Browsing by Author "Gain, James E"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Open Access
    Accelerated deconvolution of radio interferometric images using orthogonal matching pursuit and graphics hardware
    (2016) Van Belle, Jonathan; Gain, James E; Armstrong, Richard
    Deconvolution of native radio interferometric images constitutes a major computational component of the radio astronomy imaging process. An efficient and robust deconvolution operation is essential for reconstruction of the true sky signal from measured correlator data. Traditionally, radio astronomers have mostly used the CLEAN algorithm, and variants thereof. However, the techniques of compressed sensing provide a mathematically rigorous framework within which deconvolution of radio interferometric images can be implemented. We present an accelerated implementation of the orthogonal matching pursuit (OMP) algorithm (a compressed sensing method) that makes use of graphics processing unit (GPU) hardware, and show significant accuracy improvements over the standard CLEAN. In particular, we show that OMP correctly identifies more sources than CLEAN, identifying up to 82% of the sources in 100 test images, while CLEAN only identifies up to 61% of the sources. In addition, the residual after source extraction is 2.7 times lower for OMP than for CLEAN. Furthermore, the GPU implementation of OMP performs around 23 times faster than a 4-core CPU.
  • Thumbnail Image
    Item
    Open Access
    Interactive simulation and rendering of fluids on graphics hardware
    (2016) Silson, Shaun; Gain, James E
    Computational uid dynamics can be used to reproduce the complex motion of fluids for use in computer graphics, but the simulation and rendering are both highly computationally intensive. In the past performing these tasks on the CPU could take many minutes per frame, especially for large scale scenes at high levels of detail, which limited their usage to offline applications such as in film and media. However, using the massive parallelism of GPUs, it is nowadays possible to produce uid visual effects in real time for interactive applications such as games. We present such an interactive simulation using the CUDA GPU computing environment and OpenGL graphics API. Smoothed Particle Hydrodynamics (SPH) is a popular particle-based fluid simulation technique that has been shown to be well suited to acceleration on the GPU. Our work extends an existing GPU-based SPH implementation by incorporating rigid body interaction and rendering. Solid objects are represented using particles to accumulate hydrodynamic forces from surrounding fluid, while motion and collision handling are handled by the Bullet Physics library on the CPU. Our system demonstrates two-way coupling with multiple objects floating, displacing fluid and colliding with each other. For rendering we compare the performance and memory consumption of two approaches, splatting and raycasting, we also describe the visual characteristics of each. In our evaluation we consider a target of between 24 and 30 fps to be sufficient for smooth interaction and aim to determine the performance impact of our new features. We begin by establishing a performance baseline and find that the original system runs smoothly up to 216,000 fluid particles but after introducing rendering this drops to 27,000 particles with the rendering taking up the majority of the frame time in both techniques. We find that the most significant limiting factor to splatting performance to be the onscreen area occupied by fluid while the raycasting performance is primarily determined by the resolution of the 3D texture used for sampling. Finally we find that performing solid interaction on the CPU is a viable approach that does not introduce significant overhead unless solid particles vastly outnumber fluid ones.