GPU-based Version Research Articles

DualSPHysics+: An enhanced DualSPHysics with improvements in accuracy, energy conservation and resolution of the continuity equation

This paper presents an enhanced version of the well-known SPH (Smoothed Particle Hydrodynamics) open-source code DualSPHysics for the simulation of free-surface fluid flows, leading to the DualSPHysics+ code. The enhancements are made through incorporation of several schemes with respect to stability, accuracy and energy/volume conservation issues in simulating incompressible free-surface fluid flows within the weakly compressible SPH formalism. The Optimized Particle Shifting (OPS) scheme is implemented to improve the accuracy of particle shifting vectors in the free-surface region. To mitigate energy dissipation and maintain consistency, the artificial viscosity in δ-SPH is substituted with a Riemann stabilization term, leading to the δR-SPH. The Velocity divergence Error Mitigating (VEM) and Volume Conservation Shifting (VCS) schemes are adopted in DualSPHysics+ to mitigate the velocity divergence error and improve the volume conservation, and hence to enhance the resolution of the continuity equation. To further reduce both the instantaneous and accumulated errors in velocity divergence, a Hyperbolic/Parabolic Divergence Cleaning (HPDC) scheme is incorporated in addition to the VEM scheme. The implementations of the introduced schemes on both CPU and GPU-based versions of the DualSPHysics+ code along with details on the compilation, running and computational performance are presented. Validations in terms of accuracy, energy conservation and convergence of DualSPHysics+ are shown via several relevant benchmarks. It is demonstrated that a better velocity divergence error cleaning in both instantaneous and accumulated errors can be achieved by the combination of VEM and HPDC. Meanwhile, the excessive energy dissipation by the artificial viscosity is shown to be suppressed by adopting the Riemann stabilization term. Enhanced resolution of the continuity equation along with improved energy conservation of DualSPHysics+ advance the SPH-based simulation of incompressible free-surface fluid flows.

MFFT: A GPU Accelerated Highly Efficient Mixed-Precision Large-Scale FFT Framework

Fast Fourier transform (FFT) is widely used in computing applications in large-scale parallel programs, and data communication is the main performance bottleneck of FFT and seriously affects its parallel efficiency. To tackle this problem, we propose a new large-scale FFT framework, MFFT, which optimizes parallel FFT with a new mixed-precision optimization technique, adopting the “high precision computation, low precision communication” strategy. To enable “low precision communication”, we propose a shared-exponent floating-point number compression technique, which reduces the volume of data communication, while maintaining higher accuracy. In addition, we apply a two-phase normalization technique to further reduce the round-off error. Based on the mixed-precision MFFT framework, we apply several optimization techniques to improve the performance, such as streaming of GPU kernels, MPI message combination, kernel optimization, and memory optimization. We evaluate MFFT on a system with 4,096 GPUs. The results show that shared-exponent MFFT is 1.23 × faster than that of double-precision MFFT on average, and double-precision MFFT achieves performance 3.53× and 9.48× on average higher than open source library 2Decomp&FFT (CPU-based version) and heFFTe (AMD GPU-based version), respectively. The parallel efficiency of double-precision MFFT increased from 53.2% to 78.1% compared with 2Decomp&FFT, and shared-exponent MFFT further increases the parallel efficiency to 83.8%.

Efficient parallel Monte-Carlo techniques for pricing American options including counterparty credit risk

In this article we mainly propose a numerical scheme, based on the novel Stochastic Grid Bundling Method (SGBM), to price American options in the presence of counterparty credit risk. More precisely, we consider the regression techniques (regress later) employed in the SGBM method and take advantage of the bundling structure to develop an efficient parallel strategy that is implemented on a GPU architecture. Also, a novel interpolation-based technique is efficiently applied in the XVA computation. Besides the advantages obtained in the sequential version, when compared with the more classical Least Squares Method, we show the relevant speedup of the parallel GPU-based version with respect to the sequential CPU-based one.

Efficient Flow Processing in 5G-Envisioned SDN-Based Internet of Vehicles Using GPUs

In the 5G-envisioned Internet of vehicles (IoV), a significant volume of data is exchanged through networks between intelligent transport systems (ITS) and clouds or fogs. With the introduction of Software-Defined Networking (SDN), the problems mentioned above are resolved by high-speed flow-based processing of data in network systems. To classify flows of packets in the SDN network, high throughput packet classification systems are needed. Although software packet classifiers are cheaper and more flexible than hardware classifiers, they could only deliver limited performance. A key idea to resolve this problem is parallelizing packet classification on graphical processing units (GPUs). In this paper, we study parallel forms of Tuple Space Search and Pruned Tuple Space Search algorithms for the flow classification suitable for GPUs using CUDA (Compute Unified Device Architecture). The key idea behind the offered methodology is to transfer the stream of packets from host memory to the global memory of the CUDA device, then assigning each of them to a classifier thread. To evaluate the proposed method, the GPU-based versions of the algorithms were implemented on two different CUDA devices, and two different CPU-based implementations of the algorithms were used as references. Experimental results showed that GPU computing enhances the performance of Pruned Tuple Space Search remarkably more than Tuple Space Search. Moreover, results evinced the computational efficiency of the proposed method for parallelizing packet classification algorithms.

A LIST-MODE OSEM-BASED ATTENUATION AND SCATTER COMPENSATION METHOD FOR SPECT.

Reliable attenuation and scatter compensation (ASC) is a pre-requisite for quantification and beneficial for visual interpretation tasks in SPECT. In this paper, we develop a reconstruction method that uses the entire SPECT emission data, i.e. data in both the photopeak and scatter windows, acquired in list-mode format and including the energy attribute of the detected photon, to perform ASC. We implemented a GPU-based version of this method using an ordered subsets expectation maximization (OSEM) algorithm. The method was objectively evaluated using realistic simulation studies on the task of estimating uptake in the striatal regions of the brain in a 2-D dopamine transporter (DaT)-scan SPECT study. We observed that inclusion of data from the scatter window and using list-mode data yielded improved quantification compared to using data only from the photopeak window or using binned data. These results motivate further development of list-mode-based ASC methods that include scatter-window data for SPECT.

GPU-based implementation of a real-time model for atmospheric dispersion of radionuclides

This work presents the latest development related to the improvement of the atmospheric dispersion system (ADS) used in Central Nuclear Almirante Alvaro Alberto (CNAAA) nuclear power plant (NPP). The ADS upgrade includes a refinement of spatial resolution at the expense of computational time. To overcome such difficulty, the use of parallel computing (PC) has been proposed. In the considered ADS, the physical calculations are performed by 4 main modules: the source term, the wind field, the plume dispersion and the plume projection modules. As the source term module is not affected by the proposed refinements, it has not been focused until this moment. On the other hand, the remaining modules are strongly affected by the spatial resolution. In the first stage of the project, a GPU-based version of the wind field module was developed using the Compute Unified Device Architecture (CUDA), which was 30 times faster than the sequential version. Motivated by such promising results, the present work focuses in the development and evaluation of the GPU-based (also using CUDA) version of the plume dispersion module, which main feature is to calculate transport and diffusion of radionuclide in the atmosphere by using a three-dimensional puff model with Lagrangian trajectory and Gaussian diffusion. As a result, the execution time of the most refined simulation decreased from 2499 s (running on an Intel-Core I5 7500 CPU) to 68 s (running on a GTX-1070 GPU), meaning a considerable speedup of about 36 times. Here, the most important issues of the parallel implementation, as well as comparative results, are presented and discussed.

Efficient GPU-based Monte-Carlo simulation of diffusion in real astrocytes reconstructed from confocal microscopy

The primary goal of this work is to develop an efficient Monte-Carlo simulation of diffusion-weighted signal in complex cellular structures, such as astrocytes, directly derived from confocal microscopy. In this study, we first use an octree structure for spatial decomposition of surface meshes. Octree structure and radius-search algorithm help to quickly identify the faces that particles can possibly encounter during the next time step, thus speeding up the Monte-Carlo simulation. Furthermore, we propose to use a three-dimensional binary marker to describe the complex cellular structure and optimize the particle trajectory simulation. Finally, a GPU-based version of these two approaches is implemented for more efficient modeling. It is shown that the GPU-based binary marker approach yields unparalleled performance, opening up new possibilities to better understand intracellular diffusion, validate diffusion models, and create dictionaries of intracellular diffusion signatures.

A new theoretical derivation of NFFT and its implementation on GPU

Abstract An efficient calculation of NFFT (nonequispaced fast Fourier transforms) is always a challenging task in a variety of application areas, from medical imaging to radio astronomy to chemical simulation. In this article, a new theoretical derivation is proposed for NFFT based on gridding algorithm and new strategies are proposed for the implementation of both forward NFFT and its inverse on both CPU and GPU. The GPU-based version, namely CUNFFT, adopts CUDA (Compute Unified Device Architecture) technology, which supports a fine-grained parallel computing scheme. The approximation errors introduced in the algorithm are discussed with respect to different window functions. Finally, benchmark calculations are executed to illustrate the accuracy and performance of NFFT and CUNFFT. The results show that CUNFFT is not only with high accuracy, but also substantially faster than conventional NFFT on CPU.

GPU-enabled back-propagation artificial neural network for digit recognition in parallel

In this paper, we show that the GPU (graphics processing unit) can be used not only for processing graphics, but also for high speed computing. We provide a comparison between the times taken on the CPU and GPU to perform the training and testing of a back-propagation artificial neural network. We implemented two neural networks for recognizing handwritten digits; one consists of serial code executed on the CPU, while the other is a GPU-based version of the same system which executes in parallel. As an experiment for performance evaluation, a system for neural network training on the GPU is developed to reduce training time. The programming environment that the system is based on is CUDA which stands for compute unified device architecture, which allows a programmer to write code that will run on an NVIDIA GPU card. Our results over an experiment of digital image recognition using neural network confirm the speed-up advantages by tapping on the resources of GPU. Our proposed model has an advantage of simplicity, while it shows on par performance with the state-of-the-arts algorithms.

Acceleration of ensemble machine learning methods using many-core devices

We present a case study into the acceleration of ensemble machine learning methods using many-core devices in collaboration with Toshiba Medical Visualisation Systems Europe (TMVSE). The adoption of GPUs to execute a key algorithm in the classification of medical image data was shown to significantly reduce overall processing time. Using a representative dataset and pre-trained decision trees as input we will demonstrate how the decision forest classification method can be mapped onto the GPU data processing model. It was found that a GPU-based version of the decision forest method resulted in over 138 times speed-up over a single-threaded CPU implementation with further improvements possible. The same GPU-based software was then directly applied to a suitably formed dataset to benefit supervised learning techniques applied in High Energy Physics (HEP) with similar improvements in performance.

Bit-parallel approximate pattern matching: Kepler GPU versus Xeon Phi

Advanced SIMD features on GPUs and Xeon Phis promote efficient long pattern search.A tiled approach to accelerating the Wu-Manber algorithm on GPUs has been proposed.Both the GPU and Xeon Phi yield two orders-of-magnitude speedup over one CPU core.The GPU-based version with tiling runs up to 2.9 × faster than the Xeon Phi version. Approximate pattern matching (APM) targets to find the occurrences of a pattern inside a subject text allowing a limited number of errors. It has been widely used in many application areas such as bioinformatics and information retrieval. Bit-parallel APM takes advantage of the intrinsic parallelism of bitwise operations inside a machine word. This approach typically encodes non-deterministic finite automaton (NFA) states or value differences between adjacent cells of a dynamic programming matrix in the form of bit arrays. Wu-Manber (WM) is a well-known bit-parallel APM algorithm, which simulates an NFA and gains parallel efficiency by performing multiple state updates within a machine word. An important parameter is the machine word size (e.g. 32 or 64?bits for CPUs). Due to increasing vector capabilities, efficient mapping of bit-parallel APM algorithms onto modern high performance computing architectures is an interesting research topic. Prominent examples are Xeon Phi coprocessors and CUDA-enabled GPUs, which provide words of size 512?bits (by means of vector registers) and 1024?bits (by means of warps), respectively. In this paper, we investigate mappings of the WM algorithm onto these two accelerator types. Both architectures are able to achieve around two orders-of-magnitude speedups compared to a single-threaded CPU implementation. Moreover, our tile-based implementation on a GeForce Titan graphics card runs up to 2.9 × faster than our implementation on an Intel Xeon Phi 5110P. Source code is available at http://xbitpar.sourceforge.net.

An improved GPU-oriented algorithm for elastostatic analysis with boundary element method

A parallel, boundary element method procedure for elastostatic problems is here illustrated. Starting from a reference collocation–integration algorithm, a preliminary study, mainly devoted to the performance analysis of the method has been first carried out. Subsequently, three main areas that offer potential improvements have been detected. Several enhanced versions of the algorithm, incorporating the suggested modifications, have been tested and numerically characterized on test-cases with varying boundary conditions and mesh density. Finally, a parallel, high-efficiency, GPU-based version, able to process large domains fully featured with internal solutions has been developed and discussed in terms of cost/benefit ratio and overall performance.

Multi-GPU based on multicriteria optimization for motion estimation system

Graphics processor units (GPUs) offer high performance and power efficiency for a large number of data-parallel applications. Previous research has shown that a GPU-based version of a neuromorphic motion estimation algorithm can achieve a ×32 speedup using these devices. However, the memory consumption creates a bottleneck due to the expansive tree of signal processing operations performed. In the present contribution, an improvement in memory reduction was carried out, which limited accelerator viability usage. An evolutionary algorithm was used to find the best configuration. It supposes a trade-off solution between consumption resources, parallel efficiency, and accuracy. A multilevel parallel scheme was exploited: grain level by means of multi-GPU systems, and a finer level by data parallelism. In order to achieve a more relevant analysis, some optical flow benchmarks were used to validate this study. Satisfactory results opened the chance of building an intelligent motion estimation system that auto-adapted according to real-time, resource consumption, and accuracy requirements.

GPU-based approaches for real-time sound source localization using the SRP-PHAT algorithm

The aim of most microphone array applications is to localize sound sources in a noisy and reverberant environment. For that purpose, many different sound source localization (SSL) algorithms have been proposed, where the SRP-PHAT (steered response power using the phase transform) has been known as one of the state-of-the-art methods. Its original formulation allows two different practical implementations, one that is computed in the frequency domain (FDSP) and another in the time domain (TDSP), which can be enhanced by interpolation. However, the main problem of this algorithm is its high computational cost due to intensive grid scan in search for the sound source. Considering the power of graphics processing units (GPUs) for working with massively parallelizable compute-intensive algorithms, we present two highly scalable GPU-based versions of the SRP-PHAT, one for each formulation, and also an implementation of the cubic splines interpolation in the GPU. These approaches exploit the parallel aspects of the SRP-PHAT, allowing real-time execution for large search grids. Comparing our GPU approaches against traditional multithreaded CPU approaches, results show a speed up of [Formula: see text] for the FDSP, and [Formula: see text] for the TDSP with interpolation, when comparing high-end GPUs with high-end CPUs.

DecGPU: distributed error correction on massively parallel graphics processing units using CUDA and MPI

BackgroundNext-generation sequencing technologies have led to the high-throughput production of sequence data (reads) at low cost. However, these reads are significantly shorter and more error-prone than conventional Sanger shotgun reads. This poses a challenge for the de novo assembly in terms of assembly quality and scalability for large-scale short read datasets.ResultsWe present DecGPU, the first parallel and distributed error correction algorithm for high-throughput short reads (HTSRs) using a hybrid combination of CUDA and MPI parallel programming models. DecGPU provides CPU-based and GPU-based versions, where the CPU-based version employs coarse-grained and fine-grained parallelism using the MPI and OpenMP parallel programming models, and the GPU-based version takes advantage of the CUDA and MPI parallel programming models and employs a hybrid CPU+GPU computing model to maximize the performance by overlapping the CPU and GPU computation. The distributed feature of our algorithm makes it feasible and flexible for the error correction of large-scale HTSR datasets. Using simulated and real datasets, our algorithm demonstrates superior performance, in terms of error correction quality and execution speed, to the existing error correction algorithms. Furthermore, when combined with Velvet and ABySS, the resulting DecGPU-Velvet and DecGPU-ABySS assemblers demonstrate the potential of our algorithm to improve de novo assembly quality for de-Bruijn-graph-based assemblers.ConclusionsDecGPU is publicly available open-source software, written in CUDA C++ and MPI. The experimental results suggest that DecGPU is an effective and feasible error correction algorithm to tackle the flood of short reads produced by next-generation sequencing technologies.

GPU-accelerated Monte Carlo convolution/superposition implementation for dose calculation.

Dose calculation is a key component in radiation treatment planning systems. Its performance and accuracy are crucial to the quality of treatment plans as emerging advanced radiation therapy technologies are exerting ever tighter constraints on dose calculation. A common practice is to choose either a deterministic method such as the convolution/superposition (CS) method for speed or a Monte Carlo (MC) method for accuracy. The goal of this work is to boost the performance of a hybrid Monte Carlo convolution/superposition (MCCS) method by devising a graphics processing unit (GPU) implementation so as to make the method practical for day-to-day usage. Although the MCCS algorithm combines the merits of MC fluence generation and CS fluence transport, it is still not fast enough to be used as a day-to-day planning tool. To alleviate the speed issue of MC algorithms, the authors adopted MCCS as their target method and implemented a GPU-based version. In order to fully utilize the GPU computing power, the MCCS algorithm is modified to match the GPU hardware architecture. The performance of the authors' GPU-based implementation on an Nvidia GTX260 card is compared to a multithreaded software implementation on a quad-core system. A speedup in the range of 6.7-11.4x is observed for the clinical cases used. The less than 2% statistical fluctuation also indicates that the accuracy of the authors' GPU-based implementation is in good agreement with the results from the quad-core CPU implementation. This work shows that GPU is a feasible and cost-efficient solution compared to other alternatives such as using cluster machines or field-programmable gate arrays for satisfying the increasing demands on computation speed and accuracy of dose calculation. But there are also inherent limitations of using GPU for accelerating MC-type applications, which are also analyzed in detail in this article.