Search results

This paper aims to evaluate different approaches for the parallelization of compute-intensive tasks. The study compares a Java multi-threaded algorithm, distributed computing solutions with MapReduce (Apache Hadoop) and resilient distributed data set (RDD) (Apache Spark) paradigms and a graphics processing unit (GPU) approach with Numba for compute unified device architecture (CUDA).

The paper uses a simple but computationally intensive puzzle as a case study for experiments. To find all solutions using brute force search, 15! permutations had to be computed and tested against the solution rules. The experimental application comprises a Java multi-threaded algorithm, distributed computing solutions with MapReduce (Apache Hadoop) and RDD (Apache Spark) paradigms and a GPU approach with Numba for CUDA. The implementations were benchmarked on Amazon-EC2 instances for performance and scalability measurements.

The comparison of the solutions with Apache Hadoop and Apache Spark under Amazon EMR showed that the processing time measured in CPU minutes with Spark was up to 30% lower, while the performance of Spark especially benefits from an increasing number of tasks. With the CUDA implementation, more than 16 times faster execution is achievable for the same price compared to the Spark solution. Apart from the multi-threaded implementation, the processing times of all solutions scale approximately linearly. Finally, several application suggestions for the different parallelization approaches are derived from the insights of this study.

There are numerous studies that have examined the performance of parallelization approaches. Most of these studies deal with processing large amounts of data or mathematical problems. This work, in contrast, compares these technologies on their ability to implement computationally intensive distributed algorithms.

The purpose of this paper is to present a novel strategy used for acceleration of free-vibration analysis, in which the hierarchical matrices structure and Compute Unified Device Architecture (CUDA) platform is applied to improve the performance of the traditional dual reciprocity boundary element method (DRBEM).

The DRBEM is applied in forming integral equation to reduce complexity. In the procedure of optimization computation, ℋ-Matrices are introduced by applying adaptive cross-approximation method. At the same time, this paper proposes a high-efficiency parallel algorithm using CUDA and the counterpart of the serial effective algorithm in ℋ-Matrices for inverse arithmetic operation.

The analysis for free-vibration could achieve impressive time and space efficiency by introducing hierarchical matrices technique. Although the serial algorithm based on ℋ-Matrices could obtain fair performance for complex inversion operation, the CUDA parallel algorithm would further double the efficiency. Without much loss in accuracy according to the examination of the numerical example, the relative error appeared in approximation process can be fixed by increasing degrees of freedoms or introducing certain amount of internal points.

The paper proposes a novel effective strategy to improve computational efficiency and decrease memory consumption of free-vibration problems. ℋ-Matrices structure and parallel operation based on CUDA are introduced in traditional DRBEM.

The purpose of this paper is to accelerate the time‐consuming task of assembling the impedance matrix resulting from the discretization of integral equations by the moment method, accelerated using massively parallel processing scheme.

This paper provides several approaches for the implementation of moment method on compute unified device architecture (CUDA) capable general purpose video cards, as well as giving general implementation design patterns and a good overview on the topic.

The proposed method seems to be efficient in the light of the presented numerical results.

The subject of the paper is an evolving, considerably new aspect among computation techniques which could be of high interest for the scientific community.

An accurate and fast dynamic analysis innovative approach for vehicle parts is provided for engineering practice.

This paper presents an innovative dynamic analysis approach for vehicle parts based on parallel optimization algorithm with CUDA.

This project is supported by the National Science Foundation of China (No. 51805076, No. U1708255 and No. 51775093), the Fundamental Research Funds for the Central Universities (No. N170503011) and the Natural Science Foundation of Liaoning Province, China (No. 20180551058).

This paper presents an innovative approach for vehicle parts using parallel optimization algorithm based on CUDA, which can improve the computing accuracy and speed effectively.

The purpose of this paper is to achieve the acceleration of 3D object transformation using parallel techniques such as multi-core central processing unit (MC CPU) or graphic processing unit (GPU) or even both. Generating 3D animation scenes in computer graphics requires applying a 3D transformation on the vertices of the objects. These transformations consume most of the execution time. Hence, for high-speed graphic systems, acceleration of vertex transform is very much sought for because it requires many matrix operations (need) to be performed in a real time, so the execution time is essential for such processing.

In this paper, the acceleration of 3D object transformation is achieved using parallel techniques such as MC CPU or GPU or even both. Multiple geometric transformations are concatenated together at a time in any order in an interactive manner.

The performance results are presented for a number of 3D objects with paralleled implementations of the affine transform on the NVIDIA GPU series. The maximum execution time was about 0.508 s to transform 100 million vertices using LabVIEW and 0.096 s using Visual Studio. Other results also showed the significant speed-up compared to CPU, MC CPU and other previous work computations for the same object complexity.

The high-speed execution of 3D models is essential in many applications such as medical imaging, 3D games and robotics.

In FEM computations, the mesh quality improves the accuracy of the approximation solution and reduces the computation time. The dynamic bubble system meshing technique can provide high-quality meshes, but the packing process is time-consuming. This paper aims to improve the running time of the bubble meshing by using the advantages of parallel computing on graphics processing unit (GPU).

This paper is based on the analysis of the processing time on CPU. A massively parallel computing-based CUDA architecture is proposed to improve the bubble displacement and database updating. Constraints linked to hardware considerations are taken into account. Finally, speedup factors are provided on test cases and real scale examples.

The numerical experiences show the efficiency of parallel performance reaches a speedup of 35 compared to the serial implementation.

This contribution is so far limited to two-dimensional (2D) geometries although the extension to three-dimension (3D) is straightforward regarding the meshing technique itself and the GPU implementation. The authors’ works are based on a CUDA environment which is widely used by developers. C\C++ and Java were the programming languages used. Other languages may of course lead to slightly different implementations.

This approach makes it possible to use bubble meshing technique for both initial design and optimization, as excellent meshes can be built in few seconds.

Compared to previous works, this contribution shows that the scalability of the bubble meshing technique needs to solve two key issues: reach a T(N) global cost of the implementation and reach a very fast size map interpolation strategy.

The authors will review the main concepts of graphs, present the implemented algorithm, as well as explain the different techniques applied to the graph, to achieve an efficient execution of the algorithm, both in terms of the use of multiple cores that the authors have available today, and the use of massive data parallelism through the parallelization of the algorithm, bringing the graph closer to the execution through CUDA on GPUs.

In this work, the authors approach the graphs isomorphism problem, approaching this problem from a point of view very little worked during all this time, the application of parallelism and the high-performance computing (HPC) techniques to the detection of isomorphism between graphs.

Results obtained give compelling reasons to ensure that more in-depth studies on the HPC techniques should be applied in these fields, since gains of up to 722x speedup are achieved in the most favorable scenarios, maintaining an average performance speedup of 454x.

The paper is new and original.

Stochastic uncertainties in material parameters have a significant impact on the analysis of real-world electromagnetic compatibility (EMC) problems. Conventional approaches via the Monte-Carlo scheme attempt to provide viable solutions, yet at the expense of prohibitively elongated simulations and system overhead, due to the large amount of statistical implementations. The purpose of this paper is to introduce a 3-D stochastic finite-difference time-domain (S-FDTD) technique for the accurate modelling of generalised EMC applications with highly random media properties, while concurrently offering fast and economical single-run realisations.

The proposed method establishes the concept of covariant/contravariant metrics for robust tessellations of arbitrarily curved structures and derives the mean value and standard deviation of the generated fields in a single-run. Also, the critical case of geometrical and physical uncertainties is handled via an optimal parameterisation, which locally reforms the curvilinear grid. In order to pursue extra speed efficiency, code implementation is conducted through contemporary graphics processor units and parallel programming.

The curvilinear S-FDTD algorithm is proven very precise and stable, compared to existing multiple-realisation approaches, in the analysis of statistically-varying problems. Moreover, its generalised formulation allows the effective treatment of realistic structures with arbitrarily curved geometries, unlike staircase schemes. Finally, the GPU-based enhancements accomplish notably accelerated simulations that may exceed the level of 120 times. Conclusively, the featured technique can successfully attain highly accurate results with very limited system requirements.

Development of a generalised curvilinear S-FDTD methodology, based on a covariant/contravariant algorithm. Incorporation of the important geometric/physical uncertainties through a locally adaptive curved mesh. Speed advancement via modern GPU and CUDA programming which leads to reliable estimations, even for abrupt statistical media parameter fluctuations.

The purpose of this paper is to improve the computational speed of solving nonlinear dynamics by using parallel methods and mixed-precision algorithm on graphic processing units (GPUs). The computational efficiency of traditional central processing units (CPUs)-based computer aided engineering software has been difficult to satisfy the needs of scientific research and practical engineering, especially for nonlinear dynamic problems. Besides, when calculations are performed on GPUs, double-precision operations are slower than single-precision operations. So this paper implemented mixed precision for nonlinear dynamic problem simulation using Belytschko-Tsay (BT) shell element on GPU.

To minimize data transfer between heterogeneous architectures, the parallel computation of the fully explicit finite element (FE) calculation is realized using a vectorized thread-level parallelism algorithm. An asynchronous data transmission strategy and a novel dependency relationship link-based method, for efficiently solving parallel explicit shell element equations, are used to improve the GPU utilization ratio. Finally, this paper implements mixed precision for nonlinear dynamic problems simulation using the BT shell element on a GPU and compare it to the CPU-based serially executed program and a GPU-based double-precision parallel computing program.

For a car body model containing approximately 5.3 million degrees of freedom, the computational speed is improved 25 times over CPU sequential computation, and approximately 10% over double-precision parallel computing method. The accuracy error of the mixed-precision computation is small and can satisfy the requirements of practical engineering problems.

This paper realized a novel FE parallel computing procedure for nonlinear dynamic problems using mixed-precision algorithm on CPU-GPU platform. Compared with the CPU serial program, the program implemented in this article obtains a 25 times acceleration ratio when calculating the model of 883,168 elements, which greatly improves the calculation speed for solving nonlinear dynamic problems.

Massively parallel desktop computing capabilities now well within the reach of individual academics modify the environment for posterior simulation in fundamental and potentially quite advantageous ways. But to fully exploit these benefits algorithms that conform to parallel computing environments are needed. This paper presents a sequential posterior simulator designed to operate efficiently in this context. The simulator makes fewer analytical and programming demands on investigators, and is faster, more reliable, and more complete than conventional posterior simulators. The paper extends existing sequential Monte Carlo methods and theory to provide a thorough and practical foundation for sequential posterior simulation that is well suited to massively parallel computing environments. It provides detailed recommendations on implementation, yielding an algorithm that requires only code for simulation from the prior and evaluation of prior and data densities and works well in a variety of applications representative of serious empirical work in economics and finance. The algorithm facilitates Bayesian model comparison by producing marginal likelihood approximations of unprecedented accuracy as an incidental by-product, is robust to pathological posterior distributions, and provides estimates of numerical standard error and relative numerical efficiency intrinsically. The paper concludes with an application that illustrates the potential of these simulators for applied Bayesian inference.