 34% energy overhead.
Cited by 5Open access
Debapriya Chatterjee, Andrew DeOrio, Valeria Bertacco · ACM Transactions on Design Automation of Electronic Systems · 2011
Functional verification of modern digital designs is a crucial, time-consuming task impacting not only the correctness of the final product, but also its time to market. At the heart of most of today’s verification efforts is logic simulation, used heavily to verify the functional correctness of a design for a broad range of abstraction levels. In mainstream industry verification methodologies, typical setups coordinate the validation effort of a complex digital system by distributing logic simulation tasks among vast server farms for months at a time. Yet, the performance of logic simulation is not sufficient to satisfy the demand, leading to incomplete validation processes, escaped functional bugs, and continuous pressure on the EDA industry to develop faster simulation solutions. In this work we propose GCS, a solution to boost the performance of logic simulation, gate-level simulation in particular, by more than a factor of 10 using recent hardware advances in Graphic Processing Unit (GPU) technology. Noting the vast available parallelism in the hardware of modern GPUs and the inherently parallel structures of gate-level netlists, we propose novel algorithms for the efficient mapping of complex designs to parallel hardware. Our novel simulation architecture maximizes the utilization of concurrent hardware resources while minimizing expensive communication overhead. The experimental results show that our GPU-based simulator is capable of handling the validation of industrial-size designs while delivering more than an order-of-magnitude performance improvements on average, over the fastest multithreaded simulators commercially available.
Cited by 26
Lucas Henrique Moreira Silva · 2018
The data cube is a fundamental relational operator for decision support systems, thus very important for analytics. Unfortunately, a full data cube with all of its tuples has exponential complexity in terms of runtime and memory consumption as the dimensions increase linearly, so algorithms to reduce query response times continue under development. The problem stated in this work is: how can we reduce complex multidimensional queries response times from high dimensional data cubes? The problem is aggravated if recurrent updates occur and if there is a huge volume of high dimensional data to be managed. The hypothesis of this work is that clusters of CPU-GPU devices can speedup queries from high dimensional holistic data cubes that are updated constantly. The alternative solution presented in this work, named JCL-GPU-Cubing, partitions the base relation into multiple independent sub- cubes. These multiple sub-cubes represent a partial data cube to reduce the exponentiality and they are used to perform queries in CPU or in CPU-GPU computer architectures efficiently. The experimental evaluations using complex multidimensional queries demonstrated that the CPU cluster version scaled well when the base relation increased and the CPU-GPU version outperformed the CPU only version in certain scenarios.
Cited by 1
Dubois, Gert · Document Server@UHasselt (UHasselt) · 2014
Digitale video is over de jaren heen steeds belangrijker geworden in het dagelijkse leven. De compressie en encoding ervan is dan ook een zeer belangrijk aspect van digitale video. Om video encoders nog verder te versnellen en in de toekomst performant te houden dient gezocht te worden naar algoritmen die optimaal gebruik maken van multi-core processors. Een belangrijk type van deze multi-core processors zijn de GPU's, deze bieden veel rekenkracht aan maar om deze te benutten dient een hoge graad van parallellisatie bereikt te worden.\nDeze thesis onderzoekt de mogelijkheden om video encoding algoritmen te implementeren op de GPU. Uit het onderzoek blijkt dat GPU parallellisatie tot 10x snelheids verbeteringen laat voor de typische componenten nodig voor het encoderen van I-frames vergeleken met een multi-threaded CPU implementatie. Voor het encoderen van P-frames laat de GPU implementatie zelfs tot 20x snelheidsverbeteringen zien vergeleken met de multi-threaded CPU implementatie.
Cited by 0Open access
Lindberg, Amund · Bergen Open Research Archive (BORA) (University of Bergen) · 2020
Every day, a seemingly infinite amount of data is generated by social media.Analyzing this data is a complex, but rewarding task that can be solved faster with efficient algorithms and powerful computers.This thesis is concerned with identifying influential users in the Twitter micro-blog network.Designed to identify users that have an important role in spreading (mis)information, our influence measure takes both activity and network relationships into consideration and is based on breadth-first traversal of networks that arise from the sharing of content.The questions we try to answer are how to implement an efficient, parallel solution to our approach in NVIDIA CUDA using a graphical processing unit (GPU), and whether a parallel implementation gives a performance improvement over a sequential one.We give a detailed description of our influence measure in the context of social network analysis and relevant research on Twitter.An introduction to parallel programming and GPUs, as well as an explanation of the basics of CUDA is provided.This is useful as we explore known techniques on parallelizing the graph traversal algorithm Breadth-First Search (BFS), which we adapt and use in our implementation.Our experiments show that a simple implementation can achieve a considerable speedup on Twitter data.Thanks to my supervisor Fredrik Manne for help with coding, writing, and all the great advice. Thanks to Johannes Langguth
Cited by 0Open access
Song Jun Park · 2009
Parallel systems are becoming ubiquitous in the world of computing as evidenced by multi-core processors, heterogeneous Cell broadband engine, and highly parallel graphics processing units (GPUs). All parallel systems share a requirement that parallel programming is necessary to leverage multiple cores. As a result of this trend, multi-core CPUs are no longer a clear winner due to its peaked clock frequency and programming effort involved in parallelizing code for multi-core architecture. Given such drawbacks, dataparallel applications might benefit from GPU assisted computing. GPUs are the most popular and inexpensive accelerators. To evaluate GPU-based computing, a floating-point intensive algorithm for a radar imaging application is chosen for analysis. The paper attempts to present a fair performance comparison of CPU and GPU implementations.
Cited by 9
Zhe Fan, Feng Qiu, Arie Kaufman et al. · 2005
Inspired by the attractive Flops/dollar ratio and the incredible growth in the speed of modern graphics processing units (GPUs), we propose to use a cluster of GPUs for high performance scientific computing. As an example application, we have developed a parallel flow simulation using the lattice Boltzmann model (LBM) on a GPU cluster and have simulated the dispersion of airborne contaminants in the Times Square area of New York City. Using 30 GPU nodes, our simulation can compute a 480x400x80 LBM in 0.31 second/step, a speed which is 4.6 times faster than that of our CPU cluster implementation. Besides the LBM, we also discuss other potential applications of the GPU cluster, such as cellular automata, PDE solvers, and FEM.
Cited by 436
Yusuke Fujii, Takuya Azumi, Nobuhiko Nishio et al. · 2013 International Conference on Parallel and Distributed Systems · 2013
Graphics processing units (GPUs) embrace many-core compute devices where massively parallel compute threads are offloaded from CPUs. This heterogeneous nature of GPU computing raises non-trivial data transfer problems especially against latency-critical real-time systems. However even the basic characteristics of data transfers associated with GPU computing are not well studied in the literature. In this paper, we investigate and characterize currently-achievable data transfer methods of cutting-edge GPU technology. We implement these methods using open-source software to compare their performance and latency for real-world systems. Our experimental results show that the hardware-assisted direct memory access (DMA) and the I/O read-and-write access methods are usually the most effective, while on-chip micro controllers inside the GPU are useful in terms of reducing the data transfer latency for concurrent multiple data streams. We also disclose that CPU priorities can protect the performance of GPU data transfers.
Cited by 47
Yupu Song (12501028), Cailin Li (17743113), Shiyang Xiao (17743116) et al. · Figshare · 2024
<div><p>In the process of Canny edge detection, a large number of high complexity calculations such as Gaussian filtering, gradient calculation, non-maximum suppression, and double threshold judgment need to be performed on the image, which takes up a lot of operation time, which is a great challenge to the real-time requirements of the algorithm. The traditional Canny edge detection technology mainly uses customized equipment such as DSP and FPGA, but it has some problems, such as long development cycle, difficult debugging, resource consumption, and so on. At the same time, the adopted CUDA platform has the problem of poor cross-platform. In order to solve this problem, a fine-grained parallel Canny edge detection method is proposed, which is optimized from three aspects: task partition, vector memory access, and NDRange optimization, and CPU-GPU collaborative parallelism is realized. At the same time, the parallel Canny edge detection methods based on multi-core CPU and CUDA architecture are designed. The experimental results show that OpenCL accelerated Canny edge detection algorithm (OCL_Canny) achieves 20.68 times acceleration ratio compared with CPU serial algorithm at 7452 × 8024 image resolution. At the image resolution of 3500 × 3500, the OCL_Canny algorithm achieves 3.96 times the acceleration ratio compared with the CPU multi-threaded Canny parallel algorithm. At 1024 × 1024 image resolution, the OCL_Canny algorithm achieves 1.21 times the acceleration ratio compared with the CUDA-based Canny parallel algorithm. The effectiveness and performance portability of the proposed Canny edge detection parallel algorithm are verified, and it provides a reference for the research of fast calculation of image big data.</p></div>
Cited by 0Open access
Stéphane Vialle · HAL (Le Centre pour la Communication Scientifique Directe) · 2012
International audience
Cited by 0