high performance computing on graphics processing units: hgpu.org

The focus of this work is the automatic performance tuning of stencil computations on Graphics Processing Units (GPUs). A strategy is presented that uses machine learning to determine the best way to use the GPU memory followed by a heuristic that divides the remaining optimizations into groups and exhaustively explores one group at a time. […]

This paper describes usage of CUDA parallelization scheme for forward and inverse gravity problems for structural boundaries. Forward problem is calculated using the finite elements approach. This means that the whole calculation volume is split into parallelepipeds and then the gravity effect of each is calculated using known formula. Inverse problem solution is found using […]

Science and Engineering has benefited enormously from the advent of modern (digital) computing. As technology continues to grow, computation capability becomes exponentially faster, more reliable, and more efficient. While modeling and simulations have hurdled analysis past many years of trial and error, they still are restricted by resources, even with modern computing. Whether running Monte […]

This project aims to port the scientific application GADGET-3 to multiple accelerators, research on the performance achieved and compare the porting/optimisations on the given accelerators with different architectures. In this project, the most time-consuming functions of GADGET-3 was identified based on the profiling. Partial functions in GADGET-3 were ported to the accelerator NVIDIA K40 card […]

Advection is the transport of a quantity due to fluid flow, and is an important, computationally intensive part of any fluid simulation. OpenACC GPU acceleration of the advection components of MONC, an atmospheric LES, was pursued. Although this yielded no speedup, the reasons for this are examined, and the conditions under which it may become […]

The present work is an analysis of the performance of the basic vector operations AXPY, DOT and SpMV using OpenCL. The code was tested on the NVIDIA Tesla S2050 GPU and Intel Xeon Phi 3120A coprocessor. Due to the nature of the AXPY function, only two versions were implemented, the routine to be executed by […]

Fluids simulation is usually done with CFD methods which offers high precision but needs days/weeks/months to compute on desktop CPUs which limits the practical use in industrial control systems. In order to reduce the computation time Smoothed Particle Hydrodynamics (SPH) method is used. SPH is commonly used to simulate fluids in computer graphics field, especially […]

A deep generative model is developed for representation and analysis of images, based on a hierarchical convolutional dictionary-learning framework. Stochastic unpooling is employed to link consecutive layers in the model, yielding top-down image generation. A Bayesian support vector machine is linked to the top-layer features, yielding max-margin discrimination. Deep deconvolutional inference is employed when testing, […]

We present in this paper several implementations of the 3D Fast Wavelet Transform (3D-FWT) on multicore CPUs and manycore GPUs. On the GPU side, we focus on CUDA and OpenCL programming to develop methods for an efficient mapping on manycores. On multicore CPUs, OpenMP and Pthreads are used as counterparts to maximize parallelism, and renowned […]

Deep learning (DL) has achieved notable successes in many machine learning tasks. A number of frameworks have been developed to expedite the process of designing and training deep neural networks (DNNs), such as Caffe, Torch and Theano. Currently they can harness multiple GPUs on a single machine, but are unable to use GPUs that are […]

High Energy Physics (HEP) needs a huge amount of computing resources. In addition data acquisition, transfer, and analysis require a well developed infrastructure too. In order to prove new physics disciplines it is required to higher the luminosity of the accelerator facilities, which produce more-and-more data in the experimental detectors. Both testing new theories and […]

High-order numerical methods for unstructured grids combine the superior accuracy of high-order spectral or finite difference methods with the geometric flexibility of low-order finite volume or finite element schemes. The Flux Reconstruction (FR) approach unifies various high-order schemes for unstructured grids within a single framework. Additionally, the FR approach exhibits a significant degree of element […]