Use Of Graphics Processing Units Research Articles

Analysis of impact of general-purpose graphics processor units in supersonic flow modeling

Computational methods are widely used in prediction of complex flowfields associated with off-normal situations in aerospace engineering. Modern graphics processing units (GPU) provide architectures and new programming models that enable to harness their large processing power and to design computational fluid dynamics (CFD) simulations at both high performance and low cost. Possibilities of the use of GPUs for the simulation of external and internal flows on unstructured meshes are discussed. The finite volume method is applied to solve three-dimensional unsteady compressible Euler and Navier–Stokes equations on unstructured meshes with high resolution numerical schemes. CUDA technology is used for programming implementation of parallel computational algorithms. Solutions of some benchmark test cases on GPUs are reported, and the results computed are compared with experimental and computational data. Approaches to optimization of the CFD code related to the use of different types of memory are considered. Speedup of solution on GPUs with respect to the solution on central processor unit (CPU) is compared. Performance measurements show that numerical schemes developed achieve 20–50 speedup on GPU hardware compared to CPU reference implementation. The results obtained provide promising perspective for designing a GPU-based software framework for applications in CFD.

Real-time tissue offset correction system for intravital multiphoton microscopy

The development of multi-photon intravital microscopy, in particular two-photon microscopy (2PM), has been a breakthrough technique for deep-tissue imaging of dynamic cell behavior inside live organisms and has substantially advanced the field of immunology. However, intravital time-lapse imaging over prolonged time periods is complicated by slow tissue drifts caused by vital activity, leading to shifting fields of views and making the acquired image sequence partially or completely unanalyzable. To solve this issue, we have established a system that performs continuous drift offset correction in real time using fine pattern matching during 2PM acquisition. We incorporated an extensive use of graphical processing unit (GPU) for high-speed computing required for real time correction during data acquisition. This allowed us to perform prolonged acquisitions and increase the proportion of analyzable datasets to nearly 100% in lymphoid and non-lymphoid tissues. Considering the straightforward implementation of our newly developed system, we anticipate that it will be applicable for other users interested in improving the quality of live imaging data acquisition.

GPU-SME-kNN: Scalable and memory efficient kNN and lazy learning using GPUs

The k nearest neighbor (kNN) rule is one of the most used techniques in data mining and pattern recognition due to its simplicity and low identification error. However, the computational effort it requires is directly related to the dataset sizes, hence delivering a poor performance on large datasets. The use of graphics processing units (GPU) has improved the run-time performance of the kNN rule but the computational requirements of current approaches limit this performance as the dataset size increases.In this paper, we propose a new scalable and memory efficient design for a GPU-based kNN rule, called GPU-SME-kNN, that breaks the dependency between dataset size and memory footprint while delivering high performance. An experimental study of GPU-SME-kNN is presented showing a high performance, even in cases that other methods cannot address, while the computational requirements are suitable for most commercial GPU devices. Our design has also been applied to kNN-based lazy learning algorithms reducing run-times in a significant way.

Fast Hamiltonian Monte Carlo Using GPU Computing

In recent years, the Hamiltonian Monte Carlo (HMC) algorithm has been found to work more efficiently compared to other popular Markov chain Monte Carlo (MCMC) methods (such as random walk Metropolis–Hastings) in generating samples from a high-dimensional probability distribution. HMC has proven more efficient in terms of mixing rates and effective sample size than previous MCMC techniques, but still may not be sufficiently fast for particularly large problems. The use of GPUs promises to push HMC even further greatly increasing the utility of the algorithm. By expressing the computationally intensive portions of HMC (the evaluations of the probability kernel and its gradient) in terms of linear or element-wise operations, HMC can be made highly amenable to the use of graphics processing units (GPUs). A multinomial regression example demonstrates the promise of GPU-based HMC sampling. Using GPU-based memory objects to perform the entire HMC simulation, most of the latency penalties associated with transferring data from main to GPU memory can be avoided. Thus, the proposed computational framework may appear conceptually very simple, but has the potential to be applied to a wide class of hierarchical models relying on HMC sampling. Models whose posterior density and corresponding gradients can be reduced to linear or element-wise operations are amenable to significant speed ups through the use of GPUs. Analyses of datasets that were previously intractable for fully Bayesian approaches due to the prohibitively high computational cost are now feasible using the proposed framework.

An efficient GPU implementation for a faster simulation of unsteady bed-load transport

ABSTRACTComputational tools may help engineers in the assessment of sediment transport during the decision-making processes. The main requirements are that the numerical results have to be accurate and simulation models must be fast. The present work is based on the 2D shallow water equations in combination with the 2D Exner equation. The resulting numerical model accuracy was already discussed in previous work. Regarding the speed of the computation, the Exner equation slows down the already costly 2D shallow water model as the number of variables to solve is increased and the numerical stability is more restrictive. In order to reduce the computational effort required for simulating realistic scenarios, the authors have exploited the use of Graphics Processing Units in combination with non-trivial optimization procedures. The gain in computing cost obtained with the graphic hardware is compared against single-core (sequential) and multi-core (parallel) CPU implementations in two unsteady cases.

Dispersive Möbius Transform Finite-Element Time-Domain Method on Graphics Processing Units

A novel use of graphics processing units (GPUs) is presented in the execution of the dispersive finite-element time-domain (FETD) method, based upon the Mobius (bilinear) ${z}$ -transform technique. By utilizing the immense computational power of modern GPUs via NVIDIA’s compute unified device architecture (CUDA) language, a narrowing of the performance gap, which currently exists between dispersive FETD methods and their non-dispersive counterparts, can be achieved, thus facilitating the study of a wider range of physical phenomena. An analysis of the $ {z}$ -transform dispersive FETD algorithm is presented in order to both identify dispersive overhead bottlenecks and determine its suitability to parallelization. Numerical studies are then undertaken to measure the performance increase as a function of simulation parameters, such as number of variables, the amount of dispersive material present, and floating point precision.

Speedup of Magnetic-Electric Matrices Assembly Computation by Means of a Multi-GPUs Environment

This paper deals with an accelerated implementation of the assembly of the matrices accounting for the magnetic-electric interactions arising from an integral formulation of the eddy current problem. The use of integral formulations leads to fully populated matrices whose assembly requires a computational effort of $O(N^{2})$ , $N$ being the number of degrees of freedoms related to the finite-element discretization. Although the inversion procedure is $O(N^{3})$ for a direct method, the assembly for medium size problems can be very time consuming. In this paper, we prove that a significant speedup can be achieved by means of an optimized ad hoc use of graphical processing units. The price to be paid is a challenging implementation if compared with the traditional parallel systems (CPUs based). Two kinds of applications are considered: one in the framework of non-destructive testing and the other in the field of plasma fusion devices modeling.

Parallel global optimization on GPU

This work considers a parallel algorithm for solving multidimensional multiextremal optimization problems. This algorithm uses Peano-type space filling curves for dimension reduction. Conditions of non-redundant parallelization of the algorithm are considered. Efficiency of the algorithm on modern computing systems with the use of graphics processing units (GPUs) is investigated. Speedup of the algorithm using GPU as compared with the same algorithm implemented on CPU only is demonstrated experimentally. Computational experiments are carried out on a series of several hundred multidimensional multiextremal problems.

Parallel flow accumulation algorithms for graphical processing units with application to RUSLE model

Digital elevation models (DEMs) are widely used in the modeling of surface hydrology, which typically includes the determination of flow directions and flow accumulation. The use of high-resolution DEMs increases the accuracy of flow accumulation computation, but as a drawback, the computational time may become excessively long if large areas are analyzed. In this paper we investigate the use of graphical processing units (GPUs) for efficient flow accumulation calculations. We present two new parallel flow accumulation algorithms based on dependency transfer and topological sorting and compare them to previously published flow transfer and indegree-based algorithms. We benchmark the GPU implementations against industry standards, ArcGIS and SAGA. With the flow-transfer D8 flow routing model and binary input data, a speed up of 19 is achieved compared to ArcGIS and 15 compared to SAGA. We show that on GPUs the topological sort-based flow accumulation algorithm leads on average to a speedup by a factor of 7 over the flow-transfer algorithm. Thus a total speed up of the order of 100 is achieved. We test the algorithms by applying them to the Revised Universal Soil Loss Equation (RUSLE) erosion model. For this purpose we present parallel versions of the slope, LS factor and RUSLE algorithms and show that the RUSLE erosion results for an area of 12km x 24km containing 72 million cells can be calculated in less than a second. Since flow accumulation is needed in many hydrological models, the developed algorithms may find use in many other applications than RUSLE modeling. The algorithm based on topological sorting is particularly promising for dynamic hydrological models where flow accumulations are repeatedly computed over an unchanged DEM.

A GPU parallel optimised blockwise NLM algorithm in a distributed computing system

Recently, advanced computing systems are widely adopted in order to intensively elaborate a huge amount of biomedical data in the e-health field. An interesting challenge is to perform real-time diagnosis by means of complex computational environments. In this paper, we suggest to deal the most computationally expensive processing steps of a distributed cloud e-health system by the use of graphics processing units (GPUs). In the case study of the magnetic resonance imaging (MRI), for improving the quality of denoising and helping the real-time diagnosis, we have implemented a GPU parallel algorithm based on the optimised blockwise non-local means (OB-NLM) method. Experimental results have shown a significant improvement of healthcare processing practice in terms of execution time.

How General-Purpose can a GPU be?

The use of graphics processing units (GPUs) in general-purpose computation (GPGPU) is a growing field. GPU instruction sets, while implementing a graphics pipeline, draw from a range of single instruction multiple datastream (SIMD) architectures characteristic of the heyday of supercomputers. Yet only one of these SIMD instruction sets has been of application on a wide enough range of problems to survive the era when the full range of supercomputer design variants was being explored: vector instructions.
 
 This paper proposes a reconceptualization of the GPU as a multicore design with minimal exotic modes of parallelism so as to make GPGPU truly general.

More flexibility for code generation with GeNN v2.1

Background GeNN (GPU enhanced Neuronal Networks) [1,2] is a software framework that was designed to facilitate the use of GPUs (Graphics Processing Units) for the simulation of spiking neuronal networks. It is built on top of the CUDA (Common Unified Device Architecture) [3] application programming interface provided by NVIDIA Corporation and is entirely based on code generation: Users provide a compact description of a spiking neuronal network model and GeNN generates CUDA and C++ code to simulate it, also taking into account the specifics of the GPU hardware detected at compile time.

Computational Fluid Dynamics Computations Using a Preconditioned Krylov Solver on Graphical Processing Units

Graphical processing unit (GPU) computation in recent years has seen extensive growth due to advancement in both hardware and software stack. This has led to increase in the use of GPUs as accelerators across a broad spectrum of applications. This work deals with the use of general purpose GPUs for performing computational fluid dynamics (CFD) computations. The paper discusses strategies and findings on porting a large multifunctional CFD code to the GPU architecture. Within this framework, the most compute intensive segment of the software, the BiCGStab linear solver using additive Schwarz block preconditioners with point Jacobi iterative smoothing is optimized for the GPU platform using various techniques in CUDA Fortran. Representative turbulent channel and pipe flow are investigated for validation and benchmarking purposes. Both single and double precision calculations are highlighted. For a modest single block grid of 64 × 64 × 64, the turbulent channel flow computations showed a speedup of about eightfold in double precision and more than 13-fold for single precision on the NVIDIA Tesla GPU over a serial run on an Intel central processing unit (CPU). For the pipe flow consisting of 1.78 × 106 grid cells distributed over 36 mesh blocks, the gains were more modest at 4.5 and 6.5 for double and single precision, respectively.

Use of graphics processing units for efficient evaluation of derivatives of exchange integrals by means of Fourier transform of the 1/r operator and its numerical quadrature

In this paper, we propose an efficient way for evaluation of derivatives of exchange integrals. We propose an approach in which we factorize the non-local exchange kernel into a sum of separable terms. We exploit a discretized Fourier transform for the 1/r operator, and we devise a method that allows us to employ a manageable number of plane-wave functions in the Fourier expansion while still keeping necessary accuracy. Resulting formulas are amenable for efficient evaluation on graphics processing units (GPU). We discuss the GPU implementation for derivatives of two-electron repulsion integrals of the (gk|gk) type in the hybrid Gaussian and plane-wave basis. Derivatives of such integrals are needed for computation of cross sections in vibrationally inelastic electron scattering by polyatomic molecules. Speedup and accuracy achieved are demonstrated for cross sections of selected vibrational modes of cyclopropane, benzene and adamantane. The proposed factorization method is general and may be applied to any type of exchange integrals. We note briefly on its possible application to exchange integrals and their derivatives in quantum chemical computational methods.

A GPU based compressible multiphase hydrocode for modelling violent hydrodynamic impact problems

This paper presents a GPU based compressible multiphase hydrocode for modelling violent hydrodynamic impacts under harsh conditions such as slamming and underwater explosion. An effort is made to extend a one-dimensional five-equation reduced model (Kapila et al., 2001) to compute three-dimensional hydrodynamic impact problems on modern graphics hardware. In order to deal with free-surface problems such as water waves, gravitational terms, which are initially absent from the original model, are now considered and included in the governing equations. A third-order finite volume based MUSCL scheme is applied to discretise the integral form of the governing equations. The numerical flux across a mesh cell face is estimated by means of the HLLC approximate Riemann solver. The serial CPU program is firstly parallelised on multi-core CPUs with the OpenMP programming model and then further accelerated on many-core graphics processing units (GPUs) using the CUDA C programming language. To balance memory usage, computing efficiency and accuracy on multi- and many-core processors, a mixture of single and double precision floating-point operations is implemented. The most important data like conservative flow variables are handled with double-precision dynamic arrays, whilst all the other variables/arrays like fluxes, residual and source terms are treated in single precision. Several benchmark test cases including water-air shock tubes, one-dimensional liquid cavitation tube, dam break, 2D cylindrical underwater explosion near a planar rigid wall, 3D spherical explosion in a rigid cylindrical container and water entry of a 3D rigid flat plate have been calculated using the present approach. The obtained results agree well with experiments, exact solutions and other independent numerical computations. This demonstrates the capability of the present approach to deal with not only violent free-surface impact problems but also hull cavitation associated with underwater explosions. Performance analysis reveals that the running time cost of numerical simulations is dramatically reduced by use of GPUs with much less consumption of electrical energy than on the CPU.

Quantum Chemistry for Solvated Molecules on Graphical Processing Units Using Polarizable Continuum Models.

The conductor-like polarization model (C-PCM) with switching/Gaussian smooth discretization is a widely used implicit solvation model in chemical simulations. However, its application in quantum mechanical calculations of large-scale biomolecular systems can be limited by computational expense of both the gas phase electronic structure and the solvation interaction. We have previously used graphical processing units (GPUs) to accelerate the first of these steps. Here, we extend the use of GPUs to accelerate electronic structure calculations including C-PCM solvation. Implementation on the GPU leads to significant acceleration of the generation of the required integrals for C-PCM. We further propose two strategies to improve the solution of the required linear equations: a dynamic convergence threshold and a randomized block-Jacobi preconditioner. These strategies are not specific to GPUs and are expected to be beneficial for both CPU and GPU implementations. We benchmark the performance of the new implementation using over 20 small proteins in solvent environment. Using a single GPU, our method evaluates the C-PCM related integrals and their derivatives more than 10× faster than that with a conventional CPU-based implementation. Our improvements to the linear solver provide a further 3× acceleration. The overall calculations including C-PCM solvation require, typically, 20-40% more effort than that for their gas phase counterparts for a moderate basis set and molecule surface discretization level. The relative cost of the C-PCM solvation correction decreases as the basis sets and/or cavity radii increase. Therefore, description of solvation with this model should be routine. We also discuss applications to the study of the conformational landscape of an amyloid fibril.

An efficient GPU implementation for locating micro-seismic sources using 3D elastic wave time-reversal imaging

Locating micro-seismic events is of utmost importance in seismic exploration, especially when searching for unconventional oil and gas resources. The arrival-time-difference approach is the dominant source location method currently used in the field of micro-seismic source location. However, micro-seismic events can be generated by any arbitrary rock movement and are often accompanied by interference noise. Recordings show characteristics of complicated wavelets and low signal-to-noise ratios. Under such conditions, conventional triangulation methods may have difficulty producing reliable locations; time-reversal imaging micro-seismic event location techniques are more promising. Locating micro-seismic events must be performed on-site for real-time monitoring of hydraulic fracturing. Introducing wave equation imaging techniques when locating micro-seismic events will increase the computation time, thus complicating real-time site monitoring. Therefore, the use of graphics processing unit (GPU) devices to accelerate time-reversal imaging micro-seismic event location technology becomes imperative. Three-dimensional synthetic data examples have demonstrated that the GPU-based time-reversal imaging micro-seismic event location method is typically 18 times faster than the central processing unit (CPU)-based implementation. The performance boost afforded by the GPU architecture allows us to locate micro-seismic events in 3D at a lower hardware cost and in less time than has been previously possible.

NDPA: A generalized efficient parallel in-place N-Dimensional Permutation Algorithm

N-dimensional transpose/permutation is a very important operation in many large-scale data intensive and scientific applications. These applications include but not limited to oil industry i.e. seismic data processing, nuclear medicine, media production, digital signal processing and business intelligence. This paper proposes an efficient in-place N-dimensional permutation algorithm. The algorithm is based on a novel 3D transpose algorithm that was published recently. The proposed algorithm has been tested on 3D, 4D, 5D, 6D and 7D data sets as a proof of concept. This is the first contribution which is breaking the dimensions’ limitation of the base algorithm. The suggested algorithm exploits the idea of mixing both logical and physical permutations together. In the logical permutation, the address map is transposed for each data unit access. In the physical permutation, actual data elements are swapped. Both permutation levels exploit the fast on-chip memory bandwidth by transferring large amount of data and allowing for fine-grain SIMD (Single Instruction, Multiple Data) operations. Thus, the performance is improved as evident from the experimental results section. The algorithm is implemented on NVidia GeForce GTS 250 GPU (Graphics Processing Unit) containing 128 cores. The rapid increase in GPUs performance coupled with the recent and continuous improvements in its programmability proved that GPUs are the right choice for computationally demanding tasks. The use of GPUs is the second contribution which reflects how strongly they fit for high performance tasks. The third contribution is improving the proposed algorithm performance to its peak as discussed in the results section.

Use of graphics processing units for automatic synthesis of programs

Genetic programming (GP) is an evolutionary method that allows computers to solve problems automatically. However, the computational power required for the evaluation of billions of programs imposes a serious limitation on the problem size. This work focuses on accelerating GP to support the synthesis of large problems. This is done by completely exploiting the highly parallel environment of graphics processing units (GPUs). Here, we propose a new quantum-inspired linear GP approach that implements all the GP steps in the GPU and provides the following: (1) significant performance improvements in the GP steps, (2) elimination of the overhead of copying the fitness results from the GPU to the CPU, and (3) incorporation of a new selection mechanism to recognize the programs with the best evaluations. The proposed approach outperforms the previous approach for large-scale synthetic and real-world problems. Further, it provides a remarkable speedup over the CPU execution.

PSkel: A stencil programming framework for CPU‐GPU systems

SummaryThe use of Graphics Processing Units (GPUs) for high‐performance computing has gained growing momentum in recent years. Unfortunately, GPU‐programming platforms like Compute Unified Device Architecture (CUDA) are complex, user unfriendly, and increase the complexity of developing high‐performance parallel applications. In addition, runtime systems that execute those applications often fail to fully utilize the parallelism of modern CPU‐GPU systems. Typically, parallel kernels run entirely on the most powerful device available, leaving other devices idle. These observations sparked research in two directions: (1) high‐level approaches to software development for GPUs, which strike a balance between performance and ease of programming; and (2) task partitioning to fully utilize the available devices. In this paper, we propose a framework, called PSkel, that provides a single high‐level abstraction for stencil programming on heterogeneous CPU‐GPU systems, while allowing the programmer to partition and assign data and computation to both CPU and GPU. Our current implementation uses parallel skeletons to transparently leverage Intel Threading Building Blocks (Intel Corporation, Santa Clara, CA, USA) and NVIDIA CUDA (Nvidia Corporation, Santa Clara, CA, USA). In our experiments, we observed that parallel applications with task partitioning can improve average performance by up to 76% and 28% compared with CPU‐only and GPU‐only parallel applications, respectively. Copyright © 2015 John Wiley & Sons, Ltd.