Faster GPU-based convolutional gridding via thread coarsening
SKA South Africa, 3rd Floor, The Park, Park Road, 7405 South Africa
arXiv:1605.07023 [astro-ph.IM], (23 May 2016)
@article{merry2016faster,
title={Faster GPU-based convolutional gridding via thread coarsening},
author={Merry, Bruce},
year={2016},
month={may},
archivePrefix={"arXiv"},
primaryClass={astro-ph.IM}
}
Convolutional gridding is a processor-intensive step in interferometric imaging. While it is possible to use graphics processing units (GPUs) to accelerate this operation, existing methods use only a fraction of the available flops. We apply thread coarsening to improve the efficiency of an existing algorithm, and observe performance gains of up to 3.2x for single-polarization gridding and 1.9x for quad-polarization gridding on a GeForce GTX 980, and smaller but still significant gains on a Radeon R9 290X.
May 26, 2016 by hgpu
