Current Graphics Processing Units Research Articles

Efficient GPU Cloud architectures for outsourcing high-performance processing to the Cloud

AbstractThe world is becoming increasingly dependant in computing intensive applications. The appearance of new paradigms, such as Internet of Things (IoT), and advances in technologies such as Computer Vision (CV) and Artificial Intelligence (AI) are creating a demand for high-performance applications. In this regard, Graphics Processing Units (GPUs) have the ability to provide better performance by allowing a high degree of data parallelism. These devices are also beneficial in specialized fields of manufacturing industry such as CAD/CAM. For all these applications, there is a recent tendency to offload these computations to the Cloud, using a computing offloading Cloud architecture. However, the use of GPUs in the Cloud presents some inefficiencies, where GPU virtualization is still not fully resolved, as our research on what main Cloud providers currently offer in terms of GPU Cloud instances shows. To address these problems, this paper first makes a review of current GPU technologies and programming techniques that increase concurrency, to then propose a Cloud computing outsourcing architecture to make more efficient use of these devices in the Cloud.

Bio-inspired computing with magnetic skyrmions using deep learning

Spintronic systems demonstrate significant value due to variety of features that can be used in bio-inspired computing applications. Skyrmion is a magnetic nanostructure which can be used as an information carrier. In this paper, we simulate vortex-free Gaussian wave packet by incorporating spin–orbit coupling (SoC) into three-component Bose–Einstein condensates (BECs). We next observe that skyrmions can be naturally generated from vortex-free Gaussian wave packet, further these structurally asymmetric devices are constructed and initial device classifications are carried out. A spin device consisting of vertically-stacked multiple skyrmions is proposed termed as skyrmion spintronic synapse (SkSS) and skyrmion spintronic Neuron (SkSN). This paper depicts training the neural network with surrogate gradients with deep learning strategies using topologically nontrivial spin textures. The current popular Graphics Processing Unit (GPU)-based neural network computations are compared with the proposed method. This research paves a way for a neuromorphic computing systems on exotic topological solitons.

Full Core Pin-Level VVER-440 Simulation of a Rod Drop Experiment with the GPU-Based Monte Carlo Code GUARDYAN

Targeting ultimate fidelity reactor physics calculations the Dynamic Monte Carlo (DMC) method simulates reactor transients without resorting to static or quasistatic approximations. Due to the capability to harness the computing power of Graphics Processing Units, the GUARDYAN (GpU Assisted Reactor DYnamic ANalysis) code has been recently upscaled to perform pin-by-pin simulations of power plant scale systems as demonstrated in this paper. A recent rod drop experiment at a VVER-440/213 (vodo-vodyanoi enyergeticheskiy reaktor) type power plant at Paks NPP, Hungary, was considered and signals of ex-core detectors placed at three different positions were simulated successfully by GUARDYAN taking realistic fuel loading, including burn-up data into account. Results were also compared to the time-dependent Paks NPP in-house nodal diffusion code VERETINA (VERONA: VVER Online Analysis and RETINA: Reactor Thermo-hydraulics Interactive). Analysis is given of the temporal and spatial variance distribution of GUARDYAN fuel pin node-wise power estimates. We can conclude that full core, pin-wise DMC power plant simulations using realistic isotope concentrations are feasible in reasonable computing times down to 1–2% error of ex-core detector signals using current GPU (Graphics Processing Unit) High Performance Computing architectures, thereby demonstrating a technological breakthrough.

How to Ensure a Good Depth Perception on a Light Field Display

<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">In this article, we define the technical requirements for a light field display to provide effective depth cues such as binocular disparity, motion parallax, and accommodation at a pixel resolution expected by the eye. These requirements are evaluated with respect to existing technologies (e.g.,integral imaging displays) or expected future ones (e.g., micro-light-emitting diode (LED) displays). Simulation of light field displays are proposed for different display sizes (smartphone, desktop, or TV), resolution per view, and pixel pitch. These simulations convey the main impact of the pixel pitch to reach the light field display goal in terms of depth cue requirements. The last part of the paper focusses on data generation issues: how to generate the necessary amount of views at the display side and what the adapted data formats are for such a purpose. The paper is associated with supplementary videos that will illustrate the view generation topic using a dedicated data format that is optimized for current graphics processing unit (GPU) processors. Video capture of early-stage light field displays will also be provided</i> .

The potential of computational fluid dynamics simulations of airflow in the nasal cavity

Computational Fluid Dynamics (CFD) is a well-established and accepted tool for simulation and prediction of complex physical phenomena e.g., in combustion, aerodynamics or blood circulation. Recently CFD has entered the medical field due to the readily available high computational power of current graphics processing units, GPUs. Efficient numerical codes, commercial or open source, are available now.

Prediction-Based Error Correction for GPU Reliability with Low Overhead

Scientific and simulation applications are continuously gaining importance in many fields of research and industries. These applications require massive amounts of memory and substantial arithmetic computation. Therefore, general-purpose computing on graphics processing units (GPGPU), which combines the computing power of graphics processing units (GPUs) and general CPUs, have been used for computationally intensive scientific and big data processing applications. Because current GPU architectures lack hardware support for error detection in computation logic, GPGPU has low reliability. Unlike graphics applications, errors in GPGPU can lead to serious problems in general-purpose computing applications. These applications are often intertwined with human life, meaning that errors can be life threatening. Therefore, this paper proposes a novel prediction-based error correction method called Prediction-based Error Correction (PRECOR) for GPU reliability, which detects and corrects errors in GPGPU platforms with a focus on errors in computational elements. The implementation of the proposed architecture needs a small number of checkpoint buffers in order to fix errors in computational logic. The PRECOR architecture has prediction buffers and controller units for predicting erroneous outputs before performing rollback. Following a rollback, the architecture confirms the accuracy of its predictions. The proposed method effectively reduces the hardware and time overheads required to correct errors. Experimental results confirm that PRECOR efficiently fixes errors with low hardware and time overheads.

Accelerating Auxiliary-Field Quantum Monte Carlo Simulations of Solids with Graphical Processing Units.

We outline how auxiliary-field quantum Monte Carlo (AFQMC) can leverage graphical processing units (GPUs) to accelerate the simulation of solid state systems. By exploiting conservation of crystal momentum in the one- and two-electron integrals, we show how to efficiently formulate the algorithm to best utilize current GPU architectures. We provide a detailed description of different optimization strategies and profile our implementation relative to standard approaches, demonstrating a factor of 40 speedup over a CPU implementation. With this increase in computational power, we demonstrate the ability of AFQMC to systematically converge solid state calculations with respect to basis set and system size by computing the cohesive energy of carbon in the diamond structure to within 0.02 eV of the experimental result.

A GPU Algorithm for Agent-Based Models to Simulate the Integration of Cell Membrane Signals

Simulation of complex biological systems with agent-based models is becoming more relevant with the increase in Graphics Processing Unit (GPU) power. In those simulations, up to millions of virtual cells are individually computed, involving daunting processing times. An important part of computational models is the algorithm that manages how agents perceive their surroundings. This can be particularly problematic in three-dimensional environments where agents have deformable virtual membranes. This article presents a GPU algorithm that gives the possibility for agents to integrate the signals scattered on their virtual membrane. It is detailed to be coded in languages like OpenCL or Cuda. Its performances are tested to show its speed with current GPU devices. Finally, it was implemented inside an existing software to test and illustrate the possibilities it offers.

English

Bioinformatics is an emerging field, where information technology usage can significantly accelerate life science research. It is a relatively new field and the scope of exploring new tools and techniques seems immense. One major field where bioinformatics plays important role is next generation sequence analysis (NGS), in which an unknown genome is shuttered into pieces and tried to align it to a reference known genome to decipher its functions using sequence comparison. The first well known application of this technology is the human genome project which took nearly 10 years to finish. With advancements in central processing units (CPUs), the alignment time has improved, but has not reached optimal. There seems a constant need to improve this computing time, which made the scope for using graphics processing units (GPUs) and parallel programming tasks to replace CPUs. With access to high performance multi-thread, multi-core parallel computing supercomputers, several GPU based sequence alignment tools have been published recently, some of the major tools are BarraCUDA, CUSHAW, GPU-BWT, SOAP3, and SARUMAN, which claim to speed up the processes anywhere between 2x and 10x times. Most of these tools can be compiled on GCC 4.3 compilers with CUDA. This paper focuses on compiling the current GPU based alignment tools on 70.7 million read pairs (Illumina HiSeq 2000) to align them on a human genome and check its efficiency (time sensitivity and alignment specificity) compared to traditional CPU based alignment (Bowtie) tool. Resulting observations would help researchers choose the appropriate GPU alignment tool to suffice their computing needs. Key words: CUDA, sequencing, alignment, graphics processing units (GPUs), central processing units (CPUs).

Real-time ultrasound image reconstruction as an inverse problem on a GPU

Ultrasonic image reconstruction methods based on inverse problems have been shown to produce sharp, high-quality images using more information about the acquisition process in its processing. This improved reconstruction has high computational cost, usually requiring to solve large systems and making real-time imaging very difficult. Parallelizing the reconstruction using graphics processing units (GPU) can significantly accelerate this processing, but the amount of memory needed by current system models is high for current GPU capacity. This paper presents a new system model to halve this memory requirement; it exploits the symmetry of the point spread functions (PSF) of the system matrix that occurs when symmetric transducers are used for acquisition. In this case, only one of the two symmetric PSFs needs to be stored; the other function is produced by reordering the stored one. Thus, we can reconstruct ultrasound images that are twice as large, making real-time reconstruction on a GPU possible for this application.

Cellular Simulation for Distributed Sensing over Complex Terrains.

Long-range radio transmissions open new sensor application fields, in particular for environment monitoring. For example, the LoRa radio protocol enables connecting remote sensors at a distance as long as ten kilometers in a line-of-sight. However, the large area covered also brings several difficulties, such as the placement of sensing devices in regards to topology in geography, or the variability of communication latency. Sensing the environment also carries constraints related to the interest of sensing points in relation to a physical phenomenon. Thus, criteria for designs are evolving a lot from the existing methods, especially in complex terrains. This article describes simulation techniques based on geography analysis to compute long-range radio coverages and radio characteristics in these situations. As radio propagation is just a particular case of physical phenomena, it is shown how a unified approach also allows for characterizing the behavior of potential physical risks. The case of heavy rainfall and flooding is investigated. Geography analysis is achieved using segmentation tools to produce cellular systems which are in turn translated into code for high-performance computations. The paper provides results from practical complex terrain experiments using LoRa, which confirm the accuracy of the simulation, and scheduling characteristics for sample networks. Performance tables are produced for these simulations on current Graphics Processing Units (GPUs).

A Fuzzy Neural Network Based Dynamic Data Allocation Model on Heterogeneous Multi-GPUs for Large-scale Computations

The parallel computation capabilities of modern GPU (Graphics Processing Unit) processors have attracted increasing attention from researchers and engineers who have been conducting high computational throughput studies. However, current single GPU based engineering solutions are often struggle to fulfill their real-time requirements. Thus, the multi-GPU-based approach has become a popular and cost-effective choice for tackling the demands. In those cases, the computational load balancing over multiple GPU “nodes” is often the key and bottleneck that affect the quality and performance of the runtime system. The existing load balancing approaches are mainly based on the assumption that all GPU nodes in the same computer framework are of equal computational performance, which are often not the case due to cluster design and other legacy issues. This paper presents a novel dynamic load balancing (DLB) model for rapid data division and allocation on heterogeneous GPU nodes based on an innovative fuzzy neural network (FNN). In this research, a 5-state parameter feedback mechanism defining the overall cluster and node performances is proposed. The corresponding FNN-based DLB model will be capable of monitoring and predicting individual node performance under different workload scenarios. A real-time adaptive scheduler has been devised to reorganize the data inputs to each node when necessary to maintain their runtime computational performances. The devised model has been implemented on two dimensional (2D) discrete wavelet transform (DWT) tasks for evaluation. Experiment results show that this DLB model has enabled a high computational throughput while ensuring real-time and precision requirements from complex computational tasks.

CWLP: coordinated warp scheduling and locality-protected cache allocation on GPUs

As we approach the exascale era in supercomputing, designing a balanced computer system with a powerful computing ability and low power requirements has becoming increasingly important. The graphics processing unit (GPU) is an accelerator used widely in most of recent supercomputers. It adopts a large number of threads to hide a long latency with a high energy efficiency. In contrast to their powerful computing ability, GPUs have only a few megabytes of fast on-chip memory storage per streaming multiprocessor (SM). The GPU cache is inefficient due to a mismatch between the throughput-oriented execution model and cache hierarchy design. At the same time, current GPUs fail to handle burst-mode long-access latency due to GPU’s poor warp scheduling method. Thus, benefits of GPU’s high computing ability are reduced dramatically by the poor cache management and warp scheduling methods, which limit the system performance and energy efficiency. In this paper, we put forward a coordinated warp scheduling and locality-protected (CWLP) cache allocation scheme to make full use of data locality and hide latency. We first present a locality-protected cache allocation method based on the instruction program counter (LPC) to promote cache performance. Specifically, we use a PC-based locality detector to collect the reuse information of each cache line and employ a prioritised cache allocation unit (PCAU) which coordinates the data reuse information with the time-stamp information to evict the lines with the least reuse possibility. Moreover, the locality information is used by the warp scheduler to create an intelligent warp reordering scheme to capture locality and hide latency. Simulation results show that CWLP provides a speedup up to 19.8% and an average improvement of 8.8% over the baseline methods.

Solving Large Problem Sizes of Index-Digit Algorithms on GPU: FFT and Tridiagonal System Solvers

Current Graphics Processing Units (GPUs) are capable of obtaining high computational performance in scientific applications. Nevertheless, programmers have to use suitable parallel algorithms for these architectures and usually have to consider optimization techniques in the implementation in order to achieve said performance. There are many efficient proposals for limited-size problems which fit directly in the shared memory of CUDA GPUs, however, there are few GPU proposals that tackle the design of efficient algorithms for large problem sizes that exceed shared memory storage capacity. In this work, we present a tuning strategy that addresses this problem for some parallel prefix algorithms that can be represented according to a set of common permutations of the digits of each of its element indices [1], denoted as Index-Digit (ID) algorithms. Specifically, our strategy has been applied to develop flexible Multi-Stage (MS) algorithms for the Fast Fourier Transform (FFT) algorithm (MS-ID-FFT) and a tridiagonal system solver (MS-ID-TS) on the GPU. The resulting implementation is compact and outperforms other well-known and commonly used state-of-the-art libraries, with an improvement of up to 1.47x with respect to NVIDIA's complex CUFFT, and up to 33.2x in comparison with NVIDIA's CUSPARSE for real data tridiagonal systems.

Deep Learning in Medical Imaging: General Overview.

The artificial neural network (ANN)–a machine learning technique inspired by the human neuronal synapse system–was introduced in the 1950s. However, the ANN was previously limited in its ability to solve actual problems, due to the vanishing gradient and overfitting problems with training of deep architecture, lack of computing power, and primarily the absence of sufficient data to train the computer system. Interest in this concept has lately resurfaced, due to the availability of big data, enhanced computing power with the current graphics processing units, and novel algorithms to train the deep neural network. Recent studies on this technology suggest its potentially to perform better than humans in some visual and auditory recognition tasks, which may portend its applications in medicine and healthcare, especially in medical imaging, in the foreseeable future. This review article offers perspectives on the history, development, and applications of deep learning technology, particularly regarding its applications in medical imaging.

An efficiency study of a pivot-based algorithm for similarity search on a heterogeneous platform

Graphics processing units have definitely consolidated a position of privilege in the acceleration of general purpose applications. Search algorithms in large databases are a clear example of applications that benefit from computing platforms based on these devices. To obtain an efficient implementation of a given code using these platforms, it is very important to take into account their features. However, the characteristics of the application and certain overheads still introduced by these platforms, make it not always advisable to use these devices to obtain significant time reductions. In this paper, we show how different properties of current graphics processing units are exploited for improving a version of the general metric structure similarity search algorithm introduced by the authors, and compare it with a multithreaded version of the same algorithm using conventional processors. The analysis of the results provides us relevant data to determine the most appropriate computing platform.

Constructing a Bioinformatics Platform with Web and Mobile Services Based on NVIDIA Jetson TK1

Current high-end graphics processing units (abbreviate to GPUs), such as NVIDIA Tesla, Fermi, Kepler series cards which contain up to thousand cores per-chip, are widely used in the high performance computing fields. These GPU cards (called desktop GPUs) should be installed in personal computers/servers with desktop CPUs; moreover, the cost and power consumption of constructing a high performance computing platform with these desktop CPUs and GPUs are high. NVIDIA releases Tegra K1, called Jetson TK1, which contains 4 ARM Cortex-A15 CPUs and 192 CUDA cores (Kepler GPU) and is an embedded board with low cost, low power consumption and high applicability advantages for embedded applications. NVIDIA Jetson TK1 becomes a new research direction. Hence, in this paper, a bioinformatics platform was constructed based on NVIDIA Jetson TK1. ClustalWtk and MCCtk tools for sequence alignment and compound comparison were designed on this platform, respectively. Moreover, the web and mobile services for these two tools with user friendly interfaces also were provided. The experimental results showed that the cost-performance ratio by NVIDIA Jetson TK1 is higher than that by Intel XEON E5-2650 CPU and NVIDIA Tesla K20m GPU card.

Performance of a projection method for incompressible flows on heterogeneous hardware

Practical experiments on the flow in a lid-driven cavity are carried out to compare the performance of a second-order finite volume Navier–Stokes solver for incompressible fluids employing a projection method, when using various linear solver libraries on central processing units (CPUs) and graphical processing units (GPUs). The goal of the paper is to identify the potential of GPU acceleration using the built-in GPU solvers of PETSc and to provide information if the usage of GPUs is beneficial for this type of fluid solver in terms of performance and implementation effort. Additionally, energy consumption having emerged as another important goal of optimization in high-performance computing is addressed as well. In this study, the solvers available in the PETSc library, which are running on CPU as well as with GPU support, are compared with the solvers provided by the hypre library in a systematic way. The power consumption of the CPU and the GPU during the solution is measured to assess the energy efficiency in terms of the performance-per-Watt ratio. It is found that for the considered numerical scheme the usage of iterative solvers on current GPU systems is not necessarily beneficial, neither in terms of performance nor in terms of energy consumption, since both libraries, PETSc and hypre, provide highly-optimized solvers for massively parallel CPU systems.

Fine-grained GPU implementation of assembly-free iterative solver for finite element problems

This paper proposes a fine-grained implementation of matrix-free Conjugate Gradient (CG) solver for Finite Element Analysis (FEA) using Graphics Processing Unit (GPU) architectures. The use of GPU computing in FEA is today an active research field. This is primary due to current GPU sparse solvers are partially parallelizable and can hardly make use of Data-Level Parallelism (DLP) for which GPU architectures are designed. The proposed GPU instance takes advantage of Massively Parallel Processing (MPP) architectures performing well-balanced parallel calculations at the Degree-of-Freedom (DoF) level of finite elements. The numerical experiments evaluate and analyze the performance of diverse GPU instances of the matrix-free CG solver.

A study of graphics hardware accelerated particle swarm optimization with digital pheromones

Programmable Graphics Processing Units (GPUs) have lately become a promising means to perform scientific computations. Modern GPUs have proven to outperform the number of floating point operations when compared to traditional Central Processing Units (CPUs) through inherent data parallel architecture and higher bandwidth capabilities. They allow scientific computations to be performed without noticeable degradation in accuracy in a fraction of the time compared to traditional CPUs at substantially reduced costs, making them viable alternatives to expensive computer clusters or workstations. GPU programmability however, has fostered the development of a variety of programming languages making it challenging to select a computing language and use it consistently without the pitfall of being obsolete. Some GPU languages are hardware specific and are designed to rake in performance boosts when used with their host GPUs (e.g., Nvidia Cuda). Others are operating system specific (e.g., Microsoft HLSL). A few are platform agnostic lending themselves to be used on a workstation with any CPU and a GPU (e.g., GLSL, OpenCL). Of a number of companies and organizations that implement formal optimization into their processes, only a few utilize GPUs. It is either because the others are either vested much into CPU based computing or they are not fully aware of the benefits of implementing population based optimization routines in GPUs. Literature shows a large number of research publications specifically in the field of optimization utilizing GPUs. However, most of them are limited to a specific GPU hardware or addressed specific problems. The diversity in current GPU hardware and software APIs present overwhelming number of choices making it challenging to decide where and how to begin transitioning to GPU based computing, impeding promising computing avenues that relatively is very cost effective. In this paper, the authors precisely intend to address some of these issues by broadly classifying GPU APIs into three categories: 1) Hardware vendor dependent GPU APIs, 2) Graphical in context APIs, and 3) Platform agnostic APIs. Prior work by the authors demonstrated the capability of digital pheromones within Particle Swarm Optimization (PSO) for searching n-dimensional design spaces with improved accuracy, efficiency and reliability in serial and parallel CPU computing environments. To study the impact of GPUs, the authors have taken this digital pheromone variant of PSO and implemented it on three GPU APIs, each representing a category listed above, in a simplistic sense --- delegate unconstrained explicit objective function evaluations to GPUs. While this approach itself cannot be considered novel, the takeaways from implementing it on different GPU APIs provided a wealth of information that the authors believe can help optimization companies and organizations make informed decisions in implementing GPUs in their processes.