The rCUDA Team is proud to announce a new version of the rCUDA framework which will include many new functionalities as well as boosted performance. This new version, cooked for over a year, will incorporate pipelined transfers, full multi-thread and multi-node capabilities, CUDA 4.1 support, global scheduler integration, support for CUDA C extensions, and native InfiniBand support. A closed beta teting program has been started. See the complete text at http://www.rcuda.net/index.php/news/19-new-revolutionary-version-of-rcuda-to-be-launched.html.
Breadth-first search (BFS) is a core primitive for graph traversal and a basis for many higher-level graph analysis algorithms. It is also representative of a class of parallel computations whose memory accesses and work distribution are both irregular and data-dependent. Recent work has demonstrated the plausibility of GPU sparse graph traversal, but has tended to focus on asymptotically inefficient algorithms that perform poorly on graphs with non-trivial diameter.
We present a BFS parallelization focused on fine-grained task management constructed from efficient prefix sum that achieves an asymptotically optimal O(|V|+|E|) work complexity. Our implementation delivers excellent performance on diverse graphs, achieving traversal rates in excess of 3.3 billion and 8.3 billion traversed edges per second using single and quad-GPU configurations, respectively. This level of performance is several times faster than state-of-the-art implementations both CPU and GPU platforms.
(Duane Merrill, Michael Garland and Andrew Grimshaw: “Scalable GPU graph traversal”, Proceedings of the 17th ACM SIGPLAN symposium on Principles and Practice of Parallel Programming (PPoPP’12), pp.117-128, Feburary 2012. [DOI])
UKPEW is the leading UK forum for the presentation of all aspects of performance modeling and analysis of computer and telecommunication systems. Original papers are invited on all relevant topics but papers on or related to the subjects listed below are particularly welcome.
The paper submission deadline has just been extended to April 20, 2012. The conference takes place June 2 and 3, 2012, in Edinburgh, UK. More Information: http://www.ukpew.org
Accelerate your science on the Titan Supercomputer later this year, by harnessing up to 20 petaflops of parallel processing using GPUs. Open to researchers from academia, government labs, and industry, the Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program is the major means by which the scientific community gains access to some of the fastest supercomputers.
First, let INCITE know you are interested in GPU acceleration by completing a two-minute survey. Then determine if you want to submit a formal proposal by June 27, 2012.
Need help drafting your proposal? Attend a “how-to” webinar on Tuesday, April 24 to learn tips and tricks for drafting your proposal. For further questions about the call for proposals, please contact the INCITE manager at INCITE@DOEleadershipcomputing.org.
We present a new adaptive format for storing sparse matrices on GPU. We compare it with several other formats including CUSPARSE which is today probably the best choice for processing of sparse matrices on GPU in CUDA. Contrary to CUSPARSE which works with common CSR format, our new format requires conversion. However, multiplication of sparse-matrix and vector is significantly faster for many matrices. We demonstrate it on a set of 1600 matrices and we show for what types of matrices our format is profitable.
(Heller M., Oberhuber T., “Adaptive Row-Grouped CSR Format For Storing of Sparse Matrices on GPU“, preprint on Arxiv.org 2012, [PDF])
Modern GPUs are well suited for performing image processing tasks. We utilize their high computational performance and memory bandwidth for image segmentation purposes. We segment cardiac MRI data by means of numerical solution of an anisotropic partial differential equation of the Allen-Cahn type. We implement two different algorithms for solving the equation on the CUDA architecture. One of them is based on the Runge-Kutta-Merson method for the approximation of solutions of ordinary differential equations, the other uses the GMRES method for the numerical solution of systems of linear equations. In our experiments, the CUDA implementations of both algorithms are about 3–9 times faster than corresponding 12-threaded OpenMP implementations.
(Oberhuber T., Suzuki A., Vacata J., Žabka V., “Image segmentation using CUDA implementations of the Runge-Kutta-Merson and GMRES methods“, Journal of Math-for-Industry, 2011, vol. 3, pp. 73–79 [PDF])
The wall shear stress is a quantity of profound importance for clinical diagnosis of artery diseases. The lattice Boltzmann is an easily parallelizable numerical method of solving the flow problems, but it suffers from errors of the velocity field near the boundaries which leads to errors in the wall shear stress and normal vectors computed from the velocity. In this work we present a simple formula to calculate the wall shear stress in the lattice Boltzmann model and propose to compute wall normals, which are necessary to compute the wall shear stress, by taking the weighted mean over boundary facets lying in a vicinity of a wall element. We carry out several tests and observe an increase of accuracy of computed normal vectors over other methods in two and three dimensions. Using the scheme we compute the wall shear stress in an inclined and bent channel fluid flow and show a minor influence of the normal on the numerical error, implying that that the main error arises due to a corrupted velocity field near the staircase boundary. Finally, we calculate the wall shear stress in the human abdominal aorta in steady conditions using our method and compare the results with a standard finite volume solver and experimental data available in the literature. Applications of our ideas in a simplified protocol for data preprocessing in medical applications are discussed.
(Maciej Matyka, Zbigniew Koza, Łukasz Mirosław: “Wall Orientation and Shear Stress in the Lattice Boltzmann Model”, Preprint, 2012. [arXiv])
A new format for storing sparse matrices is proposed for efficient sparse matrix-vector (SpMV) product calculation on modern throughput-oriented computer architectures. This format extends the standard compressed row storage (CRS) format and is easily convertible to and from it without any memory overhead. Computational performance of an SpMV kernel for the new format is determined for over 140 sparse matrices on two Fermi-class graphics processing units (GPUs) and the efficiency of the kernel, which peaks at 36 and 25 GFLOPS at single and double precision, respectively, is compared with that of five existing generic algorithms and industrial implementations. The efficiency of the new format is also measured as a function of the mean (mu) and of the standard deviation (sigma) of the number of matrix nonzero elements per row. The largest speedup is found for matrices with mu > 20 and mu > sigma > 1.5 and can be as high as 43%.
(Zbigniew Koza, Maciej Matyka, Sebastian Szkoda, Łukasz Mirosław: “Compressed Multiple-Row Storage Format”, Preprint, 2012. [arXiv])
A new format for storing sparse matrices is suggested. It is designed to perform well mainly on GPU devices. Its implementation in CUDA is presented. Its performance is tested on 1600 different types of matrices. This format is compared in detail with a hybrid format, and strong and weak points of both formats are shown.
(Oberhuber T., Suzuki A., Vacata J.: “New Row-grouped CSR format for storing the sparse matrices on GPU with implementation in CUDA”, Acta Technica 56: 447-466, 2011 [PDF])
We present a hybrid algorithm to compute convex hull of points in three and higher dimensional spaces. Our formulation uses a GPU-based interior point filter to cull away many of the points that do not belong to the boundary. The convex hull of remaining points is computed on the CPU. The GPU-based filter proceeds in an incremental manner and computes a pseudo-hull that is contained inside the convex hull of the original points. The pseudo-hull computation involves only localized operations and therefore, maps well to GPU architectures. Furthermore, the underlying approach extends to high dimensional point sets and deforming points. In practice, our culling filter can reduce the number of candidate points by two orders of magnitude. We have implemented the hybrid algorithm on commodity GPUs, and evaluated its performance on several large point sets. In practice, the GPU-based filtering algorithm can cull up to 85M interior points per second on NVIDIA GeForce GTX 580 and the hybrid algorithm improves the overall performance of convex hull computation by 10-27 times (for static point sets) and 22-46 times (for deforming point sets).