Compute Unified Device Architecture Threads Research Articles

Розпаралелення алгоритму моделювання поведінки рідини в реальному часі

A parallel algorithm based on Compute Unified Device Architecture (CUDA) technology is proposed to accelerate fluid behavior simulation and real-time decision making capability. Three main steps were highlighted: implementation of the fluid flow simulation method, distri-bution of work between CUDA threads, and collection of results. A software product was de-veloped to analyze the obtained results. As a result, it was found that the minimum acceptable refresh rate of the simulation environment is achieved on an environment with a size of 512  512 and is an average of 51.54 FPS (number of frames per second) for both states (quiet and ac-tive simulation). An analysis of literary sources was carried out, where the current state of this scientific problem is outlined and the advantages of the proposed approach are indicated. Among the simulation methods, the method using the Navier―Strokes equation for the flow of incompressible matter was chosen because it is simple and has good possibilities for parallelization.

Multi-swarm particle swarm optimization based on CUDA for sparse reconstruction

Particle swarm optimization (PSO) has been successfully applied to the sparse reconstruction problem and achieved good results. With the dimension of the problem increases, parallelizing PSO is an effective method to reduce its running time. This paper proposes a parallel PSO framework to solve the sparse reconstruction problem based on Compute Unified Device Architecture (CUDA) platform on Graphics Processing Unit (GPU). In order to further utilize potential computing resources in the GPU and improve the performance of the algorithm, each particle is launched by CUDA threads and the swarm is divided into multiple sub-swarms in CUDA streams. A local search strategy based on gradient and a particle coding strategy is combined into PSO for the purposes of achieving better reconstruction accuracy and accelerating convergence. In addition, in order to further optimize the parallel execution process of CUDA, the reduction algorithm and dynamic parallelism are incorporated into the proposed method. In the performance experiments, the proposed algorithm achieves a maximum speedup ratio of 25 times compared to the serial version in the signal reconstruction tasks.

Real‐time multiview human pose tracking using graphics processing unit‐accelerated particle swarm optimization

SummaryThis paper describes how to achieve real‐time tracking of 3D human motion using multiview images and graphics processing unit (GPU)‐accelerated particle swarm optimization. The tracking involves configuring the 3D human model in the pose described by each particle and then rasterizing it in each 2D plane. The Compute Unified Device Architecture threads rasterize the columns of the triangles and perform the summing of the fitness values of pixels belonging to the processed columns. Such a parallel particle swarm optimization (PSO) exhibits the level of parallelism that allows us to effectively utilize the GPU resources. Image acquisition and image processing are multithreaded and run on CPU in parallel with PSO‐based searching for the best pose. Owing to such task decomposition, the tracking of the full human body can be performed at rates of 12 frames per second. For a PSO consisting of 1000 particles and executing 10 iterations, the GPU achieves an average speedup of 12 over the CPU. Using marker‐less motion capture system consisting of four calibrated and synchronized cameras, the efficiency comparisons were conducted on four CPU cores and four GTX GPUs on two cards. Copyright © 2014 John Wiley & Sons, Ltd.

PAR Model SAR Image Interpolation Algorithm on GPU with CUDA

ABSTRACTResolution of synthetic aperture radar (SAR) images is limited by imaging hardware. To improve SAR image resolution, interpolation needs to be carried out on a SAR image. An image interpolation algorithm with a good effect incurs a high computational cost and cannot be used in a real-time scenario. This paper applies Compute Unified Device Architecture (CUDA) to speed up the piece-wise autoregressive model image interpolation algorithm for acquiring high-quality SAR images. A partial differential is used to carry out image de-noising. After de-noising, divide the image interpolation based on a local window of a piece-wise autoregressive model: the SAR image is first divided into many 9 × 9 small local windows. For each window, a CUDA thread is launched to interpolate using the autoregressive interpolation algorithm. Gradient descent algorithm is used to estimate parameters of the autoregressive model in the first- and second-round interpolation. Numerical simulation indicates that this GPU-based parallel algorithm can interpolate a 2592 × 1944 image within 1/110th of the time used by a CPU-based serial algorithm. Moreover, the computation time saved increases with the image size. The experimental results show that the method in this paper can achieve high-quality image interpolation in a low computation time.

Accelerating experimental high-order spatial statistics calculations using GPUs

High-order spatial statistics have been widely used to describe the spatial phenomena in the field of geology science. Spatial statistics are subject to extremely heavy computational burden for large geostatistical models. To improve the computational efficiency, a parallel approach based on GPU (Graphics Processing Unit) is proposed for the calculation of high-order spatial statistics. The parallel scheme is achieved by utilizing a two-stage method to calculate the replicate of a moment for a given template simultaneously termed as the node-stage parallelism, and transform the spatial moments to cumulants for all lags of a template simultaneously termed as the template-stage parallelism. Also, a series of optimization strategies are proposed to take full advantage of the computational capabilities of GPUs, including the appropriate task allocation to the CUDA (Compute Unified Device Architecture) threads, proper organization of the GPU physical memory, and optimal improvement of the existed parallel routines. Tests are carried out on two training images to compare the performance of the GPU-based method with that of the serial implementation. Error analysis results indicate that the proposed parallel method can generate accurate cumulant maps, and the performance comparisons on various examples show that all the speedups for third-order, fourth-order and fifth-order cumulants calculation are over 17 times.

Curve-Fitting on Graphics Processors Using Particle Swarm Optimization

AbstractCurve fitting is a fundamental task in many research fields. In this paper we present results demonstrating the fitting of 2D images using CUDA (compute unified device architecture) on NVIDIA graphics processors via particle swarm optimization (PSO). Particle swarm optimization is particularly well-suited to implementation on graphics processors using CUDA as each CUDA thread can be made to model a single particle in a swarm with the swarm itself defined by thread blocks.The motivation for this work was the reconstruction of interferometric photoactivated localization microscopy (iPALM) data sets. The reconstruction requires the fitting of 2D curves to potentially millions of detected photoactivation peaks. Additional motivation was to search for a solution that replaces a cluster with a single desktop machine using multiple CUDA graphics cards.PSO curve fitting running on the GPU enabled a substantial performance increase over the CPU alone and scaled well with multiple CUDA cards. The performanc...

Accelerating GOR Algorithm Using CUDA

Protein secondary structure prediction is very important for its molecular structure. GOR algorithm is one of the most successful computational methods and has been widely used as an efficient analysis tool to predict secondary structure from protein sequence. However, the running time is unbearable with sharp growth in protein database. Fortunately, CUDA (Compute Unified Device Architecture) provides a promising approach to accelerate secondary structure prediction. Therefore, we propose a fine-grained parallel implementation to parallelize GOR-IV package for accelerating protein secondary structure prediction, in which each amino acid would be assigned to one single CUDA thread, hence protein secondary structure prediction would be parallelized by many CUDA threads simultaneously, and constant cache is resorted to cache parameter table. Experimental results show a speedup factor is more than 173X over original GOR-IV version.

A CUDA-based reverse gridding algorithm for MR reconstruction

MR raw data collected using non-Cartesian method can be transformed on Cartesian grids by traditional gridding algorithm (GA) and reconstructed by Fourier transform. However, its runtime complexity is O(K×N2), where resolution of raw data is N×N and size of convolution window (CW) is K. And it involves a large number of matrix calculation including modulus, addition, multiplication and convolution. Therefore, a Compute Unified Device Architecture (CUDA)-based algorithm is proposed to improve the reconstruction efficiency of PROPELLER (a globally recognized non-Cartesian sampling method). Experiment shows a write–write conflict among multiple CUDA threads. This induces an inconsistent result when synchronously convoluting multiple k-space data onto the same grid. To overcome this problem, a reverse gridding algorithm (RGA) was developed. Different from the method of generating a grid window for each trajectory as in traditional GA, RGA calculates a trajectory window for each grid. This is what “reverse” means. For each k-space point in the CW, contribution is cumulated to this grid. Although this algorithm can be easily extended to reconstruct other non-Cartesian sampled raw data, we only implement it based on PROPELLER. Experiment illustrates that this CUDA-based RGA has successfully solved the write–write conflict and its reconstruction speed is 7.5 times higher than that of traditional GA.

Performance study on CUDA GPUs for parallelizing the local ensemble transformed Kalman filter algorithm

SUMMARYModern graphics cards provide computational capabilities that exceed current CPUs. As one of the computational intensive problems, numerical weather prediction has the opportunity to benefit from the massive number of threads and large memory throughput in the graphics architecture. In this paper, we present the key steps to integrate the Compute Unified Device Architecture (CUDA) programming framework for one key component in numerical weather prediction, the data assimilation algorithm, which incorporates the observational data into the model to produce the best initial condition in the next prediction. The data assimilation algorithm we studied in this paper exhibits good localization and favors parallelism. To maximize the throughput of the graphics card, over a million CUDA threads, global memory coalescing, and fast graphics shared memory are utilized. We also demonstrate the differences in the advancement of GPU architectures from the GTX 200 series to Fermi. The experiments are carried out separately on a GTX 260 (GTX 200 series) and a GTX 460 (Fermi) graphics card. Results show an improvement of 72.1× speedup running on the GTX 260 and 92.7×speedup on the GTX 460. The results provide attractive evidence for applying CUDA GPUs to high demanding scientific computation realms. Copyright © 2011 John Wiley & Sons, Ltd.