8cdd1372-7efb-449b-82d0-ba018el469a9 PyCOOL PyCOOL (Cosmological Object-Oriented Lattice code) is a fast GPU accelerated program that solves the evolution of interacting scalar fields in an expanding universe with symplectic algorithms. The program has been written with the intention to hit a sweet spot of speed, accuracy and user friendliness. This is achieved by using the Python language with the PyCUDA interface to make a program that is very easy to adapt to different scalar field models. http://www.physics.utu.fi/tiedostot/theory/particlecosmology/pycool/ /content/cudazone/CUDABrowser/assets/images/applications/pycool-low.png /content/cudazone/CUDABrowser/assets/images/applications/pycool-med.png Academia University of Turku / Department of Physics and Astronomy 2012 01 24 01/24/2011 Open source Jani Sainio Application Code Science 8cdd1372-7efb-449b-82d0-ba018ef469a9 CUDAfy.NET An open source Microsoft .NET library that allows writing of CUDA applications including device code in languages such as C# and VB. Contains wrappers for CUSPARSE, CUBLAS, CUFFT and CURAND, as well as a growing number of specialized numerics libraries. http://www.hybriddsp.com/Products/CUDAfyNET.aspx /content/cudazone/CUDABrowser/assets/images/applications/cudafyi-low.png /content/cudazone/CUDABrowser/assets/images/applications/cudafy-med.png Commercial Hybrid DSP Systems http://www.hybriddsp.com 2011 12 07 12/07/2011 Open source Hybrid DSP Systems Code Numerics Libraries .NET,C#,VB,Solver 8cdd1372-7efb-449b-82d0-ba018ef469a6 QnDynCUDA We present a set of C++ classes which allow one to use the graphics card processors cores for quantum ab initio simulations, i.e. a direct solving of the time-dependent Schrödinger equation, gaining the benefits from the parallel architecture of the graphical processor units. We use the Chebyshev polynomial and FFT algorithm. The solution is based on NVIDIA CUDA technology. The speed-up factor in the test runs of our classes performed using the graphics card processor can even be of order of 300 in comparison with the test runs using only the single core of CPU. Not only the Schrödinger equation can be integrated using the presented solver. With only small changes it can be used for solving the nonlinear Gross–Pitaevskii equation of BECs dynamics, the heat equation, the diffusion equation or other parabolic partial differential equations of second order. http://dx.doi.org/10.1016/j.cpc.2011.11.026 /content/cudazone/CUDABrowser/assets/images/applications/QnDynCUDA-low.png /content/cudazone/CUDABrowser/assets/images/applications/QnDynCUDA-med.png Academia Nicolaus Copernicus University http://fizyka.umk.pl 2011 12 09 12/09/2011 Open source Tomasz Dziubak Jacek Matulewski Code Paper Numerics Libraries Science CUDA 8cdd1372-7efb-449b-82d0-ba08wef469a6 Efficient Decoding of QC-LDPC Codes Using GPUs In this work, we propose an efficient quasi-cyclic LDPC (QC-LDPC) decoder simulator which runs on graphics processing units (GPUs).We optimize the data structures of the messages used in the decoding process such that both the read and write processes can be performed in a highly parallel manner by the GPUs. We also propose a highly efficient algorithm to convert the data structure of the messages from one form to another with very little latency. Finally, with the use of a large number of cores in the GPU to perform the simple computations simultaneously, our GPU-based LDPC decoder is found to run at around 100 times faster than a CPU-based simulator. http://www.springerlink.com/content/j8g53w2260224wx7/ /content/cudazone/CUDABrowser/assets/images/applications/QC-LDPC-low.png /content/cudazone/CUDABrowser/assets/images/applications/QC-LDPC-med.png Academia The Hong Kong PolyU 2011 06 16 06/16/2011 100 N/A Yue Zhao Xu Chen Chiu-Wing Sham Wai M. Tam and Francis C. M. Lau Paper Signal Processing CUDA 8cdd1372-7efb-449b-82d0-ba018ef4542f Integrating CUDA & GNU Autotools One of the drawbacks to GNU Autotools is that is only provides native support for certain languages, however, it is flexible enough so that you can make it do what you want it to do... if you know how. I wanted to distribute CUDA based applications with GNU Autotools but unfortunately CUDA is not one of the languages that it supports... so I started googling around. I found several threads where other people wanted to be able to do the same thing. I found various bread crumbs here and there that enabled me to piece together a working build. Since I couldn't find all of the information that I needed in one spot, I figured I'd write it all down and publish it so others don't have to waste time figuring it out. "The ClusterChimps Guide to Integrating CUDA and GNU Autotools" is a simple guide to building CUDA targets using GNU Autotools. It will show you how to build stand alone CUDA applications, static CUDA libraries, and shared CUDA libraries. The guide comes with a companion example tarball. http://www.clusterchimps.org/autotools.php /content/cudazone/CUDABrowser/assets/images/applications/dr-zaius-low.png /content/cudazone/CUDABrowser/assets/images/applications/dr-zaius-med.png Research ClusterChimps.org http://www.clusterchimps.org 2011 11 18 11/18/2011 Open source Dr. Zaius Code Paper Tools CUDA, Autotools 8cdd1372-7efb-449b-82d0-ba018ef543e8 FMRI Analysis on the GPU Faster fMRI analysis by using the GPU. http://www.sciencedirect.com/science/article/pii/S0169260711001957 /content/cudazone/CUDABrowser/assets/images/applications/article-01.png /content/cudazone/CUDABrowser/assets/images/applications/article-02.png Research Linköping University http://www.liu.se 2011 11 12 11/12/2011 N/A Anders Eklund Mats Andersson Hans Knutsson Multimedia Paper Signal Processing Medical Imaging fMRI, GPU, permutation test 8cdd1372-7efb-449b-82d0-ba018ef567r9 True 4D Image Denoising on the GPU 4D Image denoising on the GPU http://www.hindawi.com/journals/ijbi/2011/952819/ /content/cudazone/CUDABrowser/assets/images/applications/4-dimension-01.png /content/cudazone/CUDABrowser/assets/images/applications/4-dimensions-02.png Research Linköping University http://www.liu.se 2011 11 12 11/12/2011 N/A Anders Eklund Mats Andersson Hans Knutsson Multimedia Paper Signal Processing Medical Imaging Science 4D, image denoising, CT 8cdd1372-7efb-449b-82d0-ba018ef458y9 A real-time crosstalk canceller on a notebook GPU People want to participate in the communication with the feeling of being together and sharing the same environment. Crosstalk cancellation is one of the main applications in multichannel acoustic signal processing that provides this kind of feelings. This work shows that GPU can be used as a co-processor which carries out audio processing tasks, freeing CPU resources which can be employed in other tasks. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6012072 /content/cudazone/CUDABrowser/assets/images/applications/159460_Crosstalk_01.png /content/cudazone/CUDABrowser/assets/images/applications/159460_Crosstalk_02.png Academia INCO2 (http://www.inco2.upv.es/) and GTAC (http://www.gtac.upv.es/) Groups. Universidad Politecnica de Valencia http://www.upv.es 2011 09 06 09/16/2011 N/A Jose A. Belloch Alberto Gonzalez F. J. Martínez-Zaldívar Antonio M. Vidal Application Paper Signal Processing Video & Audio 8cdd1372-7efb-449b-82d0-ba018ez454f2 Real-time massive convolution for audio applications on GPU Massive convolution is the basic operation in multichannel acoustic signal processing. This field has experienced a major development in recent years due to the growing need to incorporate new effects and the natural desire to improve the hearing experience. These acoustic effects require to compute multiples convolutions simultaneously in real-time. The work we present describes a GPU-implementation of all the operations involved in the convolution, extrapolated to multiple channels. One of the main feature in our work is the utilization of the streams on GPU. This allows to overlap computation and data-transfer from CPU to GPU. This application is the first step to carry out real-time multichannel-sound applications on GPU. http://www.springerlink.com/content/h37u46j2416m6733/ /content/cudazone/CUDABrowser/assets/images/applications/156214_Rea_Time_2.png /content/cudazone/CUDABrowser/assets/images/applications/156214_Rea_Time_1.png Academia INCO2 (http://www.inco2.upv.es/) and GTAC (http://www.gtac.upv.es/) Groups. Universidad Politecnica de Valencia http://www.upv.es 2011 04 19 04/19/2011 N/A Jose A. Belloch Alberto Gonzalez F. J. Martínez-Zaldívar Antonio M. Vidal Application Paper Signal Processing Video & Audio 8cdd1372-7efb-449b-82d0-ca018ef454f2 Exposure Render Exposure Render is a Direct Volume Rendering Application that applies progressive Monte Carlo raytracing, coupled with physically based light transport to heterogeneous volumetric data. Exposure Render enables the configuration of any number of arbitrarily shaped area lights, models a real-world camera, including its lens and aperture, and incorporates complex materials, whilst still maintaining interactive display updates. It features both surface and volumetric scattering, and applies noise reduction to remove the unwanted startup noise associated with progressive Monte Carlo rendering. The complete implementation is available in source and binary forms under a permissive free software license. http://code.google.com/p/exposure-render/downloads/list /content/cudazone/CUDABrowser/assets/images/applications/655602-example-01.png /content/cudazone/CUDABrowser/assets/images/applications/655602-example-02.png Academia TU Delft http://graphics.tudelft.nl 2011 10 19 10/19/2011 Open source T. Kroes Application Paper Code Digital Content Creation Graphics Imaging Medical Imaging Ray Tracing Monte Carlo Simulation, NVIDIA CUDA, Open Source, Ray Tracing, Volume Rendering 8cdd1372-7efb-449b-82e0-ba018ef454f2 DualSPHysics DualSPHysics is a combined CPU/GPU solver for mesh-free Smoothed Particle Hydrodynamics to be applied in CFD applications with free-surface flows. /content/cudazone/CUDABrowser/assets/images/applications/1370_283365_dualsphysics_cuda_small.png /content/cudazone/CUDABrowser/assets/images/applications/1370_283365_dualsphysics_cuda_large.png Academia EPHYSLAB--Universidade de Vigo and School of Mechanical, Aerospace and Civil Engineering-University of Manchester http://ephyslab.uvigo.es/index.php/eng/dual_sphysics/ 2011 01 11 01/11/2011 90 A.J.C. Crespo J.M. Dominguez M.G. Gesteira Multimedia Paper Computational Fluid Dynamics Science SPH, GPU, meshless method, lagrangian, fluid dynamics, free-surface flow,A.J.C. Crespo,J.M. Dominguez,M.G. Gesteira,alexbexe@uvigo.es,jmdominguez@uvigo.es,mggesteira@uvigo.es 9561fe2e-1de6-461b-b0dc-1cae41da5eb5 NeMo: real-time spiking neural network simulation Spiking neural network simulations are used to model biological brain structures. Simulating large-scale networks is computationally expensive, however, due to the number and interconnectedness of neurons in the brain. Furthermore, where such simulations are used in a embodied (i.e. robotic) setting, the simulation must be real-time in order to be useful. http://nemosim.sourceforge.net /content/cudazone/CUDABrowser/assets/images/applications/1368_193704_firing_small.png /content/cudazone/CUDABrowser/assets/images/applications/1368_193704_firing_large.png Academia Imperial College London http://www.imperial.ac.uk 2010 07 18 07/18/2010 20 Open source Andreas Fidjeland Murray Shanahan Paper Life Sciences Science neural network simulation,Andreas Fidjeland,Murray Shanahan,andreas.fidjeland@imperial.ac.uk a6fbee45-c093-4bab-958c-ce8e5e7baa64 CUDA Benoit Realtime, high resolution, high iteration, supersampled, fractal zoom. Specify the vanishing point, iteration count and colors for an animated zoom into the Mandelbrot set and then watch the zoom without having to run a separate, lengthy, rendering step. Multiple zoom specifications, called tracks, can be stored and played back in sequence, creating a continuously running fractal zooming show. /content/cudazone/CUDABrowser/assets/images/applications/1367_86365_player_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1367_86365_player_large.jpg Research 2011 04 30 04/30/2011 Open source Roger Dahl Paper Application Code Graphics Mandelbrot fractal realtime zoom log scale map,Roger Dahl,dahlsys@gmail.com 457a8222-cc5c-4ab5-a938-290aa493803f SARUMAN SARUMAN (Semiglobal Alignment of short Reads Using CUDA and NeedleMAN-Wunsch) is a mapping approach that returns all possible alignment positions of a read in a reference genome sequence under a given error threshold, together with one optimal alignment for each of these positions. Alignments are computed in parallel on graphics hardware, facilitating an considerable speedup of this normally time consuming step. http://www.cebitec.uni-bielefeld.de/brf/saruman/saruman.html /content/cudazone/CUDABrowser/assets/images/applications/1366_61128_saruman_flow_small.png /content/cudazone/CUDABrowser/assets/images/applications/1366_61128_saruman_flow_large.png Academia Center for Biotechnology, Bielefeld University http://www.cebitec.uni-bielefeld.de/ 2011 03 30 03/30/2011 Jochen Blom Tobias Jakobi Daniel Doppmeier Sebastian Jaenicke Jörn KalinowskiJens Stoye, Alexander Goesmann Jens Stoye Alexander Goesmann Paper Life Sciences Science Bioinformatics, Sequence Alignment, Short read mapping, Bioinformatics workbench, Sequence Analysis,Jochen Blom,Tobias Jakobi,Daniel Doppmeier,saruman@cebitec.uni-bielefeld.de a11c0bd6-6125-4d81-ad46-4f5c320951c7 Data Assimilation using a GPU Accelerated Path Integral Monte Carlo Approach A general approach to data assimilation (state and parameter estimation) in nonlinear dynamical systems with noisy dynamics and noisy measurements. In general terms, it is a method for extracting a few usefull pieces of information from a large amount of raw time series data. /content/cudazone/CUDABrowser/assets/images/applications/1365_90544_unobsStatesSmall_small.png /content/cudazone/CUDABrowser/assets/images/applications/1365_90544_unobsStatesSmall_large.png Academia University of California, San Diego http://physics.ucsd.edu/ 2011 04 05 04/05/2011 300 John C. Quinn Henry D.I. Abarbanel Paper Science Signal Processing Data Assimilation, Parameter Estimation, Monte Carlo, Path Integral,John C. Quinn,Henry D.I. Abarbanel,jquinn@ucsd.edu,habarbanel@ucsd.edu cf146928-c319-4ea3-97ee-6f24e3b80847 CP_select parallel selection algorithm for GPUs: calculation of the median and order statistics /content/cudazone/CUDABrowser/assets/images/applications/1364_8595_median_small.png /content/cudazone/CUDABrowser/assets/images/applications/1364_8595_median_large.png Academia Deakin University http://www.deakin.edu.au/ 2011 04 10 04/10/2011 6 Open Source Gleb Beliakov Paper Numerics Libraries Programming Tools Science selection, median, sorting,Gleb Beliakov,gleb@deakin.edu.au 5e4ba313-241d-4dc3-8146-66ddb7379614 Mesh-particle interpolations on GPUs and multicore CPUs Particle-mesh interpolations are fundamental operations for particle-in-cell codes, as implemented in vortex methods, plasma dynamics and electrostatics simulations. In these simulations, the mesh is used to solve the field equations and the gradients of the fields are used in order to advance the particles. The time integration of particle trajectories is performed through an extensive resampling of the flow field at the particle locations. The computational performance of this resampling turns out to be limited by the memory bandwidth of the underlying computer architecture. We investigate how mesh-particle interpolation can be efficiently performed on graphics processing units (GPUs) and multicore central processing units (CPUs), and we present two implementation techniques. http://rsta.royalsocietypublishing.org/content/369/1944/2164.abstract /content/cudazone/CUDABrowser/assets/images/applications/1363_259257_cyl-re40000_small.png /content/cudazone/CUDABrowser/assets/images/applications/1363_259257_cyl-re40000_large.png Academia CSE Lab, ETH Zurich www.cse-lab.ethz.ch 2011 06 01 06/01/2011 Diego Rossinelli Christian Conti Petros Koumoutsakos Paper Computational Fluid Dynamics Game Physics Numerics CPU, GPU, HPC, mesh-particle, grid-particle,Diego Rossinelli,Christian Conti,Petros Koumoutsakos,petros@inf.ethz.ch 9abc758e-c89b-4761-bfcc-57c36df7e9a1 GPU-computing in econophysics and statistical physics A recent trend in computer science and related fields is general purpose computing on graphics processing units (GPUs), which can yield impressive performance. With multiple cores connected by high memory bandwidth, today's GPUs offer resources for non-graphics parallel processing. This article provides a brief introduction into the field of GPU computing and includes examples. http://dx.doi.org/10.1140/epjst/e2011-01398-x /content/cudazone/CUDABrowser/assets/images/applications/1362_5681_econophysics_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1362_5681_econophysics_large.gif Academia ETH Zurich http://www.tobiaspreis.de/ 2011 04 07 04/07/2011 Open source Tobias Preis Paper Finance Science Compuational Finance, Computational Physics,Tobias Preis,mail@tobiaspreis.de aa2667f4-078f-444b-b136-f0065d5c014e Processing and rendering of Fourier domain optical coherence tomography images at a line rate over 524 kHz using a graphics processing unit In Fourier domain optical coherence tomography (FD-OCT), a large amount of interference data needs to be resampled from the wavelength domain to the wavenumber domain prior to Fourier transformation. We present an approach to optimize this data processing, using a graphics processing unit (GPU) and parallel processing algorithms. We demonstrate an increased processing and rendering rate over that previously reported by using GPU paged memory to render data in the GPU rather than copying back to the CPU. http://spie.org/x648.html?product_id=896535 /content/cudazone/CUDABrowser/assets/images/applications/1361_367646_Screenshot-a_small.png /content/cudazone/CUDABrowser/assets/images/applications/1361_367646_Screenshot-a_large.png Academia Aston University, Queen Mary Univ of London, NPL www.aston.ac.uk 2011 02 01 02/01/2011 Janarthanan Rasakanthan Kate Sugden Peter H. Tomlins Paper Medical Imaging Signal Processing Optical Coherence Tomography, OCT, medical Imaging,Janarthanan Rasakanthan,Kate Sugden,Peter H. Tomlins,raskanj@aston.ac.uk 13893c5b-60cd-468f-b8ec-ea6c11e367c2 Multicore/Multi-GPU Accelerated Simulations of Multiphase Compressible Flows Using Wavelet Adapted Grids We present a computational method of coupling average interpolating wavelets with high-order finite volume schemes and its implementation on heterogeneous computer architectures for the simulation of multiphase compressible flows. The method is implemented to take advantage of the parallel computing capabilities of emerging heterogeneous multicore/multi-GPU architectures. http://epubs.siam.org/sisc/resource/1/sjoce3/v33/i2/p512_s1 /content/cudazone/CUDABrowser/assets/images/applications/1360_263363_application-image_small.png /content/cudazone/CUDABrowser/assets/images/applications/1360_263363_application-image_large.png Academia ETH Zurich www.cse-lab.ethz.ch 2011 03 01 03/01/2011 Diego Rossinelli Babak Hejazialhosseini Daniele G. Spampinato Paper Computational Fluid Dynamics Numerics Signal Processing GPU, compressible flow, wavelets, multiresolution, adaptive grid, multiphase, multicore architectures, OpenCL,Diego Rossinelli,Babak Hejazialhosseini,Daniele G. Spampinato,diegor@inf.ethz.ch 089e53a3-5369-4c7c-b9ff-20d04b833618 Real-time numerical dispersion compensation using graphics processing unit for Fourier-domain optical coherence tomography Numerical dispersion compensation for both standard and full-range Fourier-domain optical coherence tomography (FD-OCT) on the graphics processing unit (GPU) architecture has been implemented. The data acquisition, processing and image display were performed on a multi-thread, CPU-GPU heterogeneous computing system. The real-time ultra-high-resolution full-range complex-conjugate-free FD-OCT imaging was demonstrated at 68.4 frame/s with a frame size of 1024 (lateral) by 2048 (axial) pixels. /content/cudazone/CUDABrowser/assets/images/applications/1359_20606_dispersion_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1359_20606_dispersion_large.jpg Academia Johns Hopkins University http://www.ece.jhu.edu/photonics/zhangkang.html 2011 03 03 03/03/2011 Kang Zhang Jin U. Kang Multimedia Paper Imaging Medical Imaging Life Sciences Signal Processing GPU, Numerical Dispersion Compensation, Optical coherence tomography,Kang Zhang,Jin U. Kang,kzhang8@jhu.edu fa369b09-c534-4b01-b7b8-28ee8d433c77 Real-time intraoperative 4D full-range FD-OCT based on the dual graphics processing units architecture for microsurgery guidance Real-time 4D full-range complex-conjugate-free Fourier-domain optical coherence tomography (FD-OCT) is implemented using a dual graphics processing units (dual-GPUs) architecture. One GPU is dedicated to the FD-OCT data processing while the second one is used for the volume rendering and display. GPU accelerated non-uniform fast Fourier transform (NUFFT) is also implemented to suppress the side lobes of the point spread function to improve the image quality. http://www.opticsinfobase.org/abstract.cfm?uri=boe-2-4-764 /content/cudazone/CUDABrowser/assets/images/applications/1358_40008_microsurgery_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1358_40008_microsurgery_large.jpg Academia Johns Hopkins University http://www.ece.jhu.edu/photonics/zhangkang.html 2011 03 01 03/01/2011 Kang Zhang Jin U. Kang Multimedia Paper Imaging Medical Imaging Life Sciences Signal Processing GPU, Optical coherence tomography , 4D imaging,Kang Zhang,Jin U. Kang,kzhang8@jhu.edu 93912935-c2cc-4303-b5b2-7355ddff8c8e IGMAS+ Three-dimensional (3D) interactive modeling with the IGMAS software provides the means for integrated processing and interpretation of geoid, gravity and magnetic fields, yielding improved geological interpretation. IGMAS 3D models are constructed using triangulated polyhedra to which constant density and/or induced and remnant susceptibility are assigned. http://www.potentialgs.com/ /content/cudazone/CUDABrowser/assets/images/applications/1357_790835_IGMAS_small.png /content/cudazone/CUDABrowser/assets/images/applications/1357_790835_IGMAS_large.png Academia Transinsight GmbH http://transinsight.com/ 2011 02 01 02/01/2011 300 Transinsight GmbH, Christan-Albrecht-Universitat zu Kiel - Department for Geophysics & Geoinformation Paper Oil & Gas Science interactive modelling, gravity, magnetic, seismic, inversion, numerical modelling, OpenCL,Transinsight GmbH, Christan-Albrecht-Universitat zu Kiel - Department for Geophysics & Geoinformation,info@potentialgs.com fb111a83-f81b-4fb8-8adf-b15c8e375ad4 Directionally Unsplit Hydrodynamic Schemes with Hybrid MPI/OpenMP/GPU Parallelization in AMR We present the implementation and performance of a class of directionally unsplit Riemann-solver-based hydrodynamic schemes on Graphic Processing Units (GPU). These schemes, including the MUSCL-Hancock method, a variant of the MUSCL-Hancock method, and the corner-transport-upwind method, are embedded into the adaptive-mesh-refinement (AMR) code GAMER. Furthermore, a hybrid MPI/OpenMP model is investigated, which enables the full exploitation of the computing power in a heterogeneous CPU/GPU cluster and significantly improves the overall performance. /content/cudazone/CUDABrowser/assets/images/applications/1356_1516457_KH_small.png /content/cudazone/CUDABrowser/assets/images/applications/1356_1516457_KH_large.png Academia National Taiwan University, Department of Physics 2011 03 22 03/22/2011 101 Hsi-Yu Schive Ui-Han Zhang Tzihong Chiueh Paper Computational Fluid Dynamics Science hybrid MPI/OpenMP/GPU, AMR,Hsi-Yu Schive,Ui-Han Zhang,Tzihong Chiueh,b88202011@ntu.edu.tw e47a37df-6e16-41c8-b775-a88790713add Horizon MHD General relativistic magnetohydrodynamics code. Used in computational astrophysics applications, particular the prediction of gravitational radiation from compact objects, and the dynamics of magnetars. http://www.horizoncode.org/ /content/cudazone/CUDABrowser/assets/images/applications/1355_172283_orszag_tang_small.png /content/cudazone/CUDABrowser/assets/images/applications/1355_172283_orszag_tang_large.png Academia University of Tuebingen, Institute for Astronomy and Astrophysics 2011 02 25 02/25/2011 200 Burkhard Zink Multimedia Paper Computational Fluid Dynamics Science mhd astrophysics simulator relativistic,Burkhard Zink,bzink@tat.uni-tuebingen.de d0c8b7d9-8adf-4aa6-82aa-4dc5d4c7a070 Practical Time Bundle Adjustment for 3D Reconstruction on GPUt We present a hybrid implementation of sparse bundle adjustment on the GPU using CUDA, with the CPU working in parallel. The algorithm is decomposed into smaller steps, each of which is scheduled on the GPU or the CPU. We develop efficient kernels for the steps and make use of existing libraries for several steps. Our implementation outperforms the CPU implementation significantly, achieving a speedup of 30-40 times over the standard CPU implementation for datasets with upto 500 images on an Nvidia Tesla C2050 GPU /content/cudazone/CUDABrowser/assets/images/applications/1354_45129_CPU-GPU-Hybrid3_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1354_45129_CPU-GPU-Hybrid3_large.jpg Academia IIIT Hyderabad www.iiit.ac.in 2011 01 01 01/01/2011 40 Siddharth Choudhary Paper Siddharth Choudhary,siddharth.choudhary@research.iiit.ac.in 5064a523-1a53-4c6e-9be0-49b519753279 Flow dynamics measurements using digital holographic PIV An in-line digital holographic (D-HPIV) setup and CUDA-accelerated algorithm were implemented in order to measure the instantaneous three-dimensional (3D), three-component (3C) velocity field of nonstationary flows. This increases dramatically the speed of digital video hologram processing. The system can measure the number, 3D position, size, 3C velocity and track of the particles. The results of the hologram reconstruction are represented using OpenGL. /content/cudazone/CUDABrowser/assets/images/applications/1353_145946_Figure3_small.png /content/cudazone/CUDABrowser/assets/images/applications/1353_145946_Figure3_large.png Academia Petrozavodsk State University www.petrsu.ru 2010 10 07 10/07/2010 1000 Dmitry Ekimov Paper Imaging Numerics Science Signal Processing Video & Audio Dmitry Ekimov,edmitr@onego.ru c3c135bf-3665-438c-b634-25ee54e81a90 Numerical simulation of flow around an oscillating cylinder This program presents a finite difference solution for 2D, low Reynolds number (1-350), unsteady flow around and heat transfer from a stationary or oscillating circular cylinder with constant surface temperature and placed in a uniform stream. The fluid is assumed to be incompressible and of constant property. The cylinder is moved mechanically and can vibrate in-line with or transverse to the main stream or can follow an elliptical or figure-8-shaped path. The governing equations are the Navier-Stokes equations, the continuity equation, a Poisson equation for pressure and the energy equation. http://www.filefactory.com/file/cac6d70/n/FlowCFD.zip /content/cudazone/CUDABrowser/assets/images/applications/1352_24973_NVIDIA_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1352_24973_NVIDIA_large.jpg Academia Department of Fluid and Heat Engineering, University of Miskolc, Hungary www.uni-miskolc.hu 2010 02 02 02/02/2010 13 Prof. Laszlo Baranyi, Laszlo Daroczy Multimedia Paper Computational Fluid Dynamics Numerics Science CFD, numerics, Computational Fluid Dynamics, oscillating cylinder, in-line oscillation, transverse oscillation, Figure-8-shape motion, SOR, successive over-relaxation, heat transfer, 2D, Reynolds number, Strouhal number, Nusselt number, incompressible, lift, drag, Poisson equation, Navier-Stokes equations, temperature,Prof. Laszlo Baranyi, Laszlo Daroczy,arambl@uni-miskolc.hu; daroczy4@freemail.hu acbbd15e-82f0-45e6-988a-f1726e4bb1ce Running the High Performance Linpack (HPL) Benchmark on NVIDIA GPUs The HPL benchmark is used to rank the world's Top500 supercomputers. This is a step by step procedure on how to run NVIDIA's version of the HPL benchmark on Tesla GPUs. We also compare the results of a normal HPL run on CPU to a hybird run on CPU-GPU to show the performance boost gained with GPUs. /content/cudazone/CUDABrowser/assets/images/applications/1349_logo_small.png /content/cudazone/CUDABrowser/assets/images/applications/1349_logo_large.png Research Saudi Aramco 2011 01 10 01/10/2011 Open source Mohamad Sindi Paper Benchmark NVIDIA, Linpack, HPL, GPU, Top500, FLOPS, High Performance Computing, HPC,Mohamad Sindi,sindimo@ieee.org be475639-1007-4bf8-bcc8-b103379fdf9d GPU Vision: Accelerating Computer Vision algorithms with Graphics Processing Units We present an introduction to the eld of GPU accelerated computer vision by examining several projects that provide the framework for researchers and developers to tap into the computational power of Graphics Processing Units (GPU). Our goal is to identify the tools and areas where GPU acceleration can provide the highest performance increases in computer vision applications by creating performance benchmarks to compare and contrast the GPU and CPU versions in realistic applications. http://c13software.com/downloads/GPUVision_2011.pdf /content/cudazone/CUDABrowser/assets/images/applications/1347_133852_haar_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1347_133852_haar_large.jpg Academia University of Connecticut 2011 02 09 02/09/2011 Tamas K. Lengyel James Gedarovich Antonio Cusano Thomas J. Peters Paper Imaging Tamas K. Lengyel,James Gedarovich,Antonio Cusano,tamas.k.lengyel@gmail.com,james.gedarovich@gmail.com,antonio.cusano@gmail.com 1deb8e36-96ab-486b-a05e-0af7a067b6b7 CUDA Image Mosaic Creates image mosaics from a database of thumbnails on a pixel by pixel basis using CUDA to perform the image comparisons. Digital Content Creation,Graphics /content/cudazone/CUDABrowser/assets/images/applications/1346_601593_cuda800_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1346_601593_cuda800_large.jpg Research Personal 2011 02 11 02/11/2011 100 Commercial Andy H Coates Multimedia Paper Photo Mosaic Image PhotoMosaic,Andy H Coates,andyhcoates@gmail.com 50827283-e4ee-4c05-bc3d-27bd9df0436b CUDA Real-time renderer /content/cudazone/CUDABrowser/assets/images/applications/1345_62423_chessRefraction_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1345_62423_chessRefraction_large.jpg Research Freelancer 2011 02 05 02/05/2011 20 Open Source Thanassis Tsiodras Multimedia Paper Graphics Ray Tracing SAH AABB BVH Triangle-meshes real-time raytracer,Thanassis Tsiodras,ttsiodras@gmail.com 7717e876-57cb-49db-9350-0ae7cc77ac63 Multi-GPU accelerated multi-spin Monte Carlo simulations of the 2D Ising model A Modern Graphics Processing unit is able to perform massively parallel scientific computations at low cost. We extend our implementation of the checkerboard algorithm for the two dimensional Ising model T. Preis et al., Journal of Computational Physics 228 2009 4468 4477 in order to overcome the memory limitations of a single GPU which enables us to simulate significantly larger systems. Using multi spin coding techniques, we are able to accelerate simulations on a single GPU by factors up to 35 compared to an optimized single Central Processor Unit core implementation which employs multispin coding. www.tobiaspreis.de/publications/bvp_cpc_2010.pdf /content/cudazone/CUDABrowser/assets/images/applications/1344_11409_preis_multi_gpu_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1344_11409_preis_multi_gpu_large.gif Academia Johannes Gutenberg University Mainz www.tobiaspreis.de 2010 08 01 08/01/2010 35 Benjamin Block Peter Virnau Tobias Preis Paper Science Computational Physics, Monte Carlo Simulation, GPU Clusters,Benjamin Block,Peter Virnau,Tobias Preis,mail@tobiaspreis.de 96a2f87d-895b-4b9f-b04e-7ceb30d28941 Hex Protein Docking Modelling protein-protein interactions (PPIs) is an important aspect of structural bioinformatics. The Hex spherical polar Fourier protein docking algorithm has been implemented on Nvidia graphics processor units (GPUs). On a GTX 285 GPU, an exhaustive six-dimensional docking search can be calculated in just 15 seconds using multiple one-dimensional fast Fourier transforms. This represents a 45-fold speed-up over the corresponding calculation on a single CPU, and is at least two orders of magnitude faster than conventional Cartesian grid-based FFT docking approaches. /content/cudazone/CUDABrowser/assets/images/applications/1342_47767_hex_3hfl_docked_rainbow_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1342_47767_hex_3hfl_docked_rainbow_large.jpg Research INRIA http://www.inria.fr 2010 04 24 04/24/2010 45 Dave Ritchie Paper Life Sciences protein docking,Dave Ritchie,dave.ritchie@inria.fr fe5a34d7-6844-4b87-8542-7ef5ce307b1c GPU-accelerated molecular dynamics simulation for study of liquidcrystalline flows We have developed a GPU-based molecular dynamics simulation for the study of flows of fluids with anisotropic molecules such as liquid crystals. An application of the simulation to the study of macroscopic flow (backflow) generation by molecular reorientation in a nematic liquid crystal under the application of an electric field is presented. The computations of intermolecular force and torque are parallelized on the GPU using the cell-list method, and an efficient algorithm to update the cell lists was proposed. http://portal.acm.org/citation.cfm?id=1808870 /content/cudazone/CUDABrowser/assets/images/applications/1340_header_r1_c1_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1340_header_r1_c1_large.jpg Academia Kochi University of Technology 2010 08 01 08/01/2010 50 Alfeus Sunarso Tomohiro Tsuji Shigeomi Chono Paper Science Alfeus Sunarso ,Tomohiro Tsuji,Shigeomi Chono,sunarso@kochi-tech.ac.jp 4672b45f-125b-480f-a8c6-f5aa647b2a75 MandelCUDA Real-time rendering of the Mandelbrot fractal /content/cudazone/CUDABrowser/assets/images/applications/1338_43133_mandel_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1338_43133_mandel_large.gif Research Freelancer 2010 03 01 03/01/2010 40 Open Source Thanassis Tsiodras Paper Graphics Application,Code,Thanassis Tsiodras,ttsiodras@gmail.com adcd077f-17fb-44e4-91de-cb028f7fe788 CUDA Accelerated Particle Engine A simple point sprite based particle engine accelerated with CUDA. /content/cudazone/CUDABrowser/assets/images/applications/1337_309888_particles (1)_small.png /content/cudazone/CUDABrowser/assets/images/applications/1337_309888_particles (1)_large.png Academia Student 2010 05 24 05/24/2010 10 Craig Mouser Multimedia Paper Digital Content Creation Graphics Video Audio Craig Mouser,mouser58907@yahoo.com 8a9de003-fdec-4da9-b81e-d3c7a991d982 Parallel Option Pricing on GPU: Barrier Options and Realized Variance Options We present parallel algorithms implemented in CUDA subroutines ready to run on Graphics Processing Units (GPUs) to price two kinds of financial derivatives, that is: continuous barrier options and realized variance options. The outstanding parallel performance of these algorithms when executed on GPUs is due to the mathematical properties of the pricing formulae used and to their software implementation. http://www.econ.univpm.it/recchioni/finance/w13 /content/cudazone/CUDABrowser/assets/images/applications/1336_Fig2GPU_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1336_Fig2GPU_large.jpg Academia Universita di Camerino, Universita Politecnica delle Marche, Universita di Roma La Sapienza 2010 11 05 11/05/2010 L. Fatone M. Giacinti F. Mariani Paper Finance L. Fatone,M. Giacinti,F. Mariani,lorella.fatone@unicam.it ,m.giacinti@univpm.it ,fra_mariani@libero.it 38d09cb5-3ff4-431c-a705-23b70259a7c1 Graphics processing unit accelerated non-uniform fast Fourier transform for ultrahigh-speed, real-time Fourier-domain OCT We implemented fast Gaussian gridding (FGG)-based non-uniform fast Fourier transform (NUFFT) on the graphics processing unit (GPU) architecture for ultrahigh-speed, real-time Fourier-domain optical coherence tomography (FD-OCT). The Vandermonde matrix-based non-uniform discrete Fourier transform (NUDFT) as well as the linear/cubic interpolation with fast Fourier transform (InFFT) methods are also implemented on GPU to compare their performance in terms of image quality and processing speed. http://www.opticsinfobase.org/oe/abstract.cfm?uri=oe-18-22-23472 /content/cudazone/CUDABrowser/assets/images/applications/1335_165404_finger_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1335_165404_finger_large.jpg Academia Johns Hopkins University www.ece.jhu.edu 2010 10 25 10/25/2010 Kang Zhang Jin U. Kang Paper Imaging Medical Imaging Signal Processing Kang Zhang,Jin U. Kang,kzhang8@jhu.edu ba0d2d88-8e95-486f-87c7-2e084e8bcc35 Real-time 4D signal processing and visualization using graphics processing unit on a regular nonlinear-k Fourier-domain OCT system We realized graphics processing unit (GPU) based real-time 4D (3D + time) signal processing and visualization on a regular Fourier-domain optical coherence tomography (FD-OCT) system with a nonlinear k-space spectrometer. An ultra-high speed linear spline interpolation (LSI) method for -to-k spectral re-sampling is implemented in the GPU architecture, which gives average interpolation speeds of >3,000,000 line/s for 1024-pixel OCT (1024-OCT) and >1,400,000 line/s for 2048-pixel OCT (2048-OCT). http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-18-11-11772 /content/cudazone/CUDABrowser/assets/images/applications/1333_61749_finger tip singles 2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1333_61749_finger tip singles 2_large.jpg Academia Johns Hopkins University www.ece.jhu.edu 2010 05 18 05/18/2010 Kang Zhang Jin U. Kang Paper Imaging Medical Imaging Signal Processing Real-time 4D Optical coherence tomography,Kang Zhang,Jin U. Kang,kzhang8@jhu.edu 4ee10503-6f9a-4136-858c-5df8aa4cf07f GPU Smoldyn Porting to CUDA of the core simulation algorithms of Smoldyn /content/cudazone/CUDABrowser/assets/images/applications/1332_167342_screenshot_small.png /content/cudazone/CUDABrowser/assets/images/applications/1332_167342_screenshot_large.png Research COSBI 2010 12 10 12/10/2010 130 Lorenzo Dematte Davide Prandi Paper Life Sciences Lorenzo Dematte,Davide Prandi,dematte@ieee.org 6933fa46-7fc4-4053-b724-feda8995296b MC-GPU: Monte Carlo Simulation of X-ray Transport for Medical Imaging Applications MC-GPU is a GPU-accelerated x-ray transport simulation code that can generate clinically-realistic radiographic projection images and computed tomography (CT) scans of the human anatomy. MC-GPU implements a massively multi-threaded Monte Carlo simulation algorithm for the transport of x rays in a voxelized geometry and uses the x-ray interaction models and cross sections from PENELOPE 2006. The code can handle realistic human anatomy phantoms, for example the freely available models from the Virtual Family. Electron transport is not implemented. The code has been developed using the CUDA programming model and MPI to address multiple GPUs in parallel. In typical diagnostic imaging simulations, a 15 to 30-fold speed up is obtained using a GPU compared to a CPU execution. /content/cudazone/CUDABrowser/assets/images/applications/1331_99177_mc-gpu_1mmDuke_50keV_1e10hist__All_and_NoScatter_LowRes_small.png /content/cudazone/CUDABrowser/assets/images/applications/1331_99177_mc-gpu_1mmDuke_50keV_1e10hist__All_and_NoScatter_LowRes_large.png Research US Food and Drug Administration http://www.fda.gov/MedicalDevices/ScienceandResearch/ucm2007489.htm 2010 07 08 07/08/2010 30 Open source Andreu Badal Aldo Badano Paper Code Medical Imaging Ray Tracing Andreu Badal,Aldo Badano,andreu_badal@hotmail.com 59b21188-3fc1-448c-bee4-6c5923cfcd67 A demonstration of Exact String Matching Algorithms in CUDA I had a simple idea: is it possible to convert some of the well-known exact string matching algorithms into CUDA versions /content/cudazone/CUDABrowser/assets/images/applications/1330_80324_Screen_shot_2011-01-07_at_PM_2012_12_43_small.png /content/cudazone/CUDABrowser/assets/images/applications/1330_80324_Screen_shot_2011-01-07_at_PM_2012_12_43_large.png Research HP Labs Singapore http://www.hp.com 2010 12 23 12/23/2010 100 Open source Raymond Tay Application Paper Code general purpose computing Raymond Tay,raymondtay1974@gmail.com f48bb746-7b1b-4da3-b204-cc93a3414cc0 Poker Simulation In GPU Simulation is a widely using technique by artificial and human players for helping the decision process in poker. In a typical texas hold'em game simulating all possible game states requires millions of hand evaluations. In this application, we port the Hand-Eval poker library to CUDA providing a generic interface for evaluations of large amounts of hand data. /content/cudazone/CUDABrowser/assets/images/applications/1329_6875_resim_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1329_6875_resim_large.jpg Academia METU 2010 01 31 01/31/2010 15 Commercial Sirin,Volkan Paper Game Simulation Sirin,Volkan,volkansirin@gmail.com 3b26e749-9952-4f86-b8f7-4ca377ef9dea Satellite Image Processing on GPU Satellite Image Processing on GPU is demonstration of performance of remote sensing algorithms such as Shadow Detection and Vegetation Detection on GPU. Also basic image processing algorithms like Contrast Normalization, Histogram Equalization, Automatic Threshold (Otsu's) are implemented. 4 band Satellite Images with 8-bit and 16-bit data are used in tests. Performance of basic and complex algorithms are compared in CPU and GPU with images with various sizes. In the tests, the effect of memory transfer and the order of bands are also considered. /content/cudazone/CUDABrowser/assets/images/applications/1328_60270_imageGPU_small.png /content/cudazone/CUDABrowser/assets/images/applications/1328_60270_imageGPU_large.png Academia Informatics Institute, Middle East Technical University http://www.vrcv.ii.metu.edu.tr 2010 01 31 01/31/2010 10 Open source Mustafa Teke Paper Code Signal Processing Mustafa Teke,mustafa.teke@gmail.com 3dcc9409-2154-48c2-8b2d-9f64294ea4a5 Parallel implementation of large scale crowd simulation Human crowd movement was simulated using texture convolution and a behavioral model inspired by smoothed particle hydrodynamics. In order to make large scale simulation possible in real-time, or almost real-time, we will implement a model for human crowd behavior on a parallel processing platform using CUDA /content/cudazone/CUDABrowser/assets/images/applications/1327_361460_accumulatorMultiColor_small.png /content/cudazone/CUDABrowser/assets/images/applications/1327_361460_accumulatorMultiColor_large.png Academia DIKU 2009 11 11 11/11/2009 Thomas Gronnelov Paper Crowd simulation Thomas Gronnelov,tag@greenleaf.dk c9f1c703-120a-4322-a545-025f92f91c95 Monte Carlo simulation of the q-state Potts Model using CUDA In this work we implement a parallel code to perform finite temperature Monte Carlo simulations of a magnetic system described by a two dimensional q-state Potts model. http://www.famaf.unc.edu.ar/grupos/GPGPU/Potts/CUDAPotts.html /content/cudazone/CUDABrowser/assets/images/applications/1326_82402_potts-nvidia_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1326_82402_potts-nvidia_large.jpg Academia GPGPU Computing Group - Fa.M.A.F. - U.N.C. http://www.famaf.unc.edu.ar/grupos/GPGPU/ 2010 01 05 01/05/2010 155 Open source Ezequiel E. Ferrero Juan Pablo De Francesco Nicolas Wolovick, Sergio A. Cannas Paper Statistical Mechanics Ezequiel E. Ferrero,Juan Pablo De Francesco,Nicolas Wolovick, Sergio A. Cannas,ferrero@famaf.unc.edu.ar,jde@famaf.unc.edu.ar,nicolasw@famaf.unc.edu.ar, cannas@famaf.unc.edu.ar ce4b9b35-490f-49f0-b1e7-4d5ec3b9841f CoroBot CUDA-enabled controller for a mobile robot. The controller takes advantage of an ION board. Machine vision algorithms are accelerated by a minimum of 8x compared to their single-threaded C++ version executed on the ION Atom CPU /content/cudazone/CUDABrowser/assets/images/applications/1325_6852_corobot_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1325_6852_corobot_large.jpg Commercial RealityFrontier http://www.realityfrontier.com 2010 09 01 09/01/2010 8 Commercial Raphael Cariou Application Multimedia Imaging Signal Processing Robotics Raphael Cariou,raphael.cariou@realityfrontier.com 9101844c-56f4-40d7-a856-c75fc81e385a Simulating spin models on GPU Simulations of the Ising, Heisenberg and spin-glass models with Metropolis and parallel tempering updates. /content/cudazone/CUDABrowser/assets/images/applications/1324_checker_small.png /content/cudazone/CUDABrowser/assets/images/applications/1324_checker_large.png Academia Johannes Gutenberg-University Mainz http://www.uni-mainz.de 2010 01 07 01/07/2010 1000 Open source Martin Weigel Paper Code Science Statistical physics Martin Weigel,weigel@uni-mainz.de be82fae2-de7e-48bc-94c2-5f68a86c8c48 iWormhole Desktop Edition Is an ultra-secure file sending Windows Application. This application was designed for consumer use with three guiding principles: 1) Speed, 2) Privacy and 3) Security. /content/cudazone/CUDABrowser/assets/images/applications/1323_36895_screenshot_small.png /content/cudazone/CUDABrowser/assets/images/applications/1323_36895_screenshot_large.png Commercial iWormhole Communications Corp http://www.iwormhole.com 2010 12 12 12/12/2010 1700 Commercial Rob Gagnon Application File Transmission Rob Gagnon,rob@iwormhole.com 56b1c6a1-88bd-4f75-9898-730bd21ce344 Simulation of 1+1 dimensional surface growth andl attices gases using GPUs Restricted solid on solid surface growth models can be mapped onto binary lattice gases. We show that efficient simulation algorithms can be realized on GPUs either by CUDA or by OpenCL programming. We consider a deposition evaporation model following Kardar-Parisi-Zhang growth in 1+1 dimensions related to the Asymmetric Simple Exclusion Process and show that for sizes, that fit into the shared memory of GPUs one can achieve the maximum parallelization speedup ( x100 for a Quadro FX 5800 graphics card with respect to a single CPU of 2.67 GHz). This permits us to study the effect of quenched columnar disorder, requiring extremely long simulation times. We compare the CUDA realization with an OpenCL implementation designed for processor clusters via MPI. A two-lane traffic model with randomized turning points is also realized and the dynamical behavior has been investigated. /content/cudazone/CUDABrowser/assets/images/applications/1322_15738_Model1d_small.png /content/cudazone/CUDABrowser/assets/images/applications/1322_15738_Model1d_large.png Academia MTA-MFA, Res. Inst. for Tech. Phys. and Materials Sci. Budapest http://www.mfa.kfki.hu 2010 12 03 12/03/2010 100 Henrik Schulz Geza Odor Gergely Odor, Mate F. Nagy Paper Statistical Physics Henrik Schulz,Geza Odor,Gergely Odor, Mate F. Nagy,odor@mfa.kfki.hu e795a2dd-798b-4cb4-9630-e8c7cd042a16 CUVI Lib CUDA for Vision and Imaging Library /content/cudazone/CUDABrowser/assets/images/applications/1321_5734_logo_small.png /content/cudazone/CUDABrowser/assets/images/applications/1321_5734_logo_large.png Commercial TunaCode 2010 08 26 08/26/2010 40 Tauseef Rehman Salman Haq Usman Aziz, Jawad Masood Code Medical Imaging Libraries Programming Tools Signal Processing Tauseef Rehman,Salman Haq,Usman Aziz, Jawad Masood,tauseef@tunacode.com c0ae46f3-d256-4c38-9f59-11cddd117c0a Interactive visualization of the largest radioastronomy cubes Astronomy is a data intensive science. The upcoming and future astronomy research facilities will systematically generate terabyte-sized data sets moving astronomy into the Petascale data era. Such increases in dataset size and dimensionality will pose serious computational challenges for many current astronomy data analysis and visualization tools. http://astronomy.swin.edu.au/~ahassan/Research.html /content/cudazone/CUDABrowser/assets/images/applications/1320_optiportal_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1320_optiportal_large.jpg Academia Swinburne University of Technology-Centre for Astrophysics and Supercomputing http://astronomy.swin.edu.au/scivis/ 2010 09 01 09/01/2010 A. H. Hassan C. J. Fluke D. G. Barnes Application A. H. Hassan,C. J. Fluke,D. G. Barnes f01e4f1c-7947-4609-b24c-4732f954bafb Simulation of 1+1 dimensional surface growth andl attices gases using GPUs Restricted solid on solid surface growth models can be mapped onto binary lattice gases. We show that efficient simulation algorithms can be realized on GPUs either by CUDA or by OpenCL programming. We consider a deposition/ evaporation model following Kardar-Parisi-Zhang growth in 1+1 dimensions related to the Asymmetric Simple Exclusion Process and show that for sizes, that fit into the shared memory of GPUs one can achieve the maximum parallelization speedup ( x100 for a Quadro FX 5800 graphics card with respect to a single CPU of 2.67 GHz). This permits us to study the effect of quenched columnar disorder, requiring extremely long simulation times. We compare the CUDA realization with an OpenCL implementation designed for processor clusters via MPI. A two-lane traffic model with randomized turning points is also realized and the dynamical behavior has been investigated. /content/cudazone/CUDABrowser/assets/images/applications/1318_15738_Model1d_small.png /content/cudazone/CUDABrowser/assets/images/applications/1318_15738_Model1d_large.png Academia MTA-MFA, Res. Inst. for Tech. Phys. and Materials Sci. Budapest http://www.mfa.kfki.hu 2010 12 03 12/03/2010 100 Henrik Schulz Geza Odor Gergely Odor, Mate F. Nagy Paper Statistical Physics Henrik Schulz,Geza Odor,Gergely Odor, Mate F. Nagy,odor@mfa.kfki.hu 4936a20a-b98f-4e39-94d2-ec174d002e9e Nonlinear Free Surface Water Waves Fast Desktop Computing for Nonlinear Free Surface Water Waves (OceanWave3D potential flow model) /content/cudazone/CUDABrowser/assets/images/applications/1317_47409_whalint3_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1317_47409_whalint3_large.jpg Academia Technical University of Denmark http://www.imm.dtu.dk/~apek 2010 12 03 12/03/2010 42 Open source Allan P. Engsig-Karup Application Computational Fluid Dynamics Numerics oceanwave3d, potential free surface flow, finite difference method, coastal engineering,Allan P. Engsig-Karup,apek@imm.dtu.dk a79d4690-9c30-4837-88bc-805873f7e5f5 Lagrangian Stochastic Particle Model using Large-Eddy Simulation Meteorology Atmospheric transport and dispersion (T D) models play an important roll in United States national defense. Due to operational time constraints, less sophisticated models have consistently dominated the defense market. Recent advances in graphics processing units (GPUs) and their programming models have made GPUs an attractive platform for commodity, low-power, high-performance parallel computing. Two GPU accelerated (using NVIDIA Corporation's CUDA technology) versions of a sophisticated, large-eddy simulation (LES) based, Lagrangian stochastic model, developed at the National Center for Atmospheric Research (NCAR), were implemented and compared against their single and multiple core CPU (Intel Harpertown) counterparts. The implementation representing the shortest route to GPU acceleration observed a single GPU speedup of 14x over the single core CPU implementation. A more robust and scalable single GPU implementation observed speedups of 20x over the single core CPU implementation. /content/cudazone/CUDABrowser/assets/images/applications/1316_27146_ave_plan_view_crop_small.png /content/cudazone/CUDABrowser/assets/images/applications/1316_27146_ave_plan_view_crop_large.png Academia University of Colorado - Boulder 2010 07 13 07/13/2010 20 Jonathan Hurst Paper Computational Fluid Dynamics Jonathan Hurst,jhurst@ucar.edu ebb0619e-dd44-4c10-ac6c-1659ff388b6f rCUDA 2.0 Allows performing CUDA calls to remote GPUs. /content/cudazone/CUDABrowser/assets/images/applications/1315_4691_rCUDA_logo_small.png /content/cudazone/CUDABrowser/assets/images/applications/1315_4691_rCUDA_logo_large.png Academia UPV / UJI 2010 11 24 11/24/2010 Open source The rCUDA Team Application Paper Code Libraries Programming Tools The rCUDA Team,apenya@gap.upv.es bab91959-3b2a-494c-8644-cf771a2f9bc0 LATTE GPU-accelerated self-consistent tight-binding molecular dynamics for materials with mixed covalent and ionic bonding. /content/cudazone/CUDABrowser/assets/images/applications/1314_main_orig_small.png /content/cudazone/CUDABrowser/assets/images/applications/1314_main_orig_large.png Research Los Alamos National Laboratory http://www.lanl.gov 2010 11 01 11/01/2010 Open source E.J. Sanville N. Bock A. M. N. Niklasson A. Odell S. Rudin M. J. Cawkwell J. Coe Code Life Sciences Science E.J. Sanville,N. Bock,J. Coe, A. M. N. Niklasson, A. Odell, S. Rudin, M. J. Cawkwell,edsanville@gmail.com,nbock@lanl.gov, jcoe@lanl.gov, amn@lanl.gov, aodell@kth.se, srudin@lanl.gov, cawkwell@lanl.gov 4dfaf12b-06c9-418f-8250-7d75fa91a932 Reverse extraction of early-age hydration kinetic equation from observed data of Portland cement. The early-age hydration of Portland cement paste has an important impact on the formation of microstructure and development of strength. However, manual derivation of hydration kinetic equation is very difficult because there are multi-phased, multi-sized and interrelated complex chemical and physical reactions during cement hydration. In this paper, early-age hydration kinetic equation is reversely extracted automatically from the observed time series of hydration degree of Portland cement using evolutionary computation method that combines gene expression programming and particle swarm optimization algorithms. In order to reduce the computing time, GPUs are used for acceleration in parallel. Studies have shown that according to the extracted kinetic equation, simulation curve of early-age hydration is in good accordance with the observed experimental data. Furthermore, this equation still has a good generalization ability even changing chemical composition, particle size and curing conditions. /content/cudazone/CUDABrowser/assets/images/applications/1313_75384_Reverse_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1313_75384_Reverse_large.jpg Academia Provincial Key Laboratory for Network-based Intelligent Computing, University of Jinan, Jinan 250022, China 2010 11 19 11/19/2010 WANG Lin YANG Bo ZHAO XiuYang, CHEN YueHui, CHANG Jun Paper Science Material WANG Lin,YANG Bo,ZHAO XiuYang, CHEN YueHui, CHANG Jun,wangplanet@gmail.com f905568d-b7d2-4ac5-b102-89b8f9a8cbdc IntelliEtch GPU module IntelliEtch is an Anisotropic Wet Etch simulator. This chemical process can be used for Silicon-based Microsystems fabrication. IntelliEtch can be used as a CAD tool for Microsystem fabrication, allowing fast and accurate simulations. /content/cudazone/CUDABrowser/assets/images/applications/1312_87164_Images-036_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1312_87164_Images-036_large.jpg Research I3M Institute(Polytechnic University of Valencia), DIPC Intitute (University of the Basque Country) 2010 10 01 10/01/2010 150 Commercial N Ferrando M A Gosalvez Multimedia Paper Computer Aided Engineering Microsystems MEMS, microsystems, cellular automata,N Ferrando,M A Gosalvez,nesferjo@upvnet.upv.es,miguelangel.gosalvez@ehu.es a63d80ac-a9bb-4cb9-a632-66d9d2563f07 Ultra Fast SOM using CUDA This paper presents an overall idea of the optimization strategies used for the parallel implementation of Basic-SOM on GPU using CUDA programming paradigm. /content/cudazone/CUDABrowser/assets/images/applications/1311_NeST-NVIDIA_Center_small.png /content/cudazone/CUDABrowser/assets/images/applications/1311_NeST-NVIDIA_Center_large.png Commercial NeST http://nestsoftware.com/ 2010 05 18 05/18/2010 Sijo Mathew Preetha Joy Paper Numerics Data mining Sijo Mathew,Preetha Joy,hpc@nestgroup.net 30fbce15-63ea-41fe-8ff3-665316b591e3 AgiSoft PhotoScan AgiSoft PhotoScan is an advanced image-based 3D modeling solution for creating professional quality 3D content from still images. Based on the latest multi-view 3D reconstruction technology, it operates on arbitrary images and is efficient in both controlled and uncontrolled conditions. The photos can be taken from any positions, providing that an object to be reconstructed is visible on at least two photos. Both image alignment and 3D model reconstruction is fully automated. /content/cudazone/CUDABrowser/assets/images/applications/1309_436476_logo-pscan-2_small.png /content/cudazone/CUDABrowser/assets/images/applications/1309_436476_logo-pscan-2_large.png Commercial AgiSoft http://www.agisoft.ru 2010 08 18 08/18/2010 20 Commercial AgiSoft Application Computer Aided Engineering Digital Content Creation Graphics image based modeling,AgiSoft,info@agisoft.ru 46c6675d-b59a-49c8-89dd-fc7a32e6484e Field Forge Field Forge brings massively parallel processing (MPP) to PostgreSQL's current single-threaded sessions. Field Forge utilizes the MPP power of the Kappa framework. The Kappa framework provides practical usage of CUDA GPU, OpenMP, and partitioned data flow scheduled processing. Field Forge make the Kappa framework from Psi Lambda LLC a new Language for defining Window and Table functions. These functions allow processing to be specified using SQL and index component notation for MPP using GPUs and CPUs. Within each Field Forge node, the Kappa framework passes (subsets) of the data sets between processing kernels and into and out of data sets. Field Forge also utilizes the Kappa framework's Apache Portable Runtime (APR) database driver SQL connections to retrieve data fields from any database source (including other Field Forge sessions and nodes), process them using the MPP capabilities of the Kappa framework, and return them as PostgreSQL table or window fields returned from table or window functions respectively. This combination of features enables a Dataset Passing Interface (DPI) for distributed MPP. DPI leverages the existing skills, protocols, connectivity, and infrastructure of an organization. /content/cudazone/CUDABrowser/assets/images/applications/1308_37625_psilambdakappa_small.png /content/cudazone/CUDABrowser/assets/images/applications/1308_37625_psilambdakappa_large.png Commercial Psi Lambda LLC http://psilambda.com 2010 11 07 11/07/2010 Commercial Psi Lambda LLC Application Finance Numerics Life Sciences Programming Tools Science PostgreSQL OpenMP CUDA Window Table Partition,Psi Lambda LLC,kappa@psilambda.com dbb00087-2b32-421a-9e65-f1295b0546b6 Introducing libflame with multi-GPU support We are happy to announce the fifth milestone release (r4648) of libflame, a modern replacement for the most-used functionality of the LAPACK linear algebra library. The main improvement since version 4.0 is that libflame now supports parallel execution using multiple GPUs through the SuperMatrix runtime system. By linking libflame with CUBLAS for the execution of BLAS routines on a single GPU, the SuperMatrix runtime system schedules operations to each GPU and manages the explicit movement of data. This release includes support for single and double precision real and complex floating point operations. /content/cudazone/CUDABrowser/assets/images/applications/1307_205958_FLAMEbanner_small.png /content/cudazone/CUDABrowser/assets/images/applications/1307_205958_FLAMEbanner_large.png Academia UT Austin / Universitat Jaume 2010 10 28 10/28/2010 Open source Ernie Chan Francisco Igual Field van Zee, Robert van de Geijn Application Code Numerics Libraries Ernie Chan,Francisco Igual,Field van Zee, Robert van de Geijn,figual@icc.uji.es 6865f9fe-2a11-47bc-9d50-c0b4026248dc alenka Alenka is a high level, high performance SQL-like language for data processing on CUDA hardware /content/cudazone/CUDABrowser/assets/images/applications/1306_53666_Cubes_small.png /content/cudazone/CUDABrowser/assets/images/applications/1306_53666_Cubes_large.png Research 2010 11 02 11/02/2010 Open source Anton K. Application Code Programming Tools databases Anton K.,antonmks@gmail.com 159ab96a-c59e-4853-bd70-03a4e7691b6f CUDA Accelerated Face Recognition We explore one of the possibilities of parallelizing and optimizing a well-known Face Recognition algorithm, Principal Component Analysis (PCA) with Eigenfaces. /content/cudazone/CUDABrowser/assets/images/applications/1305_NeST-NVIDIA_Center_small.png /content/cudazone/CUDABrowser/assets/images/applications/1305_NeST-NVIDIA_Center_large.png Commercial NeST http://nestsoftware.com 2010 07 26 07/26/2010 Numaan. A Sibi A Paper Imaging Numaan. A,Sibi A,hpc@nestgroup.net df37fa39-5f4f-43cd-8a94-797995daea91 On the Use of Small 2D Convolutions on GPUs Computing many small 2D convolutions using FFTs is a basis for a large number of applications in many domains in science and engineering, among them electromagnetic diraction modeling in physics. The GPU architecture seems to be a suitable architecture to ac- celerate these convolutions, but reaching high application performance requires substantial development time and non-portable optimizations. In this work, we present the techniques, performance results and consid- erations to accelerate small 2D convolutions using CUDA, and compare performance to a multi-threaded CPU implementation. To improve programmability and performance of applications that make heavy use of small convolutions, we argue that two improvements to software and hardware are needed: FFT libraries must be extended with a single con- volution function and communication bandwidth between CPU and GPU needs to be drastically improved. /content/cudazone/CUDABrowser/assets/images/applications/1304_2dconvolutions_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1304_2dconvolutions_large.jpg Academia TUDelft, ASML, TU/e http://www.tudelft.nl/http://www.asml.nl/http://www.tue.nl 2010 06 19 06/19/2010 Shams Al Umairy Alexander S. van Amesfoort Henk Sips, Irwan Setija, Martijn van Beurden Irwan Setija Martijn van Beurden Paper Numerics Science 2D convolution, FFT, Electromagnetic diffraction grating,GPU, CUDA, Tesla,Shams Al Umairy,Alexander S. van Amesfoort,Henk Sips, Irwan Setija, Martijn van Beurden,salumairy@gmail.com,a.s.vanamesfoort@tudelft.nl,sips@ewi.tudelft.nl,Irwan.Setija@asml.com,M.C.v.Beurden@tue.nl f0e9b09f-f104-4594-9ec8-165b10670b21 GFARGO GFARGO simulates the evolution of a gaseous protoplanetary disk subject to the gravitational perturbation of forming protoplanets embedded in it, by solving the Navier-Stokes equations on a polar mesh. It simultaneously describes how the planetary orbits expand or shrink with time, a process known as planetary migration, which plays an important role in shaping the planetary system that emerges once the disk dissipates. The actual implementation is two-dimensionnal, and performance gains ranging up to 90x are achieved with respect to CPU implementations. /content/cudazone/CUDABrowser/assets/images/applications/1303_35573_fargo_small.png /content/cudazone/CUDABrowser/assets/images/applications/1303_35573_fargo_large.png Academia Institute of Physical Sciences, UNAM, Mexico and CEA, Saclay, France 2010 10 22 10/22/2010 90 Open source Frederic Masset Application Code Computational Fluid Dynamics Science Frederic Masset,fmasset@cea.fr 6c2695c6-b17f-4357-ba6a-71715583d5ab 2 million pixel experiment This experimental application maps a HD video source (1080p) into 3D space. Each frame is processed in realtime on the GPU using NVIDIA CUDA technology. Each pixel in a frame (2.073.600 pixels per frame) is scaled by its luminance value and given the original color. The application is written in C# using DirectX11 via SlimDX, CUDA.NET and DirectShow.NET libraries. /content/cudazone/CUDABrowser/assets/images/applications/1302_114048_visualcompute_cuda_app960_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1302_114048_visualcompute_cuda_app960_large.jpg Research noumentalia.de - digital arts - visualcompute.com http://www.noumentalia.de 2010 10 22 10/22/2010 Philipp Drieger Multimedia Presentation Digital Content Creation Graphics Imaging Libraries Science Signal Processing Video & Audio HD video processing 1080p 3D CUDA .NET C# map 3D space,Philipp Drieger,info@visualcompute.com ece4bfff-3896-47dd-b0af-f6abbb592a8e powDOG: powder diffraction on GPUs Diffraction, particularly of X-rays, is a powerful technique for the investigation of structure, microstructure and dynamical properties of matter. In order to link theoretical methods, like Molecular Dynamics and other atomistic approaches, and diffraction experiments we developed a new software for calculating the powder diffraction pattern of nano-sized objects on the GPUs. The software, soon to be made available under GPL license, allows the use of GPUs on different hosts for a direct (brute-force) computation of the Debye scattering equation. /content/cudazone/CUDABrowser/assets/images/applications/1301_1322162_powDOG_small.png /content/cudazone/CUDABrowser/assets/images/applications/1301_1322162_powDOG_large.png Academia University of Trento, Trento, Italy 2010 02 08 02/08/2010 Luca Gelisio Cristy Leonor Azanza Ricardo, Matteo Leoni, Paolo Scardi. Application Science Powder diffraction, Debye scattering equation, nanostructured materials,Luca Gelisio,Cristy Leonor Azanza Ricardo, Matteo Leoni, Paolo Scardi.,luca.gelisio@unitn.it 85367756-abdb-4bc0-ab43-beed43680f51 GPU Accelerated Likelihoods for Stereo-Based Articulated Tracking For many years articulated tracking has been an active research topic in the computer vision community. While working solutions have been suggested, computational time is still problematic. We present a GPU implementation of a ray-casting based likelihood model that is orders of magnitude faster than a traditional CPU implementation. We explain the non-intuitive steps required to attain an optimized GPU implementation, where the dominant part is to hide the memory latency effectively. Benchmarks show that computations which previously required several minutes, are now performed in few seconds /content/cudazone/CUDABrowser/assets/images/applications/1299_88964_gpu_vision_2010_small.png /content/cudazone/CUDABrowser/assets/images/applications/1299_88964_gpu_vision_2010_large.png Academia The eScience Centre,Dept. of Computer Science, University of Copenhagen http://www.diku.dk 2010 09 05 09/05/2010 600 Rune Mollegaard Friborg Soren Hauberg Kenny Erleben Paper Ray Tracing Computer Vision Machine Learning Tracking Articulated Tracking,Particle Filtering,Rune Mollegaard Friborg,Soren Hauberg,Kenny Erleben ,runef@diku.dk,hauberg@diku.dk,kenny@diku.dk 8052c851-ce8c-4767-9452-e3df12796c1d Electronic Design Automation GPU-Based Robust Multigrid Preconditioned Solver for Large Scale On-Chip Power Grid Simulation /content/cudazone/CUDABrowser/assets/images/applications/1298_1108533_40-1-9_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1298_1108533_40-1-9_large.jpg Academia Michigan Technological University http://www.ece.mtu.edu/~zhuofeng/MTU_VLSI_DA.htm 2010 09 15 09/15/2010 50 Zhuo Feng Multimedia Paper Computer Aided Engineering Electronic Design Automation Multigrid, preconditioned iterative methods, power delivery network, on-chip interconnect simulation, VLSI system,Zhuo Feng,zhuofeng@mtu.edu c9e901d1-7e78-40ed-ac3d-4b59592bbb9b Engine_cudamrg for OpenSSL Engine_cudamrg is a cryptographic engine for the OpenSSL Toolkit that can accelerate some operation using a CUDA supported device, we currently support the following cipher types: * AES-128-ECB * AES-128-CBC * AES-192-ECB * AES-192-CBC * AES-256-ECB * AES-256-CBC We support both encryption and decryption for theese cipher types. For future releases we plan to optimize currently supported cipher types, add more cipher types and digest algorithms. /content/cudazone/CUDABrowser/assets/images/applications/1297_8191_engineCudamrg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1297_8191_engineCudamrg_large.png Commercial Engine_cudamrg Development Team http://groups.google.com/group/engine-cudamrg 2010 07 26 07/26/2010 Open source Paolo Margara Application Code Cryptography AES, cryptography,Paolo Margara,paolo.margara@gmail.com 8179b89f-1d5e-4dad-84d4-6b0466dcde7e Smoke Simulation for Fire Engineering using a Multigrid Method on Graphics Hardware We present a GPU-based Computational Fluid Dynamics solver for the purpose of fire engineering. We apply a multigrid method to the Jacobi solver when solving the Poisson pressure equation, supporting internal boundaries. Boundaries are handled on the coarse levels, ensuring that boundaries will never vanish after restriction. We demonstrate cases where the multigrid solver computes results up to three times more accurate than the standard Jacobi method within the same time. Providing rich visual details and flows closer to widely accepted standards in fire engineering. Making accurate interactive physical simulation for engineering purposes, has the benefit of reducing production turn-around time. We have measured speed-up improvements by a factor of up to 350, compared to existing CPU-based solvers. The present CUDA-based solver promises huge potential in economical benefits, as well as constructions of safer and more complex buildings. In this paper, the multigrid method is applied to fire engineering. However, this is not a limitation, since improvements are possible for other fields as well. Traditional Jacobi solvers are particulary suitable for the methods presented. /content/cudazone/CUDABrowser/assets/images/applications/1296_121739_vriphys2009_glimberg_erleben_teaser_small.png /content/cudazone/CUDABrowser/assets/images/applications/1296_121739_vriphys2009_glimberg_erleben_teaser_large.png Academia Department of Computer Science/University of Copenhagen http://www.diku.dk 2009 11 05 11/05/2009 350 Stefan Glimberg Kenny Erleben Jens Bennetsen Paper Computational Fluid Dynamics Computer Aided Engineering Pre-parameter studies of virtual designs Stefan Glimberg,Kenny Erleben,Jens Bennetsen,glimberg@diku.dk,kenny@diku.dk 2e28ba24-3cc8-4333-be77-fb572dc28776 GPU Accelerated Tandem Traversal of Blocked Bounding Volume Hierarchy Collision Detection for Multibody Dynamics The performance bottleneck of physics based animation, is often the collision detection. It is well-known by practitioners that the collision detection may consume more than half of the simulation time. In this work we will introduce a novel approach for collision detection using bounding volume hierarchies. Our approach makes it possible to perform non-convex object versus non-convex object collision on the GPU, using tandem traversals of bounding volume hierarchies. Prior work only supports single traversals on GPUs. We introduce a blocked hierarchy data structure, using imaginary nodes and a simultaneous descend in the tandem traversal. The data structure design and traversal are highly specialized for exploiting the parallel threads in the NVIDIA GPUs. As proof-of-concept we demonstrate a GPU implementation for a multibody dynamics simulation, showing an approximate speedup factor of up to 8 compared to a CPU implementation /content/cudazone/CUDABrowser/assets/images/applications/1295_52591_vriphys2009_damkjaer_erleben_teaser_small.png /content/cudazone/CUDABrowser/assets/images/applications/1295_52591_vriphys2009_damkjaer_erleben_teaser_large.png Academia Department of Computer Science, University of Copenhagen. http://www.diku.dk/ 2009 11 05 11/05/2009 8 Open source Jesper Damkjaer Kenny Erleben Paper Game Physics Graphics Numerics Libraries Programming Tools Science Bounding volume Hierarchies, Collision Detection, Rigid Body Simulation,Jesper Damkjaer, Kenny Erleben,damkjaer@diku.dk,kenny@diku.dk 70463033-d9d8-49f6-9067-8a982284a733 SpofetwraremGPU: Using graphics processing units in RNA microarray association studies Background: Many analyses of microarray association studies involve permutation, bootstrap resampling and crossvalidation, that are ideally formulated as embarrassingly parallel computing problems. Given that these analyses are computationally intensive, scalable approaches that can take advantage of multi-core processor systems need to be developed. http://www.gpucomputing.net/?q=node/2083 /content/cudazone/CUDABrowser/assets/images/applications/1294_bmc_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1294_bmc_large.jpg Commercial BMC Bioinformatics 2010 05 22 05/22/2010 78 Ivo D Shterev Sin-Ho Jung Stephen L George Paper Ivo D Shterev,Sin-Ho Jung,Stephen L George 796fa229-7424-4736-b68b-793cf9120ee9 High performance GPU radix sorting in CUDA This project implements a very fast, efficient radix sorting method for CUDA-capable devices. For sorting large sequences of fixed-length keys (and values), we believe our sorting primitive to be the fastest available for any fully-programmable microarchitecture: our stock NVIDIA GTX480 sorting results exceed the 1G keys/sec average sorting rate (i.e., one billion 32-bit keys sorted per second). http://code.google.com/p/back40computing/wiki/RadixSorting /content/cudazone/CUDABrowser/assets/images/applications/1291_SortingSmall_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1291_SortingSmall_large.jpg Research CUDA Developer 2010 05 27 05/27/2010 Duane Merrill Application Duane Merrill b8244c5b-21cf-4a7d-8eef-7b8e72792b98 Hardware-Assisted Projected Tetrahedra We present a flexible and highly efficient hardware-assisted volume renderer grounded on the original Projected Tetrahedra (PT) algorithm. Unlike recent similar approaches, our method is exclusively based on the rasterization of simple geometric primitives and takes full advantage of graphics hardware. Both vertex and geometry shaders are used to compute the tetrahedral projection, while the volume ray integral is evaluated in a fragment shader; hence, volume rendering is performed entirely on the GPU within a single pass through the pipeline. http://www.lcg.ufrj.br/Members/andream/papers/cgf2010.pdf /content/cudazone/CUDABrowser/assets/images/applications/1290_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1290_GPUComputing bgimg_large.png Academia University of Rio de Janeiro 2010 03 18 03/18/2010 A. Maximo R. Marroquim R. Farias Paper A. Maximo,R. Marroquim,R. Farias 1af9d850-4d52-4267-9787-72027ff4928c A Parallel Algorithm for Construction of Uniform Grids We present a fast, parallel GPU algorithm for construction of uniform grids for ray tracing, which we implement in CUDA. The algorithm performance does not depend on the primitive distribution, because we reduce the problem to sorting pairs of primitives and cell indices. Our implementation is able to take full advantage of the parallel architecture of the GPU, and construction speed is faster than CPU algorithms running on multiple cores. http://graphics.cs.uni-sb.de/fileadmin/cguds/papers/2009/kalojanov_hpg2009/kalojanov_hpg2009.pdf /content/cudazone/CUDABrowser/assets/images/applications/1289_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1289_GPUComputing bgimg_large.png Academia Saarland University 2009 06 13 06/13/2009 Javor Kalojanov Philipp Slusallek Paper Javor Kalojanov,Philipp Slusallek 17270ae2-0417-4766-88db-17f20e1e3073 Evaluation of Streaming Aggregation on Parallel Hardware Architectures We present a case study parallelizing streaming aggregation on three different parallel hardware architectures. Aggregation is a performance-critical operation for data summarization in stream computing, and is commonly found in sense-and-respond applications. Currently available commodity parallel hardware provides promise as accelerators for streaming aggregation. However, how streaming aggregation can map to the different parallel architectures is still an open question. http://people.cs.vt.edu/~scschnei/papers/debs2010.pdf /content/cudazone/CUDABrowser/assets/images/applications/1288_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1288_GPUComputing bgimg_large.png Research IBM Research Division 2010 07 12 07/12/2010 Scott Schneider Henrique Andrade Bugra Gedik Kun-Lung Weu Dimitrios S. Nikolopoulos Paper Scott Schneider,Henrique Andrade,Bugra Gedik 9f95c82d-a9c4-4324-be40-85e0a4d5ebd3 A Middleware for Efficient Stream Processing in CUDA This paper presents a middleware capable of out-of-order execution of kernels and data transfers for efficient stream processing in the compute unified device architecture (CUDA). Our middleware runs on the CUDA-compatible graphics processing unit (GPU). Using the middleware, application developers are allowed to easily overlap kernel computation with data transfer between the main memory and the video memory. http://www-hagi.ist.osaka-u.ac.jp/research/papers/201005_s-nakagw_isc.pdf /content/cudazone/CUDABrowser/assets/images/applications/1287_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1287_GPUComputing bgimg_large.png Academia University of Trier 2010 03 12 03/12/2010 Shinta Nakagawa Fumihiko Ino Kenichi Hagihara Paper Shinta Nakagawa,Fumihiko Ino,Kenichi Hagihara 5fc49232-b32f-49ab-81d9-8548d9b4b730 An Adaptive Performance Modeling Tool for GPU Architectures This paper presents an analytical model to predict the performance of general-purpose applications on a GPU architecture. The model is designed to provide performance information to an auto-tuning compiler and assist it in narrowing down the search to the more promising implementations. It can also be incorporated into a tool to help programmers better assess the performance bottlenecks in their code. We analyze each GPU kernel and identify how the kernel exercises major GPU microarchitecture features. http://impact.crhc.illinois.edu/ftp/conference/sara.pdf /content/cudazone/CUDABrowser/assets/images/applications/1286_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1286_GPUComputing bgimg_large.png Academia University of Illinois at Urbana-Champaign 2009 11 19 11/19/2009 Sara S. Baghsorkhi Matthieu Delahaye Sanjay J. Patel William D. Gropp Wen-mei W. Hwu Paper Sara S. Baghsorkhi,Matthieu Delahaye,Sanjay J. Patel,bsadeghi@illinois.edu,matthieu@illinois.edu,sjp@illinois.edu d3d1ef02-d0ff-40e9-ae3c-2f380b1f45d7 Kd-Jump: a Path-Preserving Stackless Traversal for Faster Isosurface Raytracing on GPUs Stackless traversal techniques are often used to circumvent memory bottlenecks by avoiding a stack and replacing return traversal with extra computation. This paper addresses whether the stackless traversal approaches are useful on newer hardware and technology (such as CUDA). To this end, we present a novel stackless approach for implicit kd-trees, which exploits the benefits of index-based node traversal, without incurring extra node visitation. This approach, which we term Kd-Jump, enables the traversal to immediately return to the next valid node, like a stack, without incurring extra node visitation (kd-restart). http://vplab.snu.ac.kr/lectures/09-2/graphics/lecture_notes/11%20Kd-jump.pdf /content/cudazone/CUDABrowser/assets/images/applications/1285_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1285_GPUComputing bgimg_large.png Academia Bangor University 2009 07 27 07/27/2009 David m. Hughes Ik Soo Lim Paper David m. Hughes,Ik Soo Lim,meirion@bangor.ac.uk,i.s.lim@bangor.ac.uk b82e77d4-8aab-4e0c-828d-d9c87e198557 Accelerating Flow Cytometry Data Clustering Workflows with Graphics Processing Units Flow cytometry is a mainstay technology used by biologists and immunologists for counting, sorting, and analyzing cells suspended in a fluid. The results of flow cytometry are used in a variety of important clinical and research applications such as phenotyping, DNA analysis, and cell function analysis. Like many modern scientific applications, flow cytometry produces massive amounts of data which must be clustered in order to be useful. Conventional analysis of flow cytometry data uses manual sequential bivariate gating. http://cyberaide.googlecode.com/svn/trunk/papers/thesis-pangborn/proposal/pangborn-proposal.pdf /content/cudazone/CUDABrowser/assets/images/applications/1284_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1284_GPUComputing bgimg_large.png Academia Rochester Institute of Technology 2009 09 01 09/01/2009 Andrew D. Pangborn Paper Andrew D. Pangborn b3217a94-0e15-431b-bdba-c2b96f030be4 GPU Accelerated Scientific Computing: Fluid and Particulate Flows with CUDA Simulations of particulate flows, which involve gases and liquids with suspended solid particles like dust, are generally highly CPU-time demanding. The question arises whether such computations can be performed on the GPU applying highly parallel programming models like CUDA. In this paper we demonstrate that numerical simulation in that context can greatly benefit from these emerging technologies and present results in a 2D and 3D setup. http://numhpc.math.kit.edu/download/PARS_Full_Paper_Final_Heuveline_Hahn_Rocker.pdf /content/cudazone/CUDABrowser/assets/images/applications/1283_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1283_GPUComputing bgimg_large.png Academia University of Karlsruhe 2009 10 14 10/14/2009 Tobias Hahn Vincent Heuveline Bjorn Rocker Paper Tobias Hahn,Vincent Heuveline,Bjorn Rocker,tobias.hahn@kit.edu,vincent.heuveline@kit.edu,bjoern.rocker@kit.edu 3587097f-b801-4185-ad0c-f4d4c78480c9 General-Purpose vs. GPU: Comparison of Many-Cores on Irregular Workloads XMT1 is a general-purpose many-core parallel architecture. The foremost design objective for XMT was to meet the highest standards for ease of parallel programming. GPUs, on the other hand, have acquired a strong reputation on performance, sometimes at the expense of ease of programming. The current paper presents a performance comparison on diverse workloads between XMT and an NVIDIA CUDA-enabled GPU. Configured with roughly the same amount of chip resources as the GPU, XMT achieves an average speedup of 6.05x on irregular applications, while incurring an average slowdown of 2.07x on regular ones. http://www.umiacs.umd.edu/users/vishkin/XMT/CKTV_hotpar10.pdf /content/cudazone/CUDABrowser/assets/images/applications/1282_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1282_GPUComputing bgimg_large.png Academia University of Maryland, College Park 2010 04 27 04/27/2010 George C. Caragea Fuat Keceli Alexandros Tzannes Uzi Vishkin Paper George C. Caragea,Fuat Keceli,Alexandros Tzannes,gcaragea@umd.edu,keceli@umd.edu,tzannes@umd.edu ace3d995-2e1a-4d64-a5af-450ffcfee3fd Fast Minimum Spanning Tree for Large Graphs on the GPU Graphics Processor Units are used for many general purpose processing due to high compute power available on them. Regular, data-parallel algorithms map well to the SIMD architecture of current GPU. Irregular algorithms on discrete structures like graphs are harder to map to them. Efficient data-mapping primitives can play crucial role in mapping such algorithms onto the GPU. In this paper, we present a minimum spanning tree algorithm on Nvidia GPUs under CUDA, as a recursive formulation of Boruvka's approach for undirected graphs. http://www.gpucomputing.net/?q=node/1612 /content/cudazone/CUDABrowser/assets/images/applications/1281_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1281_GPUComputing bgimg_large.png Academia International Institute of Information Technology 2009 06 07 06/07/2009 50 Vibhav Vineet Pawan Harish Suryakant Patidar P. J. Narayanan Paper Vibhav Vineet,Pawan Harish,Suryakant Patidar,vibhavvinet@research.iiit.ac.in,harishpk@research.iiit.ac.in,skp@research.iit.ac.in 81ab4f38-8401-4ac6-9e25-d5ad5edceb53 CUDA-based Triangulations of Convolution Molecular Surfaces Computing molecular surfaces is important to measure areas and volumes of molecules, as well as to infer useful information about interactions with other molecules. Over the years many algorithms have been developed to triangulate and to render molecular surfaces. However, triangulation algorithms usually are very expensive in terms of memory storage and time performance, and thus far from real-time performance. Fortunately, the massive computational power of the new generation of low-cost GPUs opens up an opportunity window to solve these problems: real-time performance and cheap computing commodities. http://salsahpc.indiana.edu/ECMLS2010/papers/066.pdf /content/cudazone/CUDABrowser/assets/images/applications/1280_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1280_GPUComputing bgimg_large.png Academia Universidade da Beira Interior 2010 06 20 06/20/2010 Sergio Dias Kuldeep Bora Abel Gomes Paper Sergio Dias,Kuldeep Bora,Abel Gomes,sdias@ubi.pt,kuldeep@iitg.ernet.in,agomes@di.ubi.pt a5df523d-c91c-4105-b8ee-fb0979362cc3 Data Parallel Three-Dimensional Cahn-Hilliard Field Equation Simulation on GPUs with CUDA Computational scientific simulations have long used parallel computers to increase their performance. Recently graphics cards have been utilised to provide this functionality. GPGPU APIs such as NVIDIA's CUDA can be used to harness the power of GPUs for purposes other than computer graphics. GPUs are designed for processing two-dimensional data. In previous work we have presented several two-dimensional Cahn-Hilliard simulations that each utilise different CUDA memory types and compared their results. http://www.massey.ac.nz/~kahawick/cstn/073/cstn-073.pdf /content/cudazone/CUDABrowser/assets/images/applications/1279_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1279_GPUComputing bgimg_large.png Academia Massey University 2009 02 01 02/01/2009 D. P. Playne K. A. Hawick Paper D. P. Playne,K. A. Hawick,d.p.playne@massey.ac.nz,k.a.hawick@massey.ac.nz 0b5f7cf7-9d2e-4318-b4c5-9c9a1dba1143 FAST VISUAL HULL AND STEREO MATCHING ON CUDA Stereo matching and visual hull are techniques that are often used in 3D reconstruction. This paper presents and evaluates implementations of these algorithms on the GPU using the CUDA architecture. Experimental results show that both, visual hull and stereo matching, have much to gain in terms of speed from the data parallel execution model. /content/cudazone/CUDABrowser/assets/images/applications/1278_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1278_GPUComputing bgimg_large.png Academia University of Surrey 2010 02 11 02/11/2010 Mykyta Fastovets Jean-Yves Guillemaut Adrian Hilton Paper Mykyta Fastovets,Jean-Yves Guillemaut,Adrian Hilton,mykyta.fastovets@surrey.ac.uk, j.guillemaut@surrey.ac.uk,a.hiltong@surrey.ac.uk 17140b63-a39b-4cdf-bd86-9a54079055b8 Speed records for NTRU In this paper NTRUEncrypt is implemented for the first time on a GPU using the CUDA platform. As is shown, this operation lends itself excellently for parallelization and performs extremely well compared to similar security levels for ECC and RSA giving speedups of around three to four orders of magnitude. The focus is on achieving a high throughput, in this case performing a large number of encryptions/decryptions in parallel. http://www.gpucomputing.net/?q=node/1573 /content/cudazone/CUDABrowser/assets/images/applications/1277_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1277_GPUComputing bgimg_large.png Academia University of Leuven 2009 09 10 09/10/2009 Jens Hermans Frederik Vercauteren Bart Preneel Paper Jens Hermans,Frederik Vercauteren,Bart Preneel 042c7881-8db7-45d5-8824-8c7f99c6ce91 Implementing a GPU Programming Model on a non-GPU Accelerator Architecture Parallel codes are written primarily for the purpose of performance. It is highly desirable that parallel codes be portable between parallel architectures without significant performance degradation or code rewrites. While performance portability and its limits have been studied thoroughly on single processor systems, this goal has been less extensively studied and is more difficult to achieve for parallel systems. Emerging single-chip parallel platforms are no exception; writing code that obtains good performance across GPUs and other many-core CMPs can be challenging. http://hal.archives-ouvertes.fr/docs/00/49/39/05/PDF/A4MMC-kofsky.pdf /content/cudazone/CUDABrowser/assets/images/applications/1275_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1275_GPUComputing bgimg_large.png Academia University of Illinois at Urbana-Champaign 2010 06 21 06/21/2010 Stephen M. Kofsky Daniel R. Johnson John A. Stratton Wen-mei W. Hwu Sanjay J. Patel Steven S. Lumetta Paper Stephen M. Kofsky,Daniel R. Johnson,John A. Stratton 93887a5d-c38e-40ec-bda5-b5bb59aae6d7 Parallelising Wavefront Applications on General-Purpose GPU Devices Pipelined wavefront applications form a large portion of the high performance scientific computing workloads at supercomputing centres such as LANL in the United States and AWE in the United Kingdom. This paper investigates the viability of utilising graphics processing units (GPUs) for the acceleration of these codes, using NVIDIA's Compute Unified Device Architecture (CUDA). Wavefront applications differ from the massively data-parallel codes typically selected for execution on GPUs in that their computation must obey a strict data dependency, limiting the achievable level of parallelism. http://www2.warwick.ac.uk/fac/sci/dcs/research/pcav/publications/pubs/ukpew-gpu-wavefronts.pdf /content/cudazone/CUDABrowser/assets/images/applications/1274_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1274_GPUComputing bgimg_large.png Academia University of Warwick 2010 06 01 06/01/2010 S. J. Pennycook G. R. Mudalige S. D. Hammond S. A. Jarvis Paper S. J. Pennycook,G. R. Mudalige,S. D. Hammond,sjp@dcs.warwick.ac.uk,g.r.mudalige@dcs.warwick.ac.uk,sdh@dcs.warwick.ac.uk f6b4d7a3-c2f6-456c-80f9-1a6666c19c99 Performance Cost Analysis of Software-Implemented Hardware Fault Tolerance Methods in General-Purpose GPU Computing Commercial off-the-shelf graphics processing units (GPUs) provide an attractive, inexpensive platform for highthroughput scientific applications. Whereas fault tolerance may be desirable for many scientific applications, off-the-shelf GPU hardware has been designed for commodity graphics applications, where fault tolerance is not necessary. http://homepages.cae.wisc.edu/~ece753/papers/Paper_4.pdf /content/cudazone/CUDABrowser/assets/images/applications/1273_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1273_GPUComputing bgimg_large.png Academia University of Wisconsin, Madison 2009 04 26 04/26/2009 Anthony E. Gregerson Ameya V. Abhyankar Paper Anthony E. Gregerson,Ameya V. Abhyankar,agregerson@wisc.edu,aabhyankar@wisc.edu a90fc8ef-5fba-447e-b091-fd5f9d5b1e50 GPU Accelerated Stylistic Augmented Reality With the introduction of programmable graphics pipeline, the highly parallel processing power of graphical processing units (GPU) is being used not only for special graphics effects but also for general purpose computation in areas such as molecular dynamics simulation, stock options pricing, and image processing. In this work, we utilize this power to increase the immersion level in an augmented reality (AR) application. http://www.vmasc.odu.edu/downloads/Capstone_Papers/Engineering/Aras.pdf /content/cudazone/CUDABrowser/assets/images/applications/1272_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1272_GPUComputing bgimg_large.png Academia Old Dominion University 2010 04 02 04/02/2010 Rifat Aras Yuzhong Shen Paper Rifat Aras,Yuzhong Shen 33549c56-faa7-4ed2-8762-b7176670e639 A Batched GPU Algorithm for Set Intersection Intersection of inverted lists is a frequently used operation in search engine systems. Efficient CPU and GPU intersection algorithms for large problem size are well studied. We propose an efficient GPU algorithm for high performance intersection of inverted index lists on CUDA platform. This algorithm feeds queries to GPU in batches, thus can take full advantage of GPU processor cores even if problem size is small. We also propose an input preprocessing method which alleviate load imbalance effectively. http://nbjl.nankai.edu.cn/Lab_Papers/2009/A%20Batched%20GPU%20Algorithm%20for%20Set%20Intersection.pdf /content/cudazone/CUDABrowser/assets/images/applications/1271_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1271_GPUComputing bgimg_large.png Academia Nankai University 2009 09 19 09/19/2009 Di Wu Fan Zhang Naiyong Ao Fang Wang Xiaoguang Liu Gang Wang Paper Di Wu,Fan Zhang,Naiyong Ao,wakensky@gmail.com,zhangfan555@gmail.com,aonaiyong@163.com c7c4497b-7a3a-4847-8010-748fc72fdd19 GPU-based ultrafast IMRT plan optimization The widespread adoption of on-board volumetric imaging in cancer radiotherapy has stimulated research efforts to develop online adaptive radiotherapy techniques to handle the inter-fraction variation of the patient's geometry. Such efforts face major technical challenges to perform treatment planning in real time. To overcome this challenge, we are developing a supercomputing online re-planning environment (SCORE) at the University of California, San Diego (UCSD). As part of the SCORE project, this paper presents our work on the implementation of an intensity-modulated radiation therapy (IMRT) optimization algorithm on graphics processing units (GPUs). http://iopscience.iop.org/0031-9155/54/21/008 /content/cudazone/CUDABrowser/assets/images/applications/1270_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1270_GPUComputing bgimg_large.png Academia University of California, San Diego 2009 10 14 10/14/2009 40 Chunhua Men Xuejun Gu Dongju Choi Amitava Majumdar Ziyi Zheng Klaus Mueller Steve B. Jiang Paper Chunhua Men,Xuejun Gu,Dongju Choi 4e362946-a2fd-4f2b-8740-0009b7348bd6 Real-time Forest Simulation for a Flight Simulator using a GPU This paper concerns the real-time simulation of forests for a flight simulator, exploiting the capacities of recent graphics cards. As we will show, these architectures coupled with recent ergonomic environments like CUDA allow C-programmers to implement highly parallelizable algorithms to be executed on GPU, without being specialized in parallel programming. http://www.ecam-rennes.fr/IMG/pdf/ICCTA2008.pdf /content/cudazone/CUDABrowser/assets/images/applications/1268_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1268_GPUComputing bgimg_large.png Academia Louis de Broglie, Graduate Engineering School 2008 02 19 02/19/2008 Jean-Marc Laferte Guillaume Daussin Pascal Haigron Jihed Flifla Paper Jean-Marc Laferte,Guillaume Daussin,Jihed Flifla,laferte@ecole-debroglie.fr,g.daussin@ecole-debroglie.fr,flifla@ecole -debroglie.fr e490331e-ba50-4ff8-ad01-f2af8b63cada cuInspiral: prototype gravitational waves detection pipeline fully coded on GPU using CUDA In this paper we report the prototype of the first coalescing binary detection pipeline fully implemented on NVIDIA GPU hardware accelerators. The code has been embedded in a GPU library, called cuInspiral and has been developed under CUDA framework. The library contains for example a PN gravitational wave signal generator, matched filtering/FFT and detection algorithms that have been profiled and compared with the corresponding CPU code with dedicated benchmark in order to provide gain factor respect to the standard CPU implementation. http://arxiv.org/PS_cache/arxiv/pdf/1006/1006.4644v1.pdf /content/cudazone/CUDABrowser/assets/images/applications/1267_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1267_GPUComputing bgimg_large.png Research National Institute of Nuclear Physics 2010 06 16 06/16/2010 Leone B. Bosi Paper Leone B. Bosi b0a54882-2b4a-4058-a6e7-b829a2a04a53 GPU Accelerated Path-planning for Multi-agents in Virtual Environments Many games are populated by synthetic humanoid actors that act as autonomous agents. The animation of humanoids in real-time applications is yet a challenge if the problem involves attaining a precise location in a virtual world (path-planning), and moving realistically according to its own personality, intentions and mood (motion planning). In this paper we present a strategy to implement - using CUDA on GPU - a path planner that produces natural steering behaviors for virtual humans using a numerical solution for boundary value problems. http://www.sbgames.org/papers/sbgames09/computing/full/cp15_09.pdf /content/cudazone/CUDABrowser/assets/images/applications/1266_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1266_GPUComputing bgimg_large.png Academia Federal University of Rio Grande do Sul 2009 10 08 10/08/2009 56 Leonardo G. Fischer Renato Silveira Luciana Nedel Paper Leonardo G. Fischer,Renato Silveira,Luciana Nedel cb2fd4bb-d313-4c57-aae7-547e4b78dc27 Real-time image segmentation on a GPU Efficient segmentation of color images is important for many applications in computer vision. Non-parametric solutions are required in situations where little or no prior knowledge about the data is available. In this paper, we present a novel parallel image segmentation algorithm which segments images in real-time in a non-parametric way. The algorithm finds the equilibrium states of a Potts model in the superparamagnetic phase of the system. http://upcommons.upc.edu/e-prints/bitstream/2117/7866/1/1104-Real-time-image-segmentation-on-a-GPU.pdf /content/cudazone/CUDABrowser/assets/images/applications/1264_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1264_GPUComputing bgimg_large.png Academia Georg-August University 2010 06 28 06/28/2010 Alexey Abramov Tomas Kulvicius Florentin Worgotter Babette Dellen Paper Alexey Abramov,Tomas Kulvicius,Florentin Worgotter,abramov@bccn-goettingen.de,tomas@bccn-goettingen.de,worgottg@bccn-goettingen.de 7f4e5772-7ed9-40a1-8da0-7694fec71c3c Development of a GPU-Based Monte Carlo Dose Calculation Code for Coupled Electron-Photon Transport Monte Carlo simulation is the most accurate method for absorbed dose calculations in radiotherapy. Its efficiency still requires improvement for routine clinical applications, especially for online adaptive radiotherapy. In this paper, 20 we report our recent development on a GPU-based Monte Carlo dose calculation code for coupled electron-photon transport. We have implemented the Dose Planning Method (DPM) Monte Carlo dose calculation package (Sempau et al, Phys. Med. Biol., 45(2000)2263-2291) on GPU architecture under CUDA platform. http://arxiv.org/ftp/arxiv/papers/0910/0910.0329.pdf /content/cudazone/CUDABrowser/assets/images/applications/1263_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1263_GPUComputing bgimg_large.png Academia University of California, San Diego 2010 03 22 03/22/2010 Xun Jia Xuejun Gu Josep Sempau Dongju Choi Amitava Majumdar Steve B. Jiang Paper Xun Jia,Xuejun Gu,Josep Sempau,Dongju Choi,Amitava Majumdar,Steve B. Jiang e41a3774-d96e-444e-a928-811d5f31b161 Performance Characterization of a GPU as a Ubiquitous Accelerator in Commodity Multiprocessor Systems Graphic processing units (GPUs) are increasingly being employed as commodity data-parallel co-processors in desktop and laptop systems due to their tremendous computational power as well as high memory bandwidth. A number of research efforts are focusing on the development of methodologies for efficient utilization of GPU hardware as a ubiquitous accelerator for CPU and memory intensive tasks to off-load the main processor(s). In order to effectively off-load parts of computation, developers need to have a clear understanding of performance trade-offs of using GPU as an accelerator for the host processor. http://www.kics.edu.pk/hpcnl/images/hpcnl_kics_tr_03.pdf /content/cudazone/CUDABrowser/assets/images/applications/1262_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1262_GPUComputing bgimg_large.png Academia Al-Khawarizmi Institute of Computer Science 2010 06 01 06/01/2010 Ghulam Mustafa Abdul Waheed Waqar Mahmood Paper Ghulam Mustafa,Abdul Waheed,Waqar Mahmood,ghulam.mustafa@kics.edu.pk,awaheed@kics.edu.pk,director@kics.edu.pk 38d916ba-ce82-48eb-9159-6f33a4396526 Implementation of Stereophonic Acoustic Echo Canceller on nVIDIA GeForce Graphics Processing Unit This paper presents an implementation of a stereophonic acoustic echo canceller on NVIDIA GeForce graphics processor and CUDA software development environment. For efficiency, fast shared memory has been used as much as possible. A tree adder is introduced to reduce the cost for summing thread outputs up. The performance evaluation results suggest that Even a low-cost GPU's with a small number of shader processor greatly helps the echo cancellation for low-cost PCbased teleconferencing. /content/cudazone/CUDABrowser/assets/images/applications/1261_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1261_GPUComputing bgimg_large.png Academia Kanazawa University 2009 12 07 12/07/2009 Akihiro Hirano Kenji Nakayama Paper Akihiro Hirano,Kenji Nakayama,hirano@t.kanazawa-u.ac.jp,nakayama@t.kanazawa-u.ac.jp 74329815-f8c3-4f35-92de-134ac83e4ada Implementing Closed-Form Expressions on FPGAs Using the NAL, with Comparison to CUDA GPU and Cell BE Implementations This paper outlines the Nallatech Accelerator Layer (NAL) and its relationship to Intel's Accelerator Abstraction Layer. The NAL is looked at in its academic context. Hardware platforms that support the NAL are discussed: the Nallatech H101, the Intel FSB-FPGA Module and the BenOne PCIe. The Intel QuickAssist Technology initiative and its associated Accelerator Abstraction Layer (AAL) are introduced. http://www.rssi2008.org/proceedings/papers/posters/07_Bruce.pdf /content/cudazone/CUDABrowser/assets/images/applications/1260_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1260_GPUComputing bgimg_large.png Research Nallatech Ltd 2008 06 17 06/17/2008 Robin Bruce Javier Setoain Richard Chamberlain Malachy Devlin Rosa M. Badia Paper Robin Bruce,Javier Setoain,Richard Chamberlain 64be2be4-6831-45f3-aa71-7ed3906590cc MITHRA: Multiple data Independent Tasks on a Heterogeneous Resource Architecture With the advent of high-performance COTS clusters, there is a need for a simple, scalable and fault-tolerant parallel programming and execution paradigm. In this paper, we show that the popular MapReduce programming model can be utilized to solve many interesting scientific simulation problems with much higher performance than regular cluster computers by leveraging GPGPU accelerators in cluster nodes. We use the Massive Unordered Distributed (MUD) formalism and establish a one-to-one correspondence between it and general Monte Carlo simulation methods. http://verma7.com/wp/wp-content/uploads/2009/09/CS597_Spring09_MITHRA_Technical_Report.pdf /content/cudazone/CUDABrowser/assets/images/applications/1259_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1259_GPUComputing bgimg_large.png Academia University of Illinois at Urbana-Champaign 2009 08 25 08/25/2009 Reza Farivar Abhishek Verma Ellick M. Chan Roy H. Campbell Paper Reza Farivar,Abhishek Verma,Ellick M. Chan,rhc@illinois.edu,Roy H. Campbell,farivar2@illinois.edu,verma7@illinois.edu,emchan@illinois.edu 82928644-d96f-4ba8-9c46-0e6bf1e1f95e Simulation of Reaction-Diffusion Processes in Three Dimensions using CUDA Numerical solution of reaction-diffusion equations in three dimensions is one of the most challenging applied mathematical problems. Since these simulations are very time consuming, any ideas and strategies aiming at the reduction of CPU time are important topics of research. A general and robust idea is the parallelization of source codes/programs. Recently, the technological development of graphics hardware created a possibility to use desktop video cards to solve numerically intensive problems. http://arxiv.org/ftp/arxiv/papers/1004/1004.0480.pdf /content/cudazone/CUDABrowser/assets/images/applications/1257_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1257_GPUComputing bgimg_large.png Academia Eotvos Lorand University 2010 04 03 04/03/2010 Ferenc Molnar Jr Ferenc Izsak Robert Meszaros Paper Ferenc Molnar Jr,Ferenc Izsak,Robert Meszaros 171b372c-fb8e-4fc3-9a02-a3db231424a7 CUDASW++2.0: enhanced Smith-Waterman Protein Database Search on CUDA-Enabled GPUs Based on SIMT and Virtualized SIMD Abstractions Due to its high sensitivity, the Smith-Waterman algorithm is widely used for biological database searches. Unfortunately, the quadratic time complexity of this algorithm makes it highly time-consuming. The exponential growth of biological databases further deteriorates the situation. To accelerate this algorithm, many efforts have been made to develop techniques in high performance architectures, especially the recently emerging many-core architectures and their associated programming models. http://www.biomedcentral.com/content/pdf/1756-0500-3-93.pdf /content/cudazone/CUDABrowser/assets/images/applications/1256_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1256_GPUComputing bgimg_large.png Academia Nanyang Technological University, Singapore 2010 04 14 04/14/2010 Yongchao Liu Bertil Schmidt Douglas L Maskell Paper Yongchao Liu,Bertil Schmidt,Douglas L Maskell,liu0039@ntu.edu.sg 2733733f-bf02-44f0-92ce-49eed1ab150c Design and Implementation of the Smith-Waterman Algorithm on the CUDA-Compatible GPU This paper describes a design and implementation of the Smith-Waterman algorithm accelerated on the graphics processing unit (GPU). Our method is implemented using compute unified device architecture (CUDA), which is available on the nVIDIA GPU. The method efficiently uses on-chip shared memory to reduce the data amount being transferred between off-chip memory and processing elements in the GPU. Furthermore, it reduces the number of data fetches by applying a data reuse technique to query and database sequences. http://www-hagi.ist.osaka-u.ac.jp/research/papers/200810_y-munekw_bibe.pdf /content/cudazone/CUDABrowser/assets/images/applications/1255_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1255_GPUComputing bgimg_large.png Academia Osaka University 2008 08 09 08/09/2008 Yuma Munekawa Fumihiko Ino Kenichi Hagihara Paper Yuma Munekawa,Fumihiko Ino,Kenichi Hagihara,y-munekw@ist.osaka-u.ac.jp,ino@ist.osaka-u.ac.jp 5f961830-f9db-421a-b7b6-cd749907f46e Tapping the Supercomputer Under Your Desk: Solving Dynamic Equilibrium Models with Graphics Processors This paper shows how to build algorithms that use graphics processing units (GPUs) installed in most modern computers to solve dynamic equilibrium models in economics. In particular, we rely on the compute unifed device architecture (CUDA) of NVIDIA GPUs. We illustrate the power of the approach by solving a simple real business cycle model with value function iteration. We document improvements in speed of around 200 times and suggest that even further gains are likely. /content/cudazone/CUDABrowser/assets/images/applications/1254_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1254_GPUComputing bgimg_large.png Academia Duke University 2010 04 10 04/10/2010 Eric M. Aldrich Jesus Fernandez-Villaverde A. Ronald Gallant Juan F. Rubio-Ramirez Paper Eric M. Aldrich,Jesus Fernandez-Villaverde,A. Ronald Gallant,ealdrich@gmail.com,jesusfv@econ.upenn.edu,aronaldg@gmail.com,Juan F. Rubio-Ramirez,jfr23@duke.edu c1f6075d-5f89-44d5-938b-f527b8a56825 Faster Matrix-Vector Multiplication on GeForce 8800GTX Recently a GPU has acquired programmability to perform general purpose computation fast by running ten thousands of threads concurrently. This paper presents a new algorithm for dense matrix-vector multiplication on NVIDIA CUDA architecture. The experimental results on GeForce 8800GTX show that the proposed algorithm runs maximum 15.69 (resp., 32.88) times faster than the sgemv routine in NVIDIA's BLAS library CUBLAS 1.1 (resp., Intel Xeon E5335 CPU with SSE3 SIMD instructions) for matrices with order 16 to 12800. http://ch.nvidia.com/docs/IO/47905/fujimoto_lspp2008.pdf /content/cudazone/CUDABrowser/assets/images/applications/1253_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1253_GPUComputing bgimg_large.png Academia Osaka University 2008 01 29 01/29/2008 15 Noriyuki Fujimoto Paper Noriyuki Fujimoto,fujimoto@ist.osaka-u.ac.jp 3939b0a0-52a9-48e0-8b31-6af60eae6ce6 Stackless KD-Tree Traversal for High Performance GPU Ray Tracing Significant advances have been achieved for realtime ray tracing recently, but realtime performance for complex scenes still requires large computational resources not yet available from the CPUs in standard PCs. Incidentally, most of these PCs also contain modern GPUs that do offer much larger raw compute power. However, limitations in the programming and memory model have so far kept the performance of GPU ray tracers well below that of their CPU counterparts. http://www.gpucomputing.net/?q=node/1293 /content/cudazone/CUDABrowser/assets/images/applications/1252_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1252_GPUComputing bgimg_large.png Academia Saarland University and MPI Informatik 2007 06 11 06/11/2007 Stefan Popov Johannes Gunther Hans-Peter Seidel Philipp Slusallek Paper Stefan Popov,Johannes Gunther,Hans-Peter Seidel c747f985-752f-4f32-be50-cf24898c527b Fast Parallel GPU-Sorting Using a Hybrid Algorithm This paper presents an algorithm for fast sorting of large lists using modern GPUs. The method achieves high speed by efficiently utilizing the parallelism of the GPU throughout the whole algorithm. Initially, a parallel bucketsort splits the list into enough sublists then to be sorted in parallel using merge-sort. The parallel bucketsort, implemented in NVIDIA's CUDA, utilizes the synchronization mechanisms, such as atomic increment, that is available on modern GPUs. http://www.gpucomputing.net/?q=node/1291 /content/cudazone/CUDABrowser/assets/images/applications/1251_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1251_GPUComputing bgimg_large.png Academia Chalmers University of Technology 2007 09 25 09/25/2007 Erik Sintorn Ulf Assarsson Paper Erik Sintorn,Ulf Assarsson,erik.sintorn@chalmers.se,uffe@chalmers.se 01c04032-f48d-40e7-ba6f-105e9e541977 Testing the Feasibility of Running a Computationally Intensive Real-Time Traffic Simulation on a Multicore Programmable Graphics Processor In the 1960s, a semiconductor scientist named Gordon Moore theorized that the number of transistors would double each year on a single integrated circuit. Through much effort, the semiconductor industry has been able to closely follow "Moore's Law", but new information shows this type of progress is not sustainable in the coming years. This realization has implications in both chip fabrication and software development. Instead of making chips with more transistors per unit area, industry now produces newer multicore chips. http://www.gpucomputing.net/?q=node/603 /content/cudazone/CUDABrowser/assets/images/applications/1250_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1250_GPUComputing bgimg_large.png Academia University of Virginia 2007 04 04 04/04/2007 Kevin Stammetti Paper Kevin Stammetti 18ea0385-7666-484e-abed-98ef81d2697a A Flexible High-Performance Lattice Boltzmann GPU Code for the Simulations of Fluid Flows in Complex Geometries We describe the porting of the Lattice Boltzmann component of MUPHY, a multi-physics/scale simulation software, to multiple graphics processing units using the Compute Unified Device Architecture. The novelty of this work is the development of ad hoc techniques for optimizing the indirect addressing that MUPHY uses for efficient simulations of irregular domains. http://www.iac.rm.cnr.it/~massimo/Papers/LBEonGPU.pdf /content/cudazone/CUDABrowser/assets/images/applications/1247_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1247_GPUComputing bgimg_large.png Academia 1 Istituto Applicazioni Calcolo, 2 NVIDIA, 3 SOFT, Istituto Nazionale Fisica della Materia, 4Harvard University School of Eng and Applied Sciences, 5 Harvard University Initiative in Innovative Computing 2009 05 11 05/11/2009 Massimo Bernaschi1 Massimiliano Fatica2 Simone Melchionna3,4 Sauro Succi1,5 Efthimios Kaxiras4 Paper Massimo Bernaschi,Massimiliano Fatica,Simone Melchionna,Sauro Succi1,Efthimios Kaxiras 74ce7180-6c27-4cf9-906a-507feae7f418 GPU Clusters for High-Performance Computing Large-scale GPU clusters are gaining popularity in the scientific computing community. However, their deployment and production use are associated with a number of new challenges. In this paper, we present our efforts to address some of the challenges with building and running GPU clusters in HPC environments. We touch upon such issues as balanced cluster architecture, resource sharing in a cluster environment, programming models, and applications for GPU clusters. http://www.ncsa.illinois.edu/~gshi/ppac09_paper.pdf /content/cudazone/CUDABrowser/assets/images/applications/1246_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1246_GPUComputing bgimg_large.png Academia University of Illinois at Urbana-Champaign 2009 08 01 08/01/2009 Volodymyr V. Kindratenko Jeremy J. Enos Guochun Shi Michael T. Showerman Galen W. Arnold John E. Stone James C. Phillips Wen-mei Hwu Paper Volodymyr V. Kindratenko,Jeremy J. Enos,Guochun Shi,kindr@ncsa.uiuc.edu,jenos@ncsa.uiuc.edu,gshi@ncsa.uiuc.edu ed2b492f-10c9-4c78-9acc-215d2de0900e Towards User Transparent Parallel Multimedia The research area of Multimedia Content Analysis (MMCA) considers all aspects of the automated extraction of knowledge from multimedia archives and data streams. To satisfy the increasing computational demands of MMCA problems, the use of High Performance Computing (HPC) techniques is essential. As most MMCA researchers are not HPC experts, there is an urgent need for 'familiar' programming models and tools that are both easy to use and efficient. http://hal.inria.fr/docs/00/49/38/83/PDF/A4MMC-werkhoven.pdf /content/cudazone/CUDABrowser/assets/images/applications/1244_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1244_GPUComputing bgimg_large.png Academia VU University 2010 06 21 06/21/2010 Ben van Werkhoven Jason Maassen Frank J. Seinstra Paper Ben van Werkhoven,Jason Maassen,Frank J. Seinstra,bjvwerkh@few.vu.nl,jason@few.vu.nl,fjseins@vew.vu.nl a4825b5f-3099-4dcc-8335-ba0f18a6a351 Axel: A Heterogeneous Cluster with FPGAs and GPUs This paper describes a heterogeneous computer cluster called Axel. Axel contains a collection of nodes; each node can include multiple types of accelerators such as FPGAs (Field Programmable Gate Arrays) and GPUs (Graphics Processing Units). A Map-Reduce framework for the Axel cluster is presented which exploits spatial and temporal locality through different types of processing elements and communication channels. The Axel system enables the first demonstration of FPGAs, GPUs and CPUs running collaboratively for N-body simulation. http://portal.acm.org/citation.cfm?id=1723112.1723134#abstract /content/cudazone/CUDABrowser/assets/images/applications/1243_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1243_logo_acm_portal2_large.jpg Academia Imperial College London 2010 02 01 02/01/2010 23 Kuen Hung Tsoi Wayne Luk Paper Kuen Hung Tsoi,Wayne Luk f6a0332b-f8ec-4d4d-9551-0f7fc57ad735 GPU-based Hierarchical Computations for View Independent Visibility With rapid improvements in the performance and programmability, Graphics Processing Units (GPUs) have fostered considerable interest in substantially reducing the running time of compute intensive problems. The solution to the view-independent mutual point-pair visibility problem (required for inter-reflections in global illumination) can, it would seem, require the capabilities of the GPUs. http://dspace.library.iitb.ac.in/jspui/bitstream/10054/1708/1/4756034.pdf /content/cudazone/CUDABrowser/assets/images/applications/1242_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1242_GPUComputing bgimg_large.png Academia Indian Institute of Technology Bombay www.cse.iitb.ac.in 2008 12 16 12/16/2008 Rhushabh Goradia Prekshu Ajmera Sharat Chandran Paper Rhushabh Goradia,Prekshu Ajmera,Sharat Chandran e3c99de6-a223-4437-af51-01bd3faf3026 Fast, GPU-based Diffuse Global Illumination For Point Models Photorealistic computer graphics attempts to match as closely as possible the rendering of a virtual scene with an actual photograph of the scene had it existed in the real world. Of the several techniques that are used to achieve this goal, physically-based approaches (i.e. those that attempt to simulate the actual physical process of illumination) provide the most striking results. http://www.cse.iitb.ac.in/~rhushabh/aps/aps4/report.pdf /content/cudazone/CUDABrowser/assets/images/applications/1241_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1241_GPUComputing bgimg_large.png Academia Indian Institute of Technology 2008 08 26 08/26/2008 Rhushabh Goradia Paper Rhushabh Goradia 5c93da6a-193d-4595-998d-07313764eef5 Fast GPU-based Adaptive Tessellation with CUDA Compact surface descriptions like higher-order surfaces are popular representations for both modeling and animation. However, for fast graphics-hardware-assisted rendering, they usually need to be converted to triangle meshes. In this paper, we introduce a new framework for performing on-the-fly crack-free adaptive tessellation of surface primitives completely on the GPU. Utilizing CUDA and its flexible memory write capabilities, we parallelize the tessellation task at the level of single surface primitives. https://www.mpi-sb.mpg.de/~mschwarz/papers/cudatess-eg09.pdf /content/cudazone/CUDABrowser/assets/images/applications/1240_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1240_GPUComputing bgimg_large.png University of Erlangen-Nuremberg 2009 04 01 04/01/2009 Michael Schwarz Marc Stamminger Michael Schwarz,Marc Stamminger 5eb012b6-230e-49e0-ac77-6be5c40b011a Using Graphics Devices in Reverse: GPU-based Image Processing and Computer Vision Graphics and vision are approximate inverses of each other. Ordinarily Graphics Processing Units are used to convert "numbers into pictures" (i.e. computer graphics). In this paper, we discus the use of GPUs in approximately the reverse way to assist in "converting pictures into numbers" (i.e. computer vision). For graphical operations, GPUs currently provide many hundreds of gigaflops of processing power. http://www.uweb.ucsb.edu/~yichuwang/ecv/paper/using_graphics_device_in_reverse.pdf /content/cudazone/CUDABrowser/assets/images/applications/1239_gpucomputing_small.png /content/cudazone/CUDABrowser/assets/images/applications/1239_gpucomputing_large.png Commercial NVIDIA Corporation 2008 08 26 08/26/2008 James Fung Steve Mann Paper James Fung,Steve Mann 1bbe2f89-4dd9-4051-9bf0-bcd1abdc7a03 CUDA SURF - A real-time implementation for SURF Keypoint detection and matching is a basic computer vision task and a necessary ingredient for several applications, e.g., object recognition, structure from motion, panorama stitching. In this work we implement the popular SURF descriptor, an approximation of SIFT, on commodity graphics hardware and achieve real-time performance even for HD images. For VGA images we achieve a speed-up of about 50x and a GTX 285 and for HD images even up to 87x. /content/cudazone/CUDABrowser/assets/images/applications/1238_633214_match_GPU_graff_img1_img2_small_small.png /content/cudazone/CUDABrowser/assets/images/applications/1238_633214_match_GPU_graff_img1_img2_small_large.png Academia TU Darmstadt http://www.tu-darmstadt.de 2010 07 13 07/13/2010 87 Open source Andre Schulz Florian Jung Sebastian Hartte Daniel Trick Christian Wojek Konrad Schindler Jens Ackermann Michael Goesele Application Code Imaging Video & Audio Keypoint detection, Keypoint description, SURF, Object detection, SfM, Image stitching,Andre Schulz,Florian Jung,Sebastian Hartte, Daniel Trick,Christian Wojek, Konrad Schindler, Jens Ackermann, Michael Goesele,wojek@cs.tu-darmstadt.de fdd8f407-1003-4323-82d5-f91b0c483a18 A Real-Time Multigrid Finite Hexahedra Method for Elasticity Simulation using CUDA We present an efficient CUDA implementation of a finite hexahedra multigrid solver for simulating elastic deformable models in real time. Due to the regular shape of the numerical stencil induced by the hexahedral regime, computations and data layout can be restructured to avoid execution divergence and to support memory access patterns enabling the hardware to coalesce multiple memory accesses into single memory transactions. This enables to effectively exploit the GPU's parallel processing units and high memory bandwidth. Performance gains of up to a factor of 12 compared to a highly optimized CPU implementation are demonstrated. /content/cudazone/CUDABrowser/assets/images/applications/1237_170404_VoxelModel_small.png /content/cudazone/CUDABrowser/assets/images/applications/1237_170404_VoxelModel_large.png Academia Computer Graphics and Visualization Group, Technische Universitat Munchen, Germany http://wwwcg.in.tum.de/ 2010 07 10 07/10/2010 12 Christian Dick Joachim Georgii Rudiger Westermann Paper Computer Aided Engineering Numerics Deformable Objects, Finite Element Methods, Multigrid,Christian Dick,Joachim Georgii,Rudiger Westermann 1cbf9658-009b-4ee7-81fe-1f144a456225 Real-time Spatiotemporal Stereo Matching Using the Dual-Cross-Bilateral Grid We introduce a real-time stereo matching technique based on a reformulation of Yoon and Kweons adaptive support weights algorithm [1]. Our implementation uses the bilateral grid to achieve a speedup of 200x compared to a straightforward full-kernel GPU implementation, which in turn is 20x faster than the original CPU implementation, thus making it the fastest technique on the Middlebury website. Published at the European Conference on Computer Vision (ECCV) 2010. /content/cudazone/CUDABrowser/assets/images/applications/1236_167313_DCBGrid-teaser_small.png /content/cudazone/CUDABrowser/assets/images/applications/1236_167313_DCBGrid-teaser_large.png Academia University of Cambridge and Microsoft Research Cambridge 2010 06 24 06/24/2010 200 Christian Richardt Douglas Orr Ian Davies Antonio Criminisi Neil A. Dodgson Multimedia Paper Code Science Video & Audio Computer Vision Christian Richardt,Douglass Orr,Ian Davies, Antonio Criminisi, Neil A. Dodgson,christian.richardt@cl.cam.ac.uk bbaba1fd-20be-44f9-9b64-fea2cd12fce2 Realtime Tracking With a Pan-Tilt Camera The human eye is amazingly adept at tracking moving objects. The process is so natural to humans that it happens without any conscious effort. While this remarkable ability depends in part on the human brain's immense processing power, the fast response of the extraocular muscles and the eyeball's light weight are also vital. Even a small point and shoot camera mounted on a servo is typically too heavy and slow to move with the agility of the human eye. How, then, can we give a computer the ability to track movement quickly and responsively? http://umassgv.blogspot.com/2010/07/realtime-tracking-with-pan-tilt-camera.html /content/cudazone/CUDABrowser/assets/images/applications/1235_123697_tracking_overview_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1235_123697_tracking_overview_large.jpg Academia University of Massachusetts, Amherst http://www.cs.umass.edu 2010 07 06 07/06/2010 Open source Blake Foster Rui Wang Erik Learned-Miller Multimedia Paper Code Video & Audio tracking, camera, fpv, pan tilt, human eye, vision,Blake Foster,Rui Wang,Erik Learned-Miller,blfoster@cs.umass.edu,ruiwang@cs.umass.edu,elm@cs.umass.edu 749bb40e-ff99-4a34-9093-c8ff4b7ab08d Thrust Graph Library Thrust Graph Library provides graph container, algorithm, and other concepts like a Boost Graph Library. This Library based on the thrust, which is a CUDA library of parallel algorithms with an interface resembling the C++ Standard Template Library (STL). /content/cudazone/CUDABrowser/assets/images/applications/1234_53418_networks_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1234_53418_networks_large.jpg Research National Institute of Advanced Industrial Science and Technology (AIST) 2010 07 06 07/06/2010 Open source kazuhiro kojima Code Libraries Graph Library,kazuhiro kojima,k.kojima@aist.go.jp aafa92f6-e4db-4ff5-bbf2-093cef1271ea Modeling Rotor Wakes with a Hybrid OVERFLOW-Vortex Method on a GPU Cluster The vortex core shed from rotorcraft blades maintains coherency---and thus dynamic relevance---many blade turns after its creation. This presents a challenge to traditional Eulerian computational methods, as fine grids are required to suppress numerical diffusion which would weaken the vortex cores after a small number of revolutions. Vortex methods have been used in the past to overcome these problems, as they require computational elements only in vorticity-containing regions, but suffer from greater computational cost per element. http://markjstock.org/research/AIAA-2010-4553.pdf /content/cudazone/CUDABrowser/assets/images/applications/1233_86624_4bladed_720_web2_small.png /content/cudazone/CUDABrowser/assets/images/applications/1233_86624_4bladed_720_web2_large.png Research Applied Scientific Research, Inc. http://www.applied-scientific.com/ 2010 06 29 06/29/2010 Commercial Mark J. Stock Adrin Gharakhani Multimedia Paper Computational Fluid Dynamics cfd rotor helicopter vortex fluid,Mark J. Stock,Adrin Gharakhani,mstock@applied-scientific.com 5eb5fa1a-f845-4b64-863f-48f16aae06bb A GPU-accelerated Boundary Element Method and Vortex Particle Method Vortex particle methods, when combined with multipole-accelerated boundary element methods (BEM), become a complete tool for direct numerical simulation (DNS) of internal or external vortex-dominated flows. In previous work, we presented a method to accelerate the vorticity-velocity inversion at the heart of vortex particle methods by performing a multipole treecode N-body method on parallel graphics hardware. The resulting method achieved a 17-fold speedup over a dual-core CPU implementation. http://markjstock.org/research/AIAA-2010-5099.pdf /content/cudazone/CUDABrowser/assets/images/applications/1232_266408_spheres_cl_vort_crop_small.png /content/cudazone/CUDABrowser/assets/images/applications/1232_266408_spheres_cl_vort_crop_large.png Research Applied Scientific Research, Inc. http://applied-scientific.com/ 2010 07 01 07/01/2010 43 Commercial Mark J. Stock Adrin Gharakhani Paper Computational Fluid Dynamics cfd vortex nbody bem fluid,Mark J. Stock,mstock@applied-scientific.com c7ec9d53-04d1-4b50-9a6a-0f92323dce34 Leukocyte Tracking: ImageJ Plugin This software is a plugin for the ImageJ image processing program. The plugin is designed to detect and track rolling leukocytes (white blood cells) through multiple frames of video. It can take advantage of a CUDA-capable GPU to dramatically accelerate video processing time; with appropriate hardware, near real-time processing can be achieved. /content/cudazone/CUDABrowser/assets/images/applications/1231_26570_leukocytes_small.png /content/cudazone/CUDABrowser/assets/images/applications/1231_26570_leukocytes_large.png Academia University of Virginia 2010 07 01 07/01/2010 26 Open source Michael Boyer David Tarjan Scott T. Acton Kevin Skadron Application Multimedia Paper Code Imaging Medical Imaging Science Video & Audio leukocyte, blood cell, tracking, video,Michael Boyer,David Tarjan,Scott T. Acton, Kevin Skadron,boyer@cs.virginia.edu d36b8ae0-3465-4b14-82f6-07712472e3ae McStas CUDA optimization project Optimize the single crystal component of McStas neutron raytracer using CUDA http://www.mcstas.org/ /content/cudazone/CUDABrowser/assets/images/applications/1229_6226_logo-left_small.png /content/cudazone/CUDABrowser/assets/images/applications/1229_6226_logo-left_large.png Academia eScience Center, University of Copenhagen http://www.escience.ku.dk 2010 01 29 01/29/2010 125 Jesper Dahlkild Martin Djurno Finn Krog Paper Code Ray Tracing Science Jesper Dahlkild,Martin Djurno,Finn Krog,jesper.dahlkild@gmail.com,djurnoe@diku.dk,fk@finnkrog.com c15f6620-26e7-4f91-b857-46380ff67782 Raytracing in participating media This work presents a CUDA-accelerated algorithm for visualization of photorealistic lighting effects which is based on Henrik Wann Jensen's method for global illumination in scenes with participating media. /content/cudazone/CUDABrowser/assets/images/applications/1228_58732_logo_small.png /content/cudazone/CUDABrowser/assets/images/applications/1228_58732_logo_large.png Academia Wroclaw University of Technology 2010 07 02 07/02/2010 10 Open source Piotr Orzechowski Application Multimedia Paper Code Graphics Ray Tracing raytracing, participating media, photon mapping,Piotr Orzechowski,piotr.orzechowski@gmail.com d35e353a-09c4-47df-ba61-10984823be50 Multi-domain, Higher Order Level Set Scheme for 3D Image Segmentation on the GPU A streaming level set PDE solver to handle large volume ( sizes more than the available GPU memory). A higher order and multi-phase solver for smooth segmentation of the volume. /content/cudazone/CUDABrowser/assets/images/applications/1227_545525_Slide2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1227_545525_Slide2_large.jpg Academia The Technical University of Denmark / The University of Texas at Austin 2010 06 16 06/16/2010 10 Open source Ojaswa Sharma Qin Zhang Francois Anton Chandrajit Bajaj Paper Ojaswa Sharma, Qin Zhang,Francois Anton, and Chandrajit Bajaj ,os@imm.dtu.dk,zqyork@ices.utexas.edu,fa@imm.dtu.dk, bajaj@cs.utexas.edu 082dbdd8-2eb2-4511-9072-d5eff768e420 A Simple Pseudo-Random Number Generator Implementation of uniformly and normally distributed pseudo random number generators as device functions. http://people.virginia.edu/~mjt5v/pf/RNG/ /content/cudazone/CUDABrowser/assets/images/applications/1226_144550627858cb6d44ceb02ba9434317_small.png /content/cudazone/CUDABrowser/assets/images/applications/1226_144550627858cb6d44ceb02ba9434317_large.png Academia University of Virginia 2010 06 08 06/08/2010 Michael Trotter Matt Goodrum Application Programming Tools Michael Trotter,Matt Goodrum,mjt5v@virginia.edu,mag6x@virginia.edu 9e6bae6c-dc38-44dd-863a-441c9440bf9e High-order finite-element seismic wave propagation modeling with MPI on a large GPU cluster We implement a high-order finite-element application, which performs the numerical simulation of seismic wave propagation resulting for instance from earthquakes at the scale of a continent or from active seismic acquisition experiments in the oil industry, on a large cluster of NVIDIA Tesla graphics cards using the CUDA programming environment and non-blocking message passing based on MPI. Contrary to many finite-element implementations, ours is implemented successfully in single precision, maximizing the performance of current generation GPUs. We discuss the implementation and optimization of the code and compare it to an existing very optimized implementation in C language and MPI on a classical cluster of CPU nodes. We use mesh coloring to efficiently handle summation operations over degrees of freedom on an unstructured mesh, and non-blocking MPI messages in order to overlap the communications across the network and the data transfer to and from the device via PCIe with calculations on the GPU. We perform a number of numerical tests to validate the single-precision CUDA and MPI implementation and assess its accuracy. We then analyze performance measurements and depending on how the problem is mapped to the reference CPU cluster, we obtain a speedup of 20x or 12x. /content/cudazone/CUDABrowser/assets/images/applications/1225_20831_seismic_paper_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1225_20831_seismic_paper_large.jpg Academia Universite de Pau (France), Florida State University (US,) TU Dortmund (Germany) 2010 06 15 06/15/2010 20 Dimitri Komatitsch Gordon Erlebacher Dominik Goddeke David Michea Paper Numerics Oil & Gas Clusters Dimitri Komatitsch,Gordon Erlebacher,Dominik Goddeke,David Michea,dimitri.komatitsch@univ-pau.fr 5f003419-e86a-4d8b-9d19-997acd44898b To GPU Synchronize or Not GPU Synchronize? The graphics processing unit (GPU) has evolved from being a fixed function processor with programmable stages into a programmable processor with many fixed function components that deliver massive parallelism. By modifying the GPUs stream processor to support general-purpose computation on the GPU (GPGPU), applications that perform massive vector operations can realize many orders of magnitude improvement in performance over a traditional processor, i.e., CPU. https://research.cs.vt.edu/synergy/pubs/papers/feng-iscas2010-gpusync.pdf /content/cudazone/CUDABrowser/assets/images/applications/1224_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1224_GPUComputing bgimg_large.png Academia Virginia Tech 2010 03 28 03/28/2010 Wu-chun Feng Shucai Xiao Paper Wu-chun Feng,Shucai Xiao,wfeng@vt.edu,shucaig@vt.edu aed6944b-13eb-478c-874a-751a332dad9d FATSEA An Architectural Simulator for General Purpose Computing on GPUs We present FATSEA, a functional and performance evaluation simulator written in C++ to handle kernels written in the CUDA programming language aimed for GPGPU computing. FATSEA takes a Parallel Thread eXecution (PTX) code as input, which is a device independent code format generated by the Nvidia CUDA compiler, to validate results and estimate performance on Nvidia platforms. http://ditec.um.es/~jlaragon/papers/FATSEA-RAPIDO-2010.pdf /content/cudazone/CUDABrowser/assets/images/applications/1223_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1223_GPUComputing bgimg_large.png Academia University of Murcia 2009 12 22 12/22/2009 K. E. Ostby J. L. Aragon J. M. Garcia M. Ujaldon Paper K. E. Ostby,J. L. Aragon,J. M. Garcya 0c25d963-909c-4a1e-8f60-81bf8fd868cf Evaluating the use of GPUs in Liver Image Segmentation and HMMER Database Searches In this paper we present the results of parallelizing two life sciences applications, Markov random fieldsbased (MRF) liver segmentation and HMMER's Viterbi algorithm, using GPUs. We relate our experiences in porting both applications to the GPU as well as the techniques and optimizations that are most beneficial. The unique characteristics of both algorithms are demonstrated by implementations on an NVIDIA 8800 GTX Ultra using the CUDA programming environment. We test multiple enhancements in our GPU kernels in order to demonstrate the effectiveness of each strategy. http://cadi.buffalo.edu/papers/2009/2009_4.pdf /content/cudazone/CUDABrowser/assets/images/applications/1222_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1222_GPUComputing bgimg_large.png Academia University at Buffalo, SUNY 2009 02 15 02/15/2009 John Paul Walters Vidyananth Balu Suryaprakash Kompalli Vipin Chaudhary Paper John Paul Walters,Vidyananth Balu,Suryaprakash Kompalli,waltersj@buffalo.edu,vbalu2@buffalo.edu,kompalli@hp.com e21e7880-81bf-46d6-9185-b56a93a0ad3b GPUCT: A GPU-Accelerated CT Reconstruction System' CT scanning is a medical imaging technique commonly used in hospitals, including the University of Virginia Hospital, to see inside the human body. Modern CT scanners can generate images of the body in three dimensions, a process called 3D reconstruction. This project illustrates the feasibility of using graphics hardware (GPUs) to process CT scans in a more efficient and inexpensive manner than current commercial reconstruction systems. Additionally, this research considers the ethical and social implications of an improved CT reconstruction system in terms of risks for hospitals and patients. http://www.cs.virginia.edu/~skadron/Papers/maier_thesis07.pdf /content/cudazone/CUDABrowser/assets/images/applications/1221_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1221_GPUComputing bgimg_large.png Academia University of Virginia 2007 03 30 03/30/2007 56 Drew Maier Paper Drew Maier 72ba00af-f128-438b-9e0f-379786953cce FPGA-Based Hardware Acceleration of Lithographic Aerial Image Simulation Lithography simulation, an essential step in design for manufacturability (DFM), is still far from computationally efficient. Most leading companies use large clusters of server computers to achieve acceptable turn-around time. Thus coprocessor acceleration is very attractive for obtaining increased computational performance with a reduced power consumption. This article describes the implementation of a customized accelerator on FPGA using a polygon-based simulation model. An application-specific memory partitioning scheme is designed to meet the bandwidth requirements for a large number of processing elements. http://cadlab.cs.ucla.edu/~cong/papers/TRETS-17.pdf /content/cudazone/CUDABrowser/assets/images/applications/1220_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1220_GPUComputing bgimg_large.png Academia University of California, Los Angeles 2009 09 17 09/17/2009 15 Jason Congyi Yi Zou Paper Jason Congyi,Yi Zou ebfbcda7-6930-45cc-83d5-1414da4c1325 Visualising Spins and Clusters in Regular and Small-World Ising Models with GPUs Visualising computational simulation models of solid state physical systems is a hard problem for dense lattice models. Fly throughs and cutaways can aid viewer understanding of a simulated system. Interactive time model parameter updates and overlaying of measurements and graticules, cluster colour labelling and other visual highlighting cues can also enhance user intuition of the model's meaning. We present some graphical and simulation optimisation techniques and various graphical rendering and explanatory techniques for computational simulation models such as the Ising model in 2 and 3 dimensions. In addition to aiding understanding of conventional algorithms such as Metropolis Monte Carlo, we try to visualise cluster updates to the system using algorithms like that of Wolff. http://www.massey.ac.nz/~dpplayne/Papers/cstn-108.pdf /content/cudazone/CUDABrowser/assets/images/applications/1219_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1219_GPUComputing bgimg_large.png Academia Massey University 2010 03 19 03/19/2010 A. Leist D. P. Playne K.A. Hawick Paper A. Leist,D. P. Playne,K.A. Hawick 3c9c3b66-7ebb-461e-884c-c820abca8856 Stereo Depth with a Unified Architecture GPU This paper describes how the calculation of depth from stereo images was accelerated using a GPU. The Compute Unified Device Architecture (CUDA) from NVIDIA was employed in novel ways to compute depth using BT cost matching and the Semi-Global Matching algorithm. The challenges of mapping a sequential algorithm to a massively parallel thread environment and performance optimization techniques are considered. http://mplab.ucsd.edu/wp-content/uploads/CVPR2008/WorkShops/data/papers/143.pdf /content/cudazone/CUDABrowser/assets/images/applications/1218_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1218_GPUComputing bgimg_large.png Academia Florida Atlantic University 2008 05 04 05/04/2008 Joel Gibson Oge Marques Paper Joel Gibson ,Oge Marques 4ea4d671-5de1-4355-8c9d-392fa35be551 3D Registration Based on Normalized Mutual Information: Performance of CPU vs. GPU Implementation Medical image registration is time-consuming but can be sped up employing parallel processing on the GPU. Normalized mutual information (NMI) is a well performing similarity measure for performing multi-modal registration. We present CUDA based solutions for computing NMI on the GPU and compare the results obtained by rigidly registering multi-modal data sets with a CPU based implementation. Our tests with RIRE data sets show a speed-up of factor 5 to 7 for our best GPU implementation. http://www.gris.informatik.tu-darmstadt.de/~swesarg/papers/1632.pdf /content/cudazone/CUDABrowser/assets/images/applications/1217_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1217_GPUComputing bgimg_large.png Academia Technische Universitat Darmstadt 2010 01 04 01/04/2010 Florian Jung Stefan Wesarg Paper Florian Jung,Stefan Wesarg,stefan.wesarg@gris.tu-darmstadt.de 93ec4473-01fb-42ff-9e39-46248ff46941 fastHOG - a real-time GPU implementation of HOG We introduce a parallel implementation of the histogram of oriented gradients algorithm for object detection. Our implementation uses the GPU and the NVIDIA CUDA framework. We achieve speedups of over 67x from the standard sequential code, using a single video card. Furthermore it supports multiple video cards so speedups of 120x or more can be achieved. This allows us to achieve real-time performance, using the full HOG algorithm for the first time in the literature. http://www.robots.ox.ac.uk/ActiveVision/Papers/prisacariu_reid_tr2310_09/prisacariu_reid_tr2310_09.pdf /content/cudazone/CUDABrowser/assets/images/applications/1216_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1216_GPUComputing bgimg_large.png Academia University of Oxford 2009 07 14 07/14/2009 120 Victor Adrian Prisacariu Ian Reid Paper Victor Adrian Prisacariu,Ian Reid,victor@robots.ox.ac.uk,ian@robots.ox.ac.uk 608510ce-a512-41c8-b32a-b55cb524284d Detection and Tracking of Human Subjects The goal of the thesis project was to devise an algorithm to detect and track people in a static video. Existing techniques are inadequate; instead a new approach based on background subtraction is used. The approach is successful with a static camera. In background subtraction, the background of the video is calculated a priori and then subtracted from each frame of the video. This isolates the foreign objects, which are detected via two simple algorithms. Both algorithms are based on the subject's center of mass, but the first algorithm traces the path of the person around the video, making it very cluttered. http://www.cs.virginia.edu/~skadron/Papers/Grosvenor_Douglas_thesis.pdf /content/cudazone/CUDABrowser/assets/images/applications/1215_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1215_GPUComputing bgimg_large.png Academia University of Virginia 2009 04 30 04/30/2009 Douglas Grosvenor Paper Douglas Grosvenor 78013ee5-9a75-42af-93a8-4a5b68a330fa Optimizing Sparse Matrix-Vector Multplication on GPUs We are witnessing the emergence of Graphics Processor units (GPUs) as powerful massively parallel systems. Furthermore, the introduction of new APIs for general-purpose comptuations on GPUs, namely, CUDA from NVIDIA, Stream SDK form AMD, and OpenCL, makes GPUs an attractive choice for high-performance numerical and scientific computing. http://domino.watson.ibm.com/library/CyberDig.nsf/papers/1D32F6D23B99F7898525752200618339/$File/rc24704.pdf /content/cudazone/CUDABrowser/assets/images/applications/1214_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1214_GPUComputing bgimg_large.png Research IBM Research Division 2009 04 02 04/02/2009 Muthu Manikandan Baskaran Rajesh Bordawekar Paper Muthu Manikandan Baskaran,Rajesh Bordawekar,baskaran@ces.ohio-state.edu,bordaw@us.ibm.com 67ac5cbe-b91c-46b0-8a4e-9635e86dec49 Acceleration of Binomial Options Pricing via Parallelizing along time-axis on a GPU Since the introduction of organized trading of options for commodities and equities, computing fair prices for options has been an important problem in financial engineering. A variety of numerical methods, including Monte Carlo methods, binomial trees, and numerical solution of stochastic differential equations, are used to compute fair prices. Traders and brokerage firms constantly strive to achieve faster calculation of option prices because timely information can mean the difference between a deal struck or missed, which translates to substantial profit or loss. Hence, the latency to compute a fair option price plays an important role in short-term trading and arbitrage. http://saahpc.ncsa.illinois.edu/09/papers/Ganesan_paper.pdf /content/cudazone/CUDABrowser/assets/images/applications/1213_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1213_GPUComputing bgimg_large.png Academia Washington University in St. Louis 2009 06 29 06/29/2009 Narayan Ganesan Roger D. Chamberlain Jeremy Buhler Paper Narayan Ganesan,Roger D. Chamberlain,Jeremy Buhler,nganesan@wustl.edu,roger@wustl.edu,jbuhler@wustl.edu 52b3b800-2810-4e18-b631-e995c9a2ed48 GPU Accelerated Cardiac Electrophysiology Numerical simulations of cellular membranes are useful for both basic science and increasingly for clinical diagnostic and therapeutic applications. A common bottleneck in such simulations arises from solving large highly complex stiff systems of ordinary di fferential equations (ODEs) thousands of times for numerous collocation points (representing cells) throughout a three-dimensional volume. For some electrophysiology simulations, over 98% of the time is spent solving these systems of ODEs when run in serial on a single core. http://cseweb.ucsd.edu/groups/hpcl/scg/papers/2010/lionetti_ms_thesis.pdf /content/cudazone/CUDABrowser/assets/images/applications/1212_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1212_GPUComputing bgimg_large.png Academia University of California, San Diego 2010 04 15 04/15/2010 280 Fred Lionetti Paper Fred Lionetti 7dbbc8bd-6029-4692-911e-5a4e725ef349 HARNESSING THE POWER OF IDLE GPUS FOR ACCELERATION OF BIOLOGICAL SEQUENCE ALIGNMENT This paper presents a parallel system capable of accelerating biological sequence alignment on the graphics processing unit (GPU) grid. The GPU grid in this paper is a desktop grid system that utilizes idle GPUs and CPUs in the office and home. Our parallel implementation employs a master-worker paradigm to accelerate an OpenGLbased algorithm that runs on a single GPU. We integrate this implementation into a screensaver-based grid system that detects idle resources on which the alignment code can run. http://www-hagi.ist.osaka-u.ac.jp/research/papers/200912_ino_ppl.pdf /content/cudazone/CUDABrowser/assets/images/applications/1211_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1211_GPUComputing bgimg_large.png Academia Osaka Univeristy 2009 08 24 08/24/2009 FUMIHIKO INO YUKI KOTANI YUMA MUNEKAWA Paper FUMIHIKO INO,YUKI KOTANI,YUMA MUNEKAWA 701ae5ed-4273-4c9a-8546-55b23244a5ca Mixing Multi-Core CPUs and GPUs for Scientific Simulation Software Recent technological and economic developments have led to widespread availability of multi-core CPUs and specialist accelerator processors such as graphical processing units (GPUs). The accelerated computational performance possible from these devices can be very high for some applications paradigms. Software languages and systems such as NVIDIA's CUDA and Khronos consortium's open compute language (OpenCL) support a number of individual parallel application programming paradigms. To scale up the performance of some complex systems simulations however, a hybrid of multi-core CPUs for coarse-grained parallelism and very many core GPUs for data parallelism is necessary. http://www.massey.ac.nz/~dpplayne/Papers/cstn-091.pdf /content/cudazone/CUDABrowser/assets/images/applications/1210_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1210_GPUComputing bgimg_large.png Academia Massey University 2009 09 21 09/21/2009 K. A. Hawick A. Leist D. P. Playne Paper K. A. Hawick,A. Leist,D. P. Playne,k.a.hawick@massey.ac.nz,a.leist@massey.ac.nz,d.p.playne@massey.ac.nz 4a0b9cb0-a366-4a75-b2c4-e00c1af60a5c Computing on GPUs The increasing power of GPUs has led to the intent to transfer computing load from CPUs to GPUs. A first example has been the porting of computing intensive algorithms like e.g. ray-tracing algorithms form CPU to GPU. Through the Compute Unified Device Architecture (CUDA [4]) GPUs can also be used to increase computing speed for High Performance Computing applications. In this paper different parallelization strategies for different processor architectures are presented. They are compared and firt experiences using GPUs for a collection of numerical applications are given. /content/cudazone/CUDABrowser/assets/images/applications/1209_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1209_GPUComputing bgimg_large.png Commercial DYNAmore GmbH 2009 05 14 05/14/2009 Dr. Uli Gohner Paper Dr. Uli Gohner 3e365933-a682-4caa-aa9c-b165f30d448d High-Performance Physics Simulations Using Multi-Core CPUs and GPGPUs in a Volunteer Computing Context This paper presents two conceptually simple methods for parallelizing a Parallel Tempering Monte Carlo simulation in a distributed volunteer computing context, where computers belonging to the general public are used. The first method uses conventional multi-threading. The second method uses CUDA, a graphics card computing system. Parallel Tempering is described, and challenges such as parallel random number generation and mapping of Monte Carlo chains to different threads are explained. While conventional multi-threading on CPUs is well-established, GPGPU programming techniques and technologies are still developing and present several challenges, such as the effective use of a relatively large number of threads. http://arxiv.org/ftp/arxiv/papers/1004/1004.0023.pdf /content/cudazone/CUDABrowser/assets/images/applications/1208_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1208_GPUComputing bgimg_large.png Commercial D-Wave Systems Inc. 2010 03 31 03/31/2010 Kamran Karimi Neil G. Dickson Firas Hamze Paper Kamran Karimi ,Neil G. Dickson ,Firas Hamze,kkarimi,@dwavesys.com,ndickson@dwavesys.com,fhamze@dwavesys.com 1e8f6f3a-a7bd-4267-9fc9-3680f9bc0449 Accelerating Large-scale Convolutional Neural Networks with Parallel Graphics Multiprocessors Training convolutional neural networks (CNNs) on large sets of high-resolution images is too computationally intense to be performed on commodity CPUs. Such architectures however achieve state-of-the-art results on low-resolution machine vision tasks such as the recognition of handwritten characters. We have adapted the inherent multi-level parallelism of CNNs for Nvidia's CUDA GPU architecture to accelerate the training by two orders of magnitude. http://www.ais.uni-bonn.de/papers/nips09ws_scherer_behnke.pdf /content/cudazone/CUDABrowser/assets/images/applications/1206_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1206_GPUComputing bgimg_large.png University of Bonn, Germany 2009 12 01 12/01/2009 Dominik Scherer Sven Behnke Paper Dominik Scherer,Sven Behnke,scherer@ais.ni-bonn.de,behnke@cs.uni-bonn.de c0bb0403-84f7-42b9-8575-ec19bf6268d2 A Practical GPU Based KNN Algorithm The KNN algorithm is a widely applied method for classification in machine learning and pattern recognition. However, we can't be able to get a satisfactory performance in many applications, as the KNN algorithm has a high computational complexity. Recent developments in programmable, highly paralleled Graphics Processing Units (GPU) have opened a new era of parallel computing which deliver tremendous computational horsepower in a single chip. In this paper, we describe a practical GPU based K Nearest Neighbor (KNN) algorithm implemented by CUDA. http://www.academypublisher.com/proc/iscsct09/papers/iscsct09p151.pdf /content/cudazone/CUDABrowser/assets/images/applications/1205_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1205_GPUComputing bgimg_large.png Academia Soochow University 2009 12 26 12/26/2009 Quansheng Kuang Lei Zhao Paper Quansheng Kuang,Lei Zhao,kqs.net@163.com,zhaol@suda.edu.cn f51cd9bf-7067-4281-9b34-95671175e688 http://www.modelica.org/events/modelica2009/Proceedings/memorystick/pages/papers/0032/0032.pdf This work focuses on the use of parallel hardware to improve the simulation speed of equation-based object-oriented Modelica models. With this intention, a method has been developed that allows for the translation of a restricted class of Modelica models to parallel simulation code, targeted for the Nvidia Tesla architecture and based on the Quantized State Systems (QSS) simulation algorithm. http://www.modelica.org/events/modelica2009/Proceedings/memorystick/pages/papers/0032/0032.pdf /content/cudazone/CUDABrowser/assets/images/applications/1204_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1204_GPUComputing bgimg_large.png 2009 09 21 09/21/2009 Martina Maggio Kristian Stavaker Filippo Donida Francesco Casella Peter Fritzson Paper Martina Maggio,Kristian Stavaker,Filippo Donida 0648ea65-f71f-4795-a1be-579a9aa03e90 An efficient GPU implementation for large scale individual-based simulation of collective behavior In this work we describe a GPU implementation for an individual-based model for fish schooling. In this model each fish aligns its position and orientation with an appropriate average of its neighbors positions and orientations. This carries a very high computational cost in the so-called nearest neighbors search. By leveraging the GPU processing power and the new programming model called CUDA we implement an efficient framework which permits to simulate the collective motion of high-density individual groups. http://www.unibas.it/utenti/erra/Papers/HiBi09.pdf /content/cudazone/CUDABrowser/assets/images/applications/1203_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1203_GPUComputing bgimg_large.png Academia Universita della Basilicata 2009 10 01 10/01/2009 Ugo Erra Bernardino Frola Vittorio Scarano Iain Couzin Paper Ugo Erra,Bernardino Frola,Vittorio Scarano,ugo.erra@unibas.it,ber.frola@gmail.com,vitsca@dia.unisa.it 9dedab39-73d9-4c13-b5d2-2104819cec81 A Hybrid Analytical DRAM Performance Model As process technology scales, the number of transistors that can in a unit area has increased exponentially. Processor throughput, memory storage, and memory throughput have all been increasing at an exponential pace. As such, DRAM has become an ever-tightening bottleneck for applications with irregular memory access patterns. Computer architects in industry sometimes use ad hoc analytical modeling techniques in lieu of cycle-accurate performance simulation to identify critical design points. https://www.ece.ubc.ca/~aamodt/papers/gyuan.mobs2009.pdf /content/cudazone/CUDABrowser/assets/images/applications/1202_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1202_GPUComputing bgimg_large.png Academia University of British Columbia 2008 05 19 05/19/2008 George L. Yuan Tor M. Aamodt Paper George L. Yuan,Tor M. Aamodt,gyuan@ece.ubc.ca,aamodt@ece.ubc.ca df66c76c-4fda-468e-9eca-124cae57e3c4 Parallelisation of Fuzzy Inference on a Graphics Processor Unit Using the Compute Unified Device Architecture The inherently parallel nature of fuzzy inference is rarely exploited by fuzzy systems researchers. Hardware implementations, such as Field Programmable Gate Arrays (FPGAs), commonly use parallel architectures to achieve fast inference speeds. In this paper, we explore the use of Graphics Processor Units (GPUs) and NVIDIA‟s Compute Unified Device Architecture (CUDA) for fast inference speeds in a scalable and flexible Mamdani type fuzzy inference system (FIS). Our goal is to provide computational intelligence researchers the skills necessary to exploit the low cost and high performance of GPUs with a minimum learning cost. http://www.cci.dmu.ac.uk/ukci2008/papers/Parallelisation-of-Fuzzy-Inference-on-a-Graphics-Processor-Unit-Using-the-Compute-Unified-Device-Architecture.pdf /content/cudazone/CUDABrowser/assets/images/applications/1201_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1201_GPUComputing bgimg_large.png Academia University of Missouri, Columbia 2008 05 01 05/01/2008 Derek Anderson Simon Coupland Paper Derek Anderson,Simon Coupland 76dcd879-098a-41f2-ab03-38c43d2a042e GPU-Based Road Sign Detection Using Particle Swarm Optimization Road Sign Detection is a major goal of Advanced Driving Assistance Systems (ADAS). Since the dawn of this discipline, much work based on different techniques has been published which shows that traffic signs can be first detected and then classified in video sequences in real time. While detection is usually performed using classical computer vision techniques based on color and/or shape matching, most often classification is performed by neural networks. In this work we present a novel approach based on both sign shape and color which uses Particle Swarm Optimization (PSO) for detection. http://www.ce.unipr.it/~mussi/downloads/papers/mussiISDA09.pdf /content/cudazone/CUDABrowser/assets/images/applications/1200_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1200_GPUComputing bgimg_large.png Academia University of Parma 2009 11 01 11/01/2009 Luca Mussi Stefano Cagnoni Fabio Daolio Paper Luca Mussi,Stefano Cagnoni,Fabio Daolio,mussi@ce.unipr.it,cagnoni@ce.unipr.it,fabio.daolio@unil.ch f0e5e186-d65b-4d6b-9f18-fb153cfcf39a LARGE-SCALE PARALLEL MULTIBODY DYNAMICS WITH FRICTIONAL CONTACT In the context of simulating the frictional contact dynamics of large systems of rigid bodies, this paper reviews a novel method for solving large cone complementarity problems by means of a fixed-point iteration algorithm. The method is an extension of the Gauss-Seidel and Gauss-Jacobi methods with overrelaxation for symmetric convex linear complementarity problems. http://www.mcs.anl.gov/uploads/cels/papers/P1487.pdf /content/cudazone/CUDABrowser/assets/images/applications/1199_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1199_GPUComputing bgimg_large.png Academia University of Wisconsin Madison 2008 10 20 10/20/2008 Dan Negrut Alessandro Tasora Mihai Anitescu Paper Dan Negrut,Alessandro Tasora,Mihai Anitescu,negrut@wisc.edu,tasora@ied.unipr.it,anitescu@mcs.anl.gov abb411b1-54e9-4a4d-8f26-7acad6754856 A characterization and analysis of PTX kernels General purpose application development for GPUs (GPGPU) has recently gained momentum as a cost-effective approach for accelerating data- and compute-intensive applications. It has been driven by the introduction of C-based programming environments such as NVIDIA's CUDA [1], OpenCL [2], and Intel's Ct [3]. While significant effort has been focused on developing and evaluating applications and software tools, comparatively little has been devoted to the analysis and characterization of applications to assist future work in compiler optimizations, application re-structuring, and micro-architecture design. http://www.computer.org/portal/web/csdl/doi/10.1109/IISWC.2009.5306801 /content/cudazone/CUDABrowser/assets/images/applications/1198_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1198_GPUComputing bgimg_large.png Academia Georgia Institute of Technology 2009 05 05 05/05/2009 Andrew Kerr Gregory Diamos Sudhakar Yalamanchili Paper Andrew Kerr,Gregory Diamos,Sudhakar Yalamanchili ed43c757-50fe-4151-a1c4-21184ce71dbd General Purpose Computation on Graphics Processing Units (GPGPU) using CUDA Graphics processing units (GPUs) are special processors which traditionally were used to accelerate computer graphics by offloading work from the CPU. Today, GPUs are highly parallel many-core processors which enable general-purpose computation on graphics processing units (GPGPU). GPGPU has already been an issue since 2002 but a huge interest did not evolve until Nvidia released the CUDA platform in 2007. Developers and researchers started to use CUDA for parallel programming. http://www.wi.uni-muenster.de/pi/lehre/ws0910/pppa/papers/gpgpu.pdf /content/cudazone/CUDABrowser/assets/images/applications/1195_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1195_GPUComputing bgimg_large.png Academia Westfalische Wilhelms-Universitat 2009 12 20 12/20/2009 Alexander Zibula Paper Alexander Zibula 993dd63c-de2a-49e1-81ba-ade7f1682b25 Simulation of one-layer shallow water systems on multicore and CUDA architectures The numerical solution of shallow water systems is useful for several applications related to geophysical flows but the big dimensions of the domains suggests the use of powerful accelerators to obtain numerical results in reasonable times. This paper addresses how to speed up the numerical solution of a first order well-balanced finite volume scheme for 2D one-layer shallow water systems by using modern Graphics Processing Units (GPUs) supporting the NVIDIA CUDA programming model. http://lsi.ugr.es/~jmantas/papers/supercomputing09.pdf /content/cudazone/CUDABrowser/assets/images/applications/1194_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1194_GPUComputing bgimg_large.png Academia 1 Universidad de Granada 2Universidad de Malaga 2010 03 01 03/01/2010 Marc de la Asuncion1 Jose M. Mantas1 Manuel J. Castro2 Paper Marc de la Asuncion1,Jose M. Mantas1,Manuel J. Castro2 78e3f6d6-b219-43a8-8c2d-f515465c3670 IDN_MFC Image denoising with bilateral filter algorithms /content/cudazone/CUDABrowser/assets/images/applications/1193_64638_Application_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1193_64638_Application_large.jpg Academia Wlroclaw University of Technology 2010 06 22 06/22/2010 100 Wojciech Korycki Application Paper Signal Processing Bilateral Filter denoising,Wojciech Korycki,wojciech.korycki@gmail.com 9a370a84-4d82-4133-b523-6f56cca33568 Hypercubic Storage Layout and Many simulations in the physical sciences are expressed in terms of rectilinear arrays of variables. It is attractive to develop such simulations for use in 1-, 2-, 3- or arbitrary physical dimensions and also in a manner that supports exploitation of data-parallelism on fast modern processing devices.We report on data layouts and transformation algorithms that support both conventional and data-parallel memory layouts. http://tur-www1.massey.ac.nz/~dpplayne/Papers/cstn-096.pdf /content/cudazone/CUDABrowser/assets/images/applications/1192_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1192_GPUComputing bgimg_large.png Academia Massey University 2009 06 23 06/23/2009 K. A. Hawick D. P. Playne Paper K. A. Hawick,D. P. Playne f33d13b3-5699-4fb2-b908-98d32866aa20 Analyzing CUDA Workloads Using a Detailed GPU Simulator Modern Graphic Processing Units (GPUs) provide sufficiently flexible programming models that understanding their performance can provide insight in designing tomorrow's manycore processors, whether those are GPUs or otherwise. The combination of multiple, multithreaded, SIMD cores makes studying these GPUs useful in understanding tradeoffs among memory, data, and thread level parallelism. While modern GPUs offer orders of magnitude more raw computing power than contemporary CPUs, many important applications, even those with abundant data level parallelism, do not achieve peak performance. https://www.ece.ubc.ca/~aamodt/papers/gpgpusim.ispass09.pdf /content/cudazone/CUDABrowser/assets/images/applications/1191_GPUComputing bgimg_small.png /content/cudazone/CUDABrowser/assets/images/applications/1191_GPUComputing bgimg_large.png Academia University of British Columbia 2009 03 01 03/01/2009 Ali Bakhoda George L. Yuan Wilson W. L. Fung Henry Wong Tor M. Aamodt Paper Ali Bakhoda,George L. Yuan,Wilson W. L. Fung,Henry Wong,Tor M. Aamodt c1d20c0c-79f9-44f0-9331-291ccbeb0ee7 Phase Based Volume Registration Using CUDA We have implemented phase based volume registration using CUDA, in contrast to all other GPU based image registration implementations that are based on the image intensity. Our registration algorithm is more robust for volumes that differ significantly in intensity. This work was presented at the IEEE conference ICASSP in Dallas 2010. /content/cudazone/CUDABrowser/assets/images/applications/1189_449881_phase_based_volume_registration_using_CUDA_small.png /content/cudazone/CUDABrowser/assets/images/applications/1189_449881_phase_based_volume_registration_using_CUDA_large.png Academia Linkoping university http://www.moviii.isy.liu.se 2010 06 22 06/22/2010 30 Anders Eklund Mats Andersson Hans Knutsson Paper Medical Imaging Image registration, local phase,Anders Eklund,Mats Andersson,Hans Knutsson,andek@imt.liu.se,matsa@imt.liu.se,knutte@imt.liu.se f9f4cdda-fab7-40fa-b8bc-43482f378a81 Towards a Software Transactional Memory for Graphics Processors The introduction of general purpose computing on many-core graphics processor systems, and the general shift in the industry towards parallelism, has created a demand for ease of parallelization. Software transactional memory (STM) simplifies development of concurrent code by allowing the programmer to mark sections of code to be executed concurrently and atomically in an optimistic manner. In contrast to locks, STMs are easy to compose and do not suffer from deadlocks. We have designed and implemented two STMs for graphics processors, one blocking and one non-blocking. The design issues involved in the development of these two STMs are described and explained in the paper together with experimental results comparing the performance of the two STMs. /content/cudazone/CUDABrowser/assets/images/applications/1188_7612_cudazonestm_small.png /content/cudazone/CUDABrowser/assets/images/applications/1188_7612_cudazonestm_large.png Academia Chalmers University of Technology http://www.chalmers.se 2010 04 14 04/14/2010 Daniel Cederman Philippas Tsigas Muhammad Tayyab Chaudhry Paper Programming Tools Daniel Cederman,Philippas Tsigas,cederman@chalmers.se,tsigas@chalmers.se a8c7fb3f-0fb8-40d2-8440-c9dedecf7051 nexiwave Speech Indexing nexiwave 2.0 the GPU Assisted Speech Indexing /content/cudazone/CUDABrowser/assets/images/applications/1187_13933_nexilogo_betawith_snowflakes_small.png /content/cudazone/CUDABrowser/assets/images/applications/1187_13933_nexilogo_betawith_snowflakes_large.png Commercial nexiwave.com http://nexiwave.com 2010 06 03 06/03/2010 75 Commercial Ben Jiang Application Signal Processing Speech Indexing Speech Indexing, GPU,Ben Jiang,ben@nexiwave.com a42593ae-afe0-46ed-85d3-7a1ab25c93ac Massive Bayesian Mixture Modelling This paper describes advances in statistical computation for large-scale data analysis in structured Bayesian mixture models via graphics processing unit (GPU) programming. The developments are partly motivated by computational challenges arising in fitting models of increasing heterogeneity to increasingly large data sets. An example context concerns common biological studies using high-throughput technologies generating many, very large data sets and requiring increasingly high-dimensional mixture models with large numbers of mixture components. /content/cudazone/CUDABrowser/assets/images/applications/1186_179430_cfse_clusters_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1186_179430_cfse_clusters_large.jpg Academia UCLA and Duke University 2010 03 01 03/01/2010 160 Open source Marc A. Suchard Quanli Wang Cliburn Chan Jacob Frelinger Andrew Cron Mike West Paper Code Numerics Life Sciences Science Computational statistics,Marc A. Suchard,Quanli Wang,Cliburn Chan, Jacob Frelinger, Andrew Cron, Mike West,msuchard@ucla.edu,mw@stat.duke.edu b5e57af8-695e-4550-9a42-a30be2716079 Accelerating Quadrature Methods for Option Valuation This paper presents an architecture for FPGA acceleration of quadrature methods used for pricing complex options, such as discrete barrier, Bermudan, and American options. The architecture can be optimized for speed and power consumption by exploiting pipelining and parallelism to produce efficient implementations in reconfigurable logic. An optimised implementation using Graphics Processing Units (GPUs) is also developed, to provide a performance and efficiency comparison with an FPGA accelerator. http://www.computer.org/portal/web/csdl/doi/10.1109/FCCM.2009.36 /content/cudazone/CUDABrowser/assets/images/applications/1185_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1185_logo_CS Digital Library_large.jpg Academia Imperial College London 09 04 01 04/01/09 Anson H. T. Tse David B. Thomas Wayne Luk Paper Anson H. T. Tse,David B. Thomas,Wayne Luk 934fb2f1-7ee6-4958-bb7d-b4ed38debaee High Resolution Program Flow Visualization of Hardware Accelerated Hybrid Multi-core Applications The advent of multi-core processors has made parallel computing techniques mandatory on main stream systems. With the recent rise of hardware accelerators, hybrid parallelism adds yet another dimension of complexity to the process of software development. This article presents a tool for graphical program flow analysis of hardware accelerated parallel programs. http://www.computer.org/portal/web/csdl/doi/10.1109/CCGRID.2010.27 /content/cudazone/CUDABrowser/assets/images/applications/1184_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1184_logo_CS Digital Library_large.jpg 2010 05 01 05/01/2010 Daniel Hackenberg Guido Juckeland Holger Brunst Paper Daniel Hackenberg,Guido Juckeland,Holger Brunst d0ffd6f3-3337-4dbe-a796-0c7d19b1cd6e An Analysis of GPU Parallel Computing Parallel systems are becoming ubiquitous in the world of computing as evidenced by multi-core processors, heterogeneous Cell broadband engine, and highly parallel graphics processing units (GPUs). All parallel systems share a requirement that parallel programming is necessary to leverage multiple cores. As a result of this trend, multi-core CPUs are no longer a clear winner due to its peaked clock frequency and programming effort involved in parallelizing code for multi-core architecture. Given such drawbacks, dataparallel applications might benefit from GPU assisted computing. http://www.computer.org/portal/web/csdl/doi/10.1109/HPCMP-UGC.2009.59 /content/cudazone/CUDABrowser/assets/images/applications/1183_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1183_logo_CS Digital Library_large.jpg Research U. S. Army Research Laboratory 2009 06 01 06/01/2009 Song Jun Park Paper Song Jun Park 79cc1232-9c03-4eca-ac3f-dd9b3743fac0 Tiling for Performance Tuning on Different Models of GPUs The strategy of using CUDA-compatible GPUs as a parallel computation solution to improve the performance of programs has been more and more widely approved during the last two years since the CUDA platform was released. Its benefit extends from the graphic domain to many other computationally intensive domains. Tiling, as the most general and important technique, is widely used for optimization in CUDA programs. New models of GPUs with better compute capabilities have, however, been released, new versions of CUDA SDKs were also released. http://www.computer.org/portal/web/csdl/doi/10.1109/ISISE.2009.60 /content/cudazone/CUDABrowser/assets/images/applications/1182_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1182_logo_CS Digital Library_large.jpg Academia Hong Kong University of Science and Technology 2002 11 01 11/01/2002 Chang Xu Steven R. Kirk Samantha Jenkins Paper Chang Xu,Steven R. Kirk,Samantha Jenkins e5e749ac-945e-42f2-a238-1209f8986eb2 Acceleration of Medical Image Registration Using Graphics Process Units in Computing Normalized Mutual Information This paper presents a computational performance analysis of an accelerated medical image registration using Graphics Processing Units (GPUs). In our previous work, a multi-resolution approach using normalized mutual information (NMI) has proven to be useful in medical image registration. In this paper, we propose an acceleration of the NMI procedure using GPU implementation because of the parallel processing capabilities. http://www.computer.org/portal/web/csdl/doi/10.1109/ICIG.2009.48 /content/cudazone/CUDABrowser/assets/images/applications/1181_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1181_logo_CS Digital Library_large.jpg Academia Kent State University 2009 09 01 09/01/2009 Wei-Hung Cheng Cheng-Chang Lu Paper Wei-Hung Cheng,Cheng-Chang Lu 705aaab7-bfe3-4433-ae2d-e1490bf77dbb MAX-MIN Ant System on GPU with CUDA We propose a parallel MAX-MIN Ant System (MMAS) algorithm that is suitable for an implementation on graphics processing units (GPUs). Multi ant colonies with respective parameter settings are whole offloaded to the GPU in parallel. We have implemented this GPU-based MMAS on the GPU with compute unified device architecture (CUDA). Some performance optimization means for kernel program of GPU are introduced. Experimental results that are based on simulations for the traveling salesperson problem are presented to evaluate the proposed techniques. /content/cudazone/CUDABrowser/assets/images/applications/1180_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1180_logo_CS Digital Library_large.jpg 2009 12 07 12/07/2009 Hongtao Bai Dantong Ou Yang Ximing Li Lili He Haihong Yu Paper Hongtao Bai,Dantong Ou Yang,Ximing Li f3d7658d-4572-4345-a1bd-3fe05ca6ce37 Scene Recognition Acceleration Using CUDA and OpenMP Scene recognition has become a remarkable field in image processing area, and many methods have been proposed in recent years, in which the idea of extracting the scene gist from global features has been proved to have higher retrieval accuracy compared with many other methods. However, the process of extracting gist is heavily time-consuming and not suitable for real-time application. In this paper, the CUDA architecture is deployed to accelerate this process. http://www.computer.org/portal/web/csdl/doi/10.1109/ICISE.2009.1045 /content/cudazone/CUDABrowser/assets/images/applications/1179_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1179_logo_CS Digital Library_large.jpg Academia Dalian University of Technology 2009 12 01 12/01/2009 Yuxin Wang Zhen Feng He Guo Changqin He Yuansheng Yang Paper Yuxin Wang,Zhen Feng,He Guo b717867c-c959-4577-b6a5-afbc0e42fdae A Stream Processor Cluster Architecture Model with the Hybrid Technology of MPI and CUDA Nowadays, the compute capability of traditional cluster system can't keep up with the computing needs of a practical application, and these aspects of energy, space technology, etc. have become a huge problem. However, as parallel computing equipment, the stream processor (SP) has a high performance of floating-point operations. NVIDIA GPUs is a typical stream processor device, CUDA technology enables the way to develop a better parallel program on GPUs to become flexible. In this paper, we make use of the hybrid parallel computing programming environment (HPCPE) with MPI and CUDA technology to build the simple CPU + GPU-based stream processor cluster system. http://www.computer.org/portal/web/csdl/doi/10.1109/ICISE.2009.171 /content/cudazone/CUDABrowser/assets/images/applications/1177_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1177_logo_CS Digital Library_large.jpg Academia University of Shanghai for Science and Technology 2009 12 01 12/01/2009 Qing-kui Chen Jia-kang Zhang Paper Qing-kui Chen,Jia-kang Zhang 5da86d39-2d35-4792-a0b5-e400e3383959 Formal Description and Optimization Based High - Performance Computing on CUDA In recent years, with the development of GPU, based on the general purpose computation on graphics processors has became a new field. Aiming at the processing of GPU, this paper provides the formal description for data parallel mode, a detailed description of the CUDA programming mode land the principle of optimization. It shows by the comparative experiment that CUDA owns strongly of the ability to the parallel processing and provides new methods and ideas to GPGPU. /content/cudazone/CUDABrowser/assets/images/applications/1176_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1176_logo_CS Digital Library_large.jpg Academia Hong Kong University of Science and Technology 2009 12 01 12/01/2009 Bo Li Huacheng Zhao JingJing Liang Paper Bo Li,Huacheng Zhao,JingJing Liang 2287f161-8378-4190-ae1f-bd428d9ca3c3 Password Recovery for RAR Files Using CUDA Driven by the insatiable demand of real-time graphics, especially from the market of computer games, Graphics Processing Unit (GPU) is becoming a major computing horsepower during recent years since the performance of GPU is surpassing that of the contemporary CPU. This paper presents our study on how to efficiently recover the passwords for encrypted RAR files. http://www.computer.org/portal/web/csdl/doi/10.1109/DASC.2009.123 /content/cudazone/CUDABrowser/assets/images/applications/1175_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1175_logo_CS Digital Library_large.jpg 2009 12 01 12/01/2009 Guang Hu Jianhua Ma Benxiong Huang Paper Guang Hu,Jianhua Ma,Benxiong Huang 2b933813-11b7-4f81-8e63-9ce67eba045f Password Recovery for RAR Files Using CUDA Driven by the insatiable demand of real-time graphics, especially from the market of computer games, Graphics Processing Unit (GPU) is becoming a major computing horsepower during recent years since the performance of GPU is surpassing that of the contemporary CPU. This paper presents our study on how to efficiently recover the passwords for encrypted RAR files. Our research focus is on the AES key generation processing, which is the most time consuming stage in the whole RAR encryption/decryption process. http://www.computer.org/portal/web/csdl/doi/10.1109/DASC.2009.123 /content/cudazone/CUDABrowser/assets/images/applications/1174_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1174_logo_CS Digital Library_large.jpg 2009 12 01 12/01/2009 Guang Hu Jianhua Ma Benxiong Huang Paper Guang Hu,Jianhua Ma,Benxiong Huang f928f65a-86c9-4a31-8299-3e40f02d03fa GPU-Assisted Computation of Centroidal Voronoi Tessellation Centroidal Voronoi tessellations (CVT) are widely used in computational science and engineering. The most commonly used method is Lloyd's method, and recently the L-BFGS method is shown to be faster than Lloyd's method for computing the CVT. However, these methods run on the CPU and are still too slow for many practical applications. We present techniques to implement these methods on the GPU for computing the CVT on 2D planes and on surfaces, and demonstrate significant speedup of these GPU-based methods over their CPU counterparts. http://www.computer.org/portal/web/csdl/doi/10.1109/TVCG.2010.53 /content/cudazone/CUDABrowser/assets/images/applications/1173_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1173_logo_CS Digital Library_large.jpg Academia University of Texas at Dallas 2010 03 16 03/16/2010 Guodong Rong Yang Liu Wenping Wang Xiaotian Yin David Gu Xiaohu Guo Paper Guodong Rong,Yang Liu,Wenping Wang,Xiaotian Yin,David Gu,Xiaohu Guo 7e3424c0-b0ed-4476-8830-eb60da8a80c7 Designing Efficient Many-Core Parallel Algorithms for All-Pairs Shortest-Paths Using CUDA Finding the all-pairs shortest-paths on a large graph is a fundamental problem in many practical applications such as bioinformatics, internet node traffic and network routing. In this paper, we present the designs of two efficient parallel algorithms for many-core GPUs using CUDA. Our algorithms expose substantial fine-grained parallelism while maintaining minimal global communication. By using the global scope of the GPU's global memory, coalescing the global memory reads and writes, and avoiding on-chip shared memory bank conflicts, we are able to achieve a large performance benefit with a speed-up of 2,500x on a desktop computer in comparison with a single core program. http://www.computer.org/portal/web/csdl/doi/10.1109/ITNG.2010.230 /content/cudazone/CUDABrowser/assets/images/applications/1172_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1172_logo_CS Digital Library_large.jpg Academia Lamar University 2010 04 01 04/01/2010 Quoc-Nam Tran Paper Quoc-Nam Tran 0315bfde-f758-4fc6-8cf5-85aac810ca12 Record Setting Software Implementation of DES Using CUDA The increase in computational power of off-the-shelf hardware offers more and more advantageous tradeoffs among efficiency, cost and availability, thus enhancing the feasibility of of cryptanalytic attacks aiming to lower the security of widely used cryptosystems. In this paper we illustrate an GPU-based software implementation of the most efficent variant of Data Encryption Standard (DES), showing the performance of a software breaker which effectively exploits the multi-core Nvidia GT200 graphic architecture. http://www.computer.org/portal/web/csdl/doi/10.1109/ITNG.2010.43 /content/cudazone/CUDABrowser/assets/images/applications/1171_logo_CS Digital Library_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1171_logo_CS Digital Library_large.jpg 2010 04 01 04/01/2010 Giovanni Agosta Allessandro Barenghi Fabrizio De Santis Gerardo Pelosi Paper Giovanni Agosta,Allessandro Barenghi,Fabrizio De Santis 7df8d14f-4c52-460a-8881-ad932fd45292 Eye-Full Tower: A GPU-based variable multibaseline omnidirectional stereovision system with automatic baseline selection for outdoor mobile robot navigation In recent years, it can be observed that there is a gradual increase in the number of researchers and projects involved with the development of omnidirectional vision systems for various applications. The primary factors, which contributed towards this positive trend, are the availability of inexpensive and high resolution vision sensors, robust and fast computers and the advantages of using such systems over perspective vision systems. In this paper, a novel variable multibaseline omnidirectional stereovision system is presented. http://portal.acm.org/citation.cfm?id=1805342.1805504&coll=Portal&dl=GUIDE&CFID=92176503&CFTOKEN=39358289 /content/cudazone/CUDABrowser/assets/images/applications/1170_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1170_logo_acm_portal2_large.jpg Academia Monash University 2010 06 01 06/01/2010 Wen Lik Dennis Lui Ray Jarvis Paper Wen Lik Dennis Lui,Ray Jarvis 56e6ce9f-ae50-427d-8d3f-23b1e24c6683 Optimized high speed pixel sorting and its application in watershed based image segmentation Efficient sorting of image pixels based on their grayscale value is traditionally implemented using an algorithm based on distribution or counting sort methods. We show that an elegant alternative can be used which outperforms the traditional method both in terms of processing speed and main memory access. We discuss both theoretically analyzed and real-life performance results, and demonstrate the improvements that can be obtained when our algorithm is combined with a well-known watershed image segmentation method. /content/cudazone/CUDABrowser/assets/images/applications/1169_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1169_logo_acm_portal2_large.jpg Research The National Institute for Criminalistics and Criminology (NICC) 2010 07 01 07/01/2010 Patrick De Smet Paper Patrick De Smet 729f484e-7759-401b-a6ef-78c85f290bd6 GPU-accelerated molecular dynamics simulation for study of liquid crystalline flows We have developed a GPU-based molecular dynamics simulation for the study of flows of fluids with anisotropic molecules such as liquid crystals. An application of the simulation to the study of macroscopic flow (backflow) generation by molecular reorientation in a nematic liquid crystal under the application of an electric field is presented. The computations of intermolecular force and torque are parallelized on the GPU using the cell-list method, and an efficient algorithm to update the cell lists was proposed. Some important issues in the implementation of computations that involve a large number of arithmetic operations and data on the GPU that has limited high-speed memory resources are addressed extensively. http://portal.acm.org/citation.cfm?id=1808372.1808870&coll=Portal&dl=GUIDE&CFID=92176503&CFTOKEN=39358289 /content/cudazone/CUDABrowser/assets/images/applications/1168_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1168_logo_acm_portal2_large.jpg Academia Kochi University of Technology 2010 08 01 08/01/2010 Alfeus Sunarso Tomohiro Tsuji Shigeomi Chono Paper Alfeus Sunarso,Tomohiro Tsuji,Shigeomi Chono 2b5f83d2-9afa-42a5-8582-8a9b3ae48841 Parallel implementation of wavelet-based image denoising on programmable PC-grade graphics hardware The discrete wavelet transform (DWT) has been extensively used for image compression and denoising in the areas of image processing and computer vision. However, the intensive computation of DWT due to its inherent multilevel data decomposition and reconstruction operations brings a bottleneck that drastically reduces its performance and implementations for real-time applications when facing large size digital images and/or high-definition videos. Although various software-based acceleration solutions, such as the lifting scheme, have been devised and achieved a higher performance in general, the pure software accelerated DWT still struggle to cope with the demands from real-time and interactive applications. With the growing capacity and popularity of graphics hardware, personal computers (PCs) nowadays are often equipped with programmable graphics processing units (GPUs) for graphics acceleration. The GPU offers a cost-effective parallel data processing mechanism for operations on large amount of data, even for applications beyond graphics. This practice is commonly referred as general-purpose computing on GPU (GPGPU). http://portal.acm.org/citation.cfm?id=1786816.1787181&coll=Portal&dl=GUIDE&CFID=92176503&CFTOKEN=39358289 /content/cudazone/CUDABrowser/assets/images/applications/1167_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1167_logo_acm_portal2_large.jpg Academia University of Huddersfield 2010 08 01 08/01/2010 Yang Su Zhijie Xu Paper Yang Su,Zhijie Xu a4486be2-fdef-4db0-89b0-879b296f6681 GPU Computing for Atmospheric Modeling Experience with a small kernel and implications for a full model Much success has been achieved using GPUs to accelerate existing applications that are highly data parallel, or that are dominated by small, intense computational kernels. What are the prospects for porting existing large scientific models that do not fit this mold? We take an expensive routine from the CAM atmosphere model, and port it to a GPU using CUDA. We use the experience gained as a guide in thinking about porting the full application to an accelerator based system. We consider the best path forward for getting large scientific models running on accelerator based systems, and identify cases where porting may be feasible, and where a complete redesign may be the best option. /content/cudazone/CUDABrowser/assets/images/applications/1166_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1166_logo_xplore_large.gif Research National Center for Atmospheric Research 2010 02 04 02/04/2010 Rory Kelly Paper Rory Kelly ce8e0150-e62b-4a0c-890b-0442b2e058a6 Design and Performance Evaluation of Image Processing Algorithms on GPUs In this paper, we construe key factors in design and evaluation of image processing algorithms on the massive parallel GPU (graphics processing units) using the CUDA (compute unified device architecture) programming model. A set of metrics, customized for image processing, are proposed to quantitatively evaluate algorithm characteristics. In addition, we show that a range of image processing algorithms map readily to CUDA using multiview stereo matching, linear feature extraction, JPEG2000 image encoding, and non-photorealistic rendering (NPR) as our example applications. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5477417&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26pageNumber%3D2%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1165_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1165_logo_xplore_large.gif Academia Inha University 2010 06 03 06/03/2010 In Kyu Park Nitin Singhal Man Hee Lee Sungdae Cho Chris Kim Paper In Kyu Park,Nitin Singhal,Man Hee Lee 750d6910-dfb0-4eee-9c8e-cec1320d7f09 CUDA-Based Linear Solvers for Stable Fluids In the field of computer graphics, physically-based fluids simulations (i.e., simulations that solve the equations that govern fluids behaviour) are performed using, among others, Stam's stable fluids method. This method requires the solution of two sparse linear systems that can be solved using an iterative solver (e.g., Jacobi, Gauss-Seidel, conjugate gradient, etc.). Focusing on real-time 3D applications, we provide and analyze the performance of the parallel GPU-based (using CUDA) algorithms of the Jacobi, Gauss-Seidel, and conjugate gradient solvers. /content/cudazone/CUDABrowser/assets/images/applications/1164_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1164_logo_xplore_large.gif 2010 04 21 04/21/2010 Goncalo Amador Abel Gomes Paper Goncalo Amador,Abel Gomes d43dc906-8a5f-4bb4-bcb2-006f2d9be085 Implementation of Variable Preconditioned GCR with mixed precision on GPU using CUDA The Variable Preconditioned GVR (VPGCR) with mixed precision on Graphics Processing Unit (GPU) using Compute Unified Device Architecture (CUDA) is numerically investigated. The convergence theorem of VPGCR is guaranteed that the residual equation for the preconditioned procedure can be solved in the range of single precision operation. The results of computations show that VPGCR with mixed precision operation on GPU demonstrated significant achievement than that of CPU. Especially, VPGCR on GPU with mixed precision operation is 22.53 times faster than that of Central Processing Unit (CPU). /content/cudazone/CUDABrowser/assets/images/applications/1163_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1163_logo_xplore_large.gif Academia Tokyo University of Technology 2010 05 09 05/09/2010 Soichiro Ikuno Norihisa Fujita Susumu Yamamoto Susumu Nakata Paper Soichiro Ikuno,Norihisa Fujita,Susumu Yamamoto 55d72cf8-150e-4281-be46-1a00cd588e1e A CUDA-Based Implementation of Stable Fluids in 3D with Internal and Moving Boundaries Fluid simulation has been an active research field in computer graphics for the last 30 years. Stam's stable fluids method, among others, is used for solving the equations that govern fluids (i.e. Navier-Stokes equations). An implementation of stable fluids in 3D using NVIDIA Compute Unified Architecture, shortly CUDA, is provided in this paper. This CUDA-based implementation also features the accurate physical treatment of internal (i.e. static boundaries inside the simulation domain) and moving boundaries. The performance gains of the presented implementation vs a sequential CPU-based implementation, and points of further improvement are also addressed. /content/cudazone/CUDABrowser/assets/images/applications/1161_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1161_logo_xplore_large.gif 2010 03 23 03/23/2010 Goncalo Amador Abel Gomes Paper Goncalo Amador,Abel Gomes d41f4ce3-8c86-4dd4-835b-8954c9caef44 Hybrid Core Acceleration of UWB SIRE Radar Signal Processing To move High Performance Computing (HPC) closer to forward operating environments and missions, the Army Research Laboratory is developing approaches using hybrid, asymmetric core computing. By blending capabilities found in Graphics Processing Units (GPUs) and traditional von Neumann multi-core Central Processing Units (CPUs), approaches are being developed and optimized to provide at or near real-time processing speeds for research project applications. Algorithms are designed to partition work to resources best designed to handle the processing load. The use of commodity resources allows the design to be flexible throughout the life-cycle without the costly and time-consuming delays associated with Application Specific Integrated Circuit (ASIC) development. This paradigm allows for rapid technology transfer to end users. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5477419&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26pageNumber%3D2%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1160_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1160_logo_xplore_large.gif Research U. S. Army Researc Laboratory 2010 06 03 06/03/2010 Song Jun Park James Ross Dale Shires David Richie Brian Henz Lam Nguyen Paper Song Jun Park,James Ross,Dale Shires 78482823-4944-4b0a-ab95-15e1aad00454 Optimal loop unrolling for GPGPU programs Graphics Processing Units (GPUs) are massively parallel, many-core processors with tremendous computational power and very high memory bandwidth. With the advent of general purpose programming models such as NVIDIA's CUDA and the new standard OpenCL, general purpose programming using GPUs (GPGPU) has become very popular. However, the GPU architecture and programming model have brought along with it many new challenges and opportunities for compiler optimizations. One such classical optimization is loop unrolling. Current GPU compilers perform limited loop unrolling. In this paper, we attempt to understand the impact of loop unrolling on GPGPU programs. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470423&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26pageNumber%3D2%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1159_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1159_logo_xplore_large.gif Academia The Ohio State University 2010 04 19 04/19/2010 Giridhar Murthy Sreenivasa Mahesh Ravishankar Muthu Manikandan Baskaran P. Sadayappan Paper Giridhar Murthy Sreenivasa,Mahesh Ravishankar,Muthu Manikandan Baskaran 86fc0781-21d5-4b31-9fb0-7061d02a703b Using CUDA enabled FDTD simulations to solve multi-gigahertz EMI challenges Thanks to the application of GPU-CUDA acceleration technology to EM simulation tools, more and more complicated EMI challenges can be efficiently investigated and solved very early in the design process. This paper presents a novel methodology to predict EMI emission due to memory SSO noise from a real, commercial graphics card by means of a commercially available CUDA accelerated full-wave FDTD simulator. It is shown that thanks to the CUDA acceleration one can estimate the influence of on-board decoupling capacitors on the EMI emission within hours. /content/cudazone/CUDABrowser/assets/images/applications/1158_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1158_logo_xplore_large.gif Research KHBO Flanders Mechatronics Engineering Centre 2010 04 12 04/12/2010 Davy Pissoort Chen Wang Hany Fahmy Paper Davy Pissoort,Chen Wang,Hany Fahmy 860722bd-ba53-4391-9e5c-7197a5574713 Dynamic load balancing on single- and multi-GPU systems The computational power provided by many-core graphics processing units (GPUs) has been exploited in many applications. The programming techniques currently employed on these GPUs are not sufficient to address problems exhibiting irregular, and unbalanced workload. The problem is exacerbated when trying to effectively exploit multiple GPUs concurrently, which are commonly available in many modern systems. In this paper, we propose a task-based dynamic load-balancing solution for single-and multi-GPU systems. The solution allows load balancing at a finer granularity than what is supported in current GPU programming APIs, such as NVIDIA's CUDA. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470413&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26pageNumber%3D2%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1157_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1157_logo_xplore_large.gif Academia University of Delaware 2010 04 19 04/19/2010 Long Chen Oreste Villa Sriram Krishnamoorthy Paper Long Chen,Oreste Villa,Sriram Krishnamoorthy 8bc36492-7c2b-4e62-ba6d-a9664ee84f10 Automatic Generation of Multi-Core Chemical Kernels This work presents KPPA (the Kinetics PreProcessor: Accelerated), a general analysis and code generation tool that achieves significantly reduced time-to-solution for chemical kinetics kernels on three multi-core platforms: NVIDIA GPUs using CUDA, the Cell Broadband Engine, and Intel Quad-Core Xeon CPUs. A comparative performance analysis of chemical kernels from WRF-Chem and the Community Multiscale Air Quality Model (CMAQ) is presented for each platform in double and single precision on coarse and fine grids. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5473221&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26pageNumber%3D2%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1156_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1156_logo_xplore_large.gif Academia Virginia Polytechnic Institute and State University 2010 05 27 05/27/2010 John Linford John Michalakes Manish Vachharajani Adrian Sandu Paper J. Linford,J. Michalakes,M. Vachharajani f0ed0b4c-5d78-4283-b786-d977d462b699 Speculative execution on multi-GPU systems The lag of parallel programming models and languages behind the advance of heterogeneous many-core processors has left a gap between the computational capability of modern systems and the ability of applications to exploit them. Emerging programming models, such as CUDA and OpenCL, force developers to explicitly partition applications into components (kernels) and assign them to accelerators in order to utilize them effectively. An accelerator is a processor with a different ISA and micro-architecture than the main CPU. These static partitioning schemes are effective when targeting a system with only a single accelerator. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470427&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26pageNumber%3D2%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1155_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1155_logo_xplore_large.gif Academia Georgia Institute of Technology 2010 04 19 04/19/2010 Gregory Diamos Sudhakar Yalamanchili Paper Gregory Diamos,Sudhakar Yalamanchili 829884ae-a849-4ab7-a5db-1ebb4290798a AUTO-GC: Automatic translation of data mining applications to GPU clusters Because of the very favorable price to performance ratio of the GPUs, a popular parallel programming configuration today is a cluster of GPUs. However, extracting performance on such a configuration would typically require programming in both MPI and CUDA, thus requiring a high degree of expertise and effort. It is clearly desirable to be able to support higher-level programming of this emerging high-performance computing platform. /content/cudazone/CUDABrowser/assets/images/applications/1154_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1154_logo_xplore_large.gif Academia The Ohio State University 2010 04 19 04/19/2010 Wenjing Ma Gagan Agrawal Paper Wenjing Ma,Gagan Agrawal 4401d416-56a2-4a55-88a0-a8ccbb66c75d Pricing of cross-currency interest rate derivatives on Graphics Processing Units We present a Graphics Processing Unit (GPU) parallelization of the computation of the price of cross-currency interest rate derivatives via a Partial Differential Equation (PDE) approach. In particular, we focus on the GPU-based parallel computation of the price of long-dated foreign exchange interest rate hybrids, namely Power Reverse Dual Currency (PRDC) swaps with Bermudan cancelable features. We consider a three-factor pricing model with foreign exchange skew which results in a time-dependent parabolic PDE in three spatial dimensions. Finite difference methods on uniform grids are used for the spatial discretization of the PDE, and the Alternating Direction Implicit (ADI) technique is employed for the time discretization. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470708&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1153_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1153_logo_xplore_large.gif Academia University of Toronto 2010 04 19 04/19/2010 Duy Minh Dang Paper Duy Minh Dang fb5436fc-e3e8-4c43-b70a-2dc6ba8e4f18 Study on GPU-accelerated extraction of interconnects parasitic using CUDA and MPI Parallel computation is application-oriented, particularly for the GPU (Graphics Processing Unit) with the inherent parallelism. This paper shows the architecture of a GPU cluster based on MPI (Message Passing Interface) and CUDA (Compute Unified Device Architecture). Results show that the acceleration ratio is obviously improved but the acceleration effect seems decelerated in large-scale GPU cluster. The parallel algorithm is mainly focused on task partitioning sparse matrix-vector multiplications (SpVM) in GPUs. /content/cudazone/CUDABrowser/assets/images/applications/1151_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1151_logo_xplore_large.gif Academia Chinese Academy of Sciences 2010 05 09 05/09/2010 Xiaoyu Xu Guoqiang Liu Hui Qu Wei Xu Yang Zhang Paper Xiaoyu Xu,Guoqiang Liu,Hui Qu eb22bb0c-f56a-47bc-8b54-6bc2fa978435 Performance study of mapping irregular computations on GPUs Recently, Graphical Processing Units (GPUs) have become increasingly more capable and well-suited to general purpose applications. As a result of the GPUs high degree of parallelism and computational power, there has been a great deal of interest directed toward the platform for parallel application development. Much of the focus, however, has been on very regular applications that exhibit a high degree of data parallelism, as these applications map well to the GPU. Irregular applications, such as the Breadth First Search discussed in this paper, have not been as extensively studied and are more difficult to implement in an efficient fashion on the GPU. We will present both an implementation of the Breadth First Search algorithm as well as that of a Matrix Parenthesization algorithm. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470770&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1150_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1150_logo_xplore_large.gif Academia University of Manitoba 2010 04 19 04/19/2010 Steven Solomon Parimala Thulasiraman Paper Steven Solomon,Parimala Thulasiraman 2aa51865-7e62-41be-8adf-a461a0ae58d7 Design and implementation of MPEG audio layer III decoder using graphics processing units This paper describes a new implemented method for the MPEG audio layer III (MP3) decoder. The proposed architecture is based on a graphic process unit (GPU) using CUDA environment, where it can effectively take advantage of modern GPU's parallel computing power. The implemented system with this architecture employs a multi-thread model and memory optimization to process MP3 decoding in parallel, so it is significant to minimize the computational overhead. Experimental results on a GTX260+ graphics card showed that the proposed architecture is over five times faster than traditional MP3 library based on CPU. /content/cudazone/CUDABrowser/assets/images/applications/1148_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1148_logo_xplore_large.gif Academia Chinese Academy of Sciences 2010 04 09 04/09/2010 Chen Xiaoliang Zheng Chengshi Ma Longhua Cheng Xiaobin Li Xiaodong Paper Chen Xiaoliang ,Zheng Chengshi ,Ma Longhua 0e47069d-420f-4d76-9565-d04f4341f8d2 Dynamically tuned push-relabel algorithm for the maximum flow problem on CPU-GPU-Hybrid platforms The maximum flow problem is a fundamental graph theory problem with many important applications. Max-flow algorithms based on the push-relabel method are known to have better complexity bound and faster practical execution speed than others. However, existing push-relabel algorithms are designed for uniprocessors or parallel processors that support locking primitives, thus making it very difficult to apply the push-relabel technique to CUDA-based GPUs. In this paper, we present a first generic parallel push-relabel algorithm for CUDA devices. We model the parallelization efficiency of the algorithm, which reveals that, for a given input graph, the level of parallelism varies during the execution of the algorithm. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470401&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1147_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1147_logo_xplore_large.gif Academia Georgia Institute of Technology 2010 04 19 04/19/2010 Zhengyu He Bo Hong Paper Zhengyu He,Bo Hong a23a06cf-4a9e-46f4-9171-718369793c99 An auto-tuning framework for parallel multicore stencil computations Although stencil auto-tuning has shown tremendous potential in effectively utilizing architectural resources, it has hitherto been limited to single kernel instantiations; in addition, the large variety of stencil kernels used in practice makes this computation pattern difficult to assemble into a library. This work presents a stencil auto-tuning framework that significantly advances programmer productivity by automatically converting a straightforward sequential Fortran 95 stencil expression into tuned parallel implementations in Fortran, C, or CUDA, thus allowing performance portability across diverse computer architectures, including the AMD Barcelona, Intel Nehalem, Sun Victoria Falls, and the latest NVIDIA GPUs. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470421&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1146_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1146_logo_xplore_large.gif Research CRD/NERSC, Lawrence Berkeley National Laboratory 2010 04 19 04/19/2010 Shoaib Kamil Cy Chan Leonid Oliker John Shalf Samuel Williams Paper Shoaib Kamil,Cy Chan,Leonid Oliker 5448d0e9-eebf-4e90-8c29-59f7f4c224c1 Comparing Hardware Accelerators in Scientific Applications: A Case Study Multi-core processors and a variety of accelerators have allowed scientific applications to scale to larger problem sizes. We present a performance, design methodology, platform, and architectural comparison of several application accelerators executing a Quantum Monte Carlo application. We compare the application's performance and programmability on a variety of platforms including CUDA with Nvidia GPUs, Brook+ with ATI graphics accelerators, OpenCL running on both multi-core and graphics processors, C++ running on multi-core processors, and a VHDL implementation running on a Xilinx FPGA. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5482576&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1145_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1145_logo_xplore_large.gif Academia University of Tennessee 2010 06 06 06/06/2010 R. Weber A. Gothandaraman R. Hinde G. Peterson Paper R. Weber,A. Gothandaraman,R. Hinde 1aa1156a-6f55-4e59-93bb-4cf5c1a6b6a2 Demystifying GPU Microarchitecture Through Microbenchmarking Graphics processors (GPU) offer the promise of more than an order of magnitude speedup over conventional processors for certain non-graphics computations. Because the GPU is often presented as a C-like abstraction (e.g., Nvidia's CUDA), little is known about the characteristics of the GPU's architecture beyond what the manufacturer has documented. This work develops a microbechmark suite and measures the CUDA-visible architectural characteristics of the Nvidia GT200 (GTX280) GPU. Various undisclosed characteristics of the processing elements and the memory hierarchies are measured. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5452013&queryText%3Dcuda+2010%26openedRefinements%3D*%26sortType%3Ddesc_Publication+Year%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1144_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1144_logo_xplore_large.gif Academia University of Toronto 2010 03 28 03/28/2010 H. Wong M. Papadopoulou M. Sadooghi-Alvandi A. Moshovos Paper H. Wong,M. Papadopoulou,M. Sadooghi-Alvandi 56a2b66b-15bc-46e7-bce1-2f1b937dfe11 SelfAudience Audience Measurement - real time video analysis for counting people, face detection and tracking /content/cudazone/CUDABrowser/assets/images/applications/1143_428449_selfadvert_small.png /content/cudazone/CUDABrowser/assets/images/applications/1143_428449_selfadvert_large.png Commercial SelfAdvert http://www.selfadvert.com 2010 05 15 05/15/2010 300 SelfAdvert Application Multimedia Video & Audio freeware audience measurement,SelfAdvert,sales@selfadvert.com 41b685bb-4e0a-49ba-af2f-b938f11bae36 Cellular Automata Evolver Evolver of Cellular Automata 1D rules plus inference tools with the state of the art technology /content/cudazone/CUDABrowser/assets/images/applications/1142_caev2_small.png /content/cudazone/CUDABrowser/assets/images/applications/1142_caev2_large.png Research Cellular Automata Evolver 2010 06 02 06/02/2010 10 Denis Antiga Application Science Cellular Automata,Denis Antiga,a.denis1@yahoo.com a71cdec6-989a-487e-bfe7-36278925ca5d Statistical constraints on binary black hole inspiral dynamics We perform a statistical analysis of the binary black hole problem in the post-Newtonian approximation by systematically sampling and evolving the parameter space of initial configurations for quasi-circular inspirals. Through a principal component analysis of spin and orbital angular momentum variables we systematically look for uncorrelated quantities and find three of them which are highly conserved in a statistical sense, both as functions of time and with respect to variations in initial spin orientations. http://arxiv.org/abs/1005.5560 /content/cudazone/CUDABrowser/assets/images/applications/1141_bh_small.png /content/cudazone/CUDABrowser/assets/images/applications/1141_bh_large.png Academia University of Maryland 2010 05 30 05/30/2010 50 Chad Galley Frank Herrmann John Silberholz Manuel Tiglio Gustavo Guerberoff Paper Numerics Science Chad Galley,Frank Herrmann,John Silberholz, Manuel Tiglio, Gustavo Guerberoff,tiglio@umd.edu f5c5c329-3a57-4d10-a40d-475b6d59423c Object-oriented stream programming using aspects High-performance parallel programs that efficiently utilize heterogeneous CPU+GPU accelerator systems require tuned coordination among multiple program units. However, using current programming frameworks such as CUDA leads to tangled source code that combines code for the core computation with that for device and computational kernel management, data transfers between memory spaces, and various optimizations. In this paper, we propose a programming system based on the principles of Aspect-Oriented Programming, to un-clutter the code and to improve programmability of these heterogeneous parallel systems. Specifically, we use standard C++ to describe the core computations and aspects to encapsulate all other support parts. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470472&queryText%3Dcuda%26searchWithin%3D2010%26openedRefinements%3D*%26pageNumber%3D2%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1140_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1140_logo_xplore_large.gif Academia Rutgers University 2010 04 19 04/19/2010 Mingliang Wang Manish Parashar Paper Mingliang Wang,Manish Parashar 60a13ea4-55ee-4aa3-9fbe-2d1ee29bca6c The GPU Computing Era GPU computing is at a tipping point, becoming more widely used in demanding consumer applications and high-performance computing. This article describes the rapid evolution of GPU architectures-from graphics processors to massively parallel many-core multiprocessors, recent developments in GPU computing architectures, and how the enthusiastic adoption of CPU+GPU coprocessing is accelerating parallel applications. /content/cudazone/CUDABrowser/assets/images/applications/1139_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1139_logo_xplore_large.gif Commercial NVIDIA 2010 03 01 03/01/2010 J. Nickolls W. J. Dally Paper J. Nickolls,W. J. Dally d5846aba-4896-45b5-b4d7-371b91ef56e5 Fast implementation of Wyner-Ziv Video codec using GPGPU In this paper, we report a fast implementation of Wyner-Ziv video decoder using general-purpose computing on graphics processing units (GPGPU). Despite of its many advantages, Wyner-Ziv video coding has a problem of huge decoding complexity. Since Slepian-Wolf decoding with rate adaptive LDPC accumulate code takes up more than 90% of entire Wyner-Ziv video decoding complexity, in this paper, we focus on fast implementation of the Slepian-Wolf decoder using the CUDA (Compute Unified Device Architecture) which is a GPGPU architecture developed by NVIDIA. Our implementation is shown to be 4 5 times (QCIF size) or 15 20 times (CIF size) faster compared to conventional Slepian-Wolf decoding. /content/cudazone/CUDABrowser/assets/images/applications/1138_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1138_logo_xplore_large.gif Academia Sungkyunkwan University 2010 03 24 03/24/2010 20 Ryanggeun Oh Jongbing Park Byeungwoo Jeon Paper Ryanggeun Oh,Jongbing Park,Byeungwoo Jeon b37872a6-2751-4cb1-9d68-64b433ae6da1 Efficient parallel algorithms for maximum-density segment problem One of the fundamental problems involving DNA sequences is to find high density segments of certain widths, for example, those regions with intensive guanine and cytosine (GC). Formally, given a sequence, each element of which has a value and a width, the maximum-density segment problem asks for the segment with the maximum density while satisfying minimum and possibly maximum width constraints. While several linear-time sequential algorithms have emerged recently due to its primitive-like utility, to our knowledge, no nontrivial parallel algorithm has yet been proposed for this topical problem. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470390&queryText%3Dcuda%26searchWithin%3D2010%26openedRefinements%3D*%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1136_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1136_logo_xplore_large.gif Academia Georgia State University 2010 04 19 04/19/2010 Xue Wang Fasheng Qiu Sushil K. Prasad Paper Xue Wang,Fasheng Qiu,Sushil K. Prasad ad4bb0eb-1147-4ed3-ae05-d7d49cb8d9b4 Fast binding site mapping using GPUs and CUDA Binding site mapping refers to the computational prediction of the regions on a protein surface that are likely to bind a small molecule with high affinity. The process involves flexibly docking a variety of small molecule probes and finding a consensus site that binds most of those probes. Due to the computational complexity of flexible docking, the process is often split into two steps: the first performs rigid docking between the protein and the probe; the second models the side chain flexibility by energy-minimizing the (few thousand) top scoring protein-probe complexes generated by the first step. Both these steps are computationally very expensive, requiring many hours of runtime per probe on a serial CPU. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470895&queryText%3Dcuda%26searchWithin%3D2010%26openedRefinements%3D*%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1134_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1134_logo_xplore_large.gif Academia Boston University 2010 04 19 04/19/2010 Bharat Sukhwani Martin C. Herbordt Paper Bharat Sukhwani,Martin C. Herbordt 4fb1d9f2-99b7-4237-b4f0-46e4bf9cf25a Parallel external sorting for CUDA-enabled GPUs with load balancing and low transfer overhead Sorting is a well-investigated topic in Computer Science in general and by now many efficient sorting algorithms for CPUs and GPUs have been developed. There is no swapping, paging, etc. available on GPUs to provide more virtual memory than physically available, thus if one wants to sort sequences that exceed GPU memory using the GPU the problem of external sorting arises. /content/cudazone/CUDABrowser/assets/images/applications/1133_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1133_logo_xplore_large.gif Academia Christian-Albrechts-University 2010 04 19 04/19/2010 Hagen Peters Ole Schulz-Hildebrandt Norbert Luttenberger Paper Hagen Peters,Ole Schulz-Hildebrandt,Norbert Luttenberger 5668b6fb-9541-43d3-b426-da3fbc93395c A tile-based parallel Viterbi algorithm for biological sequence alignment on GPU with CUDA The Viterbi algorithm is the compute-intensive kernel in Hidden Markov Model (HMM) based sequence alignment applications. In this paper, we investigate extending several parallel methods, such as the wave-front and streaming methods for the Smith-Waterman algorithm, to achieve a significant speed-up on a GPU. The wave-front method can take advantage of the computing power of the GPU but it cannot handle long sequences because of the physical GPU memory limit. On the other hand, the streaming method can process long sequences but with increased overhead due to the increased data transmission between CPU and GPU. To further improve the performance on GPU, we propose a new tile-based parallel algorithm. We take advantage of the homological segments to divide long sequences into many short pieces and each piece pair (tile) can be fully held in the GPU's memory. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470903&queryText%3Dcuda%26searchWithin%3D2010%26openedRefinements%3D*%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1132_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1132_logo_xplore_large.gif Academia Tsinghua University 2010 04 19 04/19/2010 Zhihui Du Zhaoming Yin David A. Bader Paper Zhihui Du,Zhaoming Yin,David A. Bader a2642850-909e-475e-bd95-fcb458609914 Designing scalable many-core parallel algorithms for min graphs using CUDA Removing redundant edges on a large graph is a fundamental problem in many practical applications such as verification of real-time systems and network routing. In this paper, we present the designs of scalable and efficient parallel algorithms for multiple many-core GPU devices using CUDA. Our algorithms expose substantial fine-grained parallelism while maintaining minimal global communication. By using the global scope of the GPU's global memory, coalescing the global memory reads and writes, and avoiding on-chip shared memory bank conflicts, we are able to achieve a large performance benefit with a speed-up of 2,500x on a desktop computer in comparison with a single core CPU program. We report our experiments on large graphs with up to 29K vertices using multiple GPU devices. /content/cudazone/CUDABrowser/assets/images/applications/1131_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1131_logo_xplore_large.gif Academia Lamar University 2010 04 19 04/19/2010 Quoc-Nam Tran Paper Quoc-Nam Tran ebaed060-0f46-49b2-b9b3-5bfbb7e21ff5 Implementing the Himeno benchmark with CUDA on GPU clusters This paper describes the use of CUDA to accelerate the Himeno benchmark on clusters with GPUs. The implementation is designed to optimize memory bandwidth utilization. Our approach achieves over 83% of the theoretical peak bandwidth on a NVIDIA Tesla C1060 GPU and performs at over 50 GFlops. A multi-GPU implementation that utilizes MPI alongside CUDA streams to overlap GPU execution with data transfers allows linear scaling and performs at over 800 GFlops on a cluster with 16 GPUs. The paper presents the optimizations required to achieve this level of performance. /content/cudazone/CUDABrowser/assets/images/applications/1130_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1130_logo_xplore_large.gif Commercial NVIDIA 2010 04 19 04/19/2010 Everett H. Phillips Massimiliano Fatica Paper Everett H. Phillips,Massimiliano Fatica adad926b-73e9-492a-a3d5-69301fb1d791 CUDA-based AES parallelization with fine-tuned GPU memory utilization Current Graphics Processing Unit (GPU) presents large potentials in speeding up computationally intensive data parallel applications over traditional parallelization approaches since there are much more hardware threads inside GPUs than the computational cores available to common CPU threads. NVIDIA developed a generic GPU programming platform, CUDA, which allows programmers to utilize GPU through C programming language and parallelize applications in a similar way as in traditional multithreading approach. However, not all applications are suitable for this new platform. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470766&queryText%3Dcuda%26searchWithin%3D2010%26openedRefinements%3D*%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1129_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1129_logo_xplore_large.gif Academia Arkansas State University 2010 04 19 04/19/2010 Chonglei Mei Hai Jiang Jeff Jenness Paper Chonglei Mei,Hai Jiang,Jeff Jenness 59dd2453-f984-4ab7-9a4c-49d2350b0f09 Optimization of linked list prefix computations on multithreaded GPUs using CUDA We present a number of optimization techniques to compute prefix sums on linked lists and implement them on multithreaded GPUs using CUDA. Prefix computations on linked structures involve in general highly irregular fine grain memory accesses that are typical of many computations on linked lists, trees, and graphs. While the current generation of GPUs provides substantial computational power and extremely high bandwidth memory accesses, they may appear at first to be primarily geared toward streamed, highly data parallel computations. In this paper, we introduce an optimized multithreaded GPU algorithm for prefix computations through a randomization process that reduces the problem to a large number of fine-grain computations. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470455&queryText%3Dcuda%26searchWithin%3D2010%26openedRefinements%3D*%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1128_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1128_logo_xplore_large.gif Academia University of Maryland 2010 04 19 04/19/2010 Zheng Wei Joseph JaJa Paper Zheng Wei,Joseph JaJa 9ecfd491-9553-4775-b59e-87718a3593fc Parallel computing with CUDA NVIDIA's CUDA architecture provides a powerful platform for writing highly parallel programs. By providing simple abstractions for hierarchical thread organization, memories, and synchronization, the CUDA programming model allows programmers to write scalable programs without the burden of learning a multitude of new programming constructs. http://www.ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=5470378&queryText%3Dcuda%26searchWithin%3D2010%26openedRefinements%3D*%26searchField%3DSearch+All /content/cudazone/CUDABrowser/assets/images/applications/1127_logo_xplore_small.gif /content/cudazone/CUDABrowser/assets/images/applications/1127_logo_xplore_large.gif Commercial NVIDIA 2010 04 19 04/19/2010 Michael Garland Paper Michael Garland 7473646a-8f1b-4e8b-b578-cabe90a66678 Efficient compilation of fine-grained SPMD-threaded programs for multicore CPUs In this paper we describe techniques for compiling fine-grained SPMD-threaded programs, expressed in programming models such as OpenCL or CUDA, to multicore execution platforms. Programs developed for manycore processors typically express finer thread-level parallelism than is appropriate for multicore platforms. We describe options for implementing fine-grained threading in software, and find that reasonable restrictions on the synchronization model enable significant optimizations and performance improvements over a baseline approach. We evaluate these techniques in a production-level compiler and runtime for the CUDA programming model targeting modern CPUs. http://portal.acm.org/citation.cfm?id=1772954.1772971&coll=Portal&dl=ACM&CFID=91959390&CFTOKEN=70859630 /content/cudazone/CUDABrowser/assets/images/applications/1126_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1126_logo_acm_portal2_large.jpg Research NVIDIA Corporation 2010 04 01 04/01/2010 John A Stratton Vinod Grover Jaydeep Marathe Bastiaan Aarts Mike Murphy Ziang Hu Wen-mei W. Hwu Paper John A Stratton,Vinod Grover,Jaydeep Marathe 7702a523-e58c-4e1e-8e05-207e1430c47c Non-blocking programming on multi-core graphics processors: extended asbtract This paper investigates the synchronization power of coalesced memory accesses, a family of memory access mechanisms introduced in recent large multicore architectures like the CUDA graphics processors. We first design three memory access models to capture the fundamental features of the new memory access mechanisms. Subsequently, we prove the exact synchronization power of these models in terms of their consensus numbers. These tight results show that the coalesced memory access mechanisms can facilitate strong synchronization between the threads of multicore processors, without the need of synchronization primitives other than reads and writes. http://portal.acm.org/citation.cfm?id=1556444.1556448&coll=Portal&dl=ACM&CFID=91959390&CFTOKEN=70859630 /content/cudazone/CUDABrowser/assets/images/applications/1125_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1125_logo_acm_portal2_large.jpg Academia University of Tromse 2009 06 01 06/01/2009 Phuong Hoai Ha Philippas Tsigas Otto J. Anshus Paper Phuong Hoai Ha,Philippas Tsigas,Otto J. Anshus 018fa5a9-ae5d-498c-87ba-c505061b01c5 Application-guided tool development for architecturally diverse computation Architecturally diverse computation exploits non-traditional computing platforms (e.g., field-programmable gate arrays, graphics processors, heterogeneous chip multiprocessors) to execute user applications. We have designed the Auto-Pipe tool set with the goal of easing the task of developing applications for architecturally diverse systems. Prior to and during the course of Auto-Pipe's design, we have developed a number of real, substantial applications, and the the lessons learned during the development of these applications has had a direct bearing on the capabilities of Auto-Pipe. In this paper, we describe the relationship between our application development experience and Auto-Pipe. In short, how have applications guided the tools' evolution and development? /content/cudazone/CUDABrowser/assets/images/applications/1124_logo_acm_portal2_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1124_logo_acm_portal2_large.jpg Academia Washington University in St. Louis 2010 03 01 03/01/2010 R. D. Chamberlain J. Buhler M. Franklin J. H. Buckley Paper R. D. Chamberlain,J. Buhler,M. Franklin 7c408604-7b5b-4079-8a47-1aeb09371dde NeuroSolutions CUDA Add-on The NeuroSolutions CUDA Add-on implements high performance parallel computing of Neural Networks using Levenberg-Marquardt - one of the most powerful form of back-propagation learning available. Neural Networks are a form of artificial intelligence (AI) that have proved to be effective in solving a wide range of data mining and data modeling problems including credit card fraud detection, cancer diagnosis and financial forecasting to name a few. As problems become more and more complex, so does the demand for processing power. By parallelizing advanced learning algorithms on a GPU (Graphics Processing Unit), NeuroSolutions can achieve up to 50 times greater performance than that of processing on a traditional CPU (Central Processing Unit). A free evaluation version of NeuroSolutions is available for download on our website. /content/cudazone/CUDABrowser/assets/images/applications/1123_v6-nscuda-large_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1123_v6-nscuda-large_large.jpg Commercial NeuroDimension, Inc. http://www.nd.com 2010 06 13 05/13/2010 50 Commercial Gary Lynn Brian Kachnowski Application Presentation Computational Fluid Dynamics Finance Imaging Medical Imaging Numerics Life Sciences Libraries Oil & Gas Science Signal Processing Neural Networks Data Mining Machine Learning neural network, Levenberg-Marquardt, CUDA, Mutlilayer Perceptron, GPU, parallel processing,Gary Lynn,Brian Kachnowski,info@nd.com e1b2a932-da54-4115-9913-ef21d09b12cb Bayesian Real-Time Perception Algorithms on GPU Real-time implementation of a Bayesian framework for robotic multisensory perception using the Compute Unified Device Architecture (CUDA). /content/cudazone/CUDABrowser/assets/images/applications/1122_1bayesoccupancyfilter_small.jpg /content/cudazone/CUDABrowser/assets/images/applications/1122_1bayesoccupancyf