Techniques for Mapping Synthetic Aperture Radar Processing Algorithms to Multi-GPU Clusters
Department of CISE, University of Florida, Gainesville, FL 32611-6120, USA
IEEE International Symposium on Signal Processing and Information Technology (ISSPIT), 2012
@article{hayden2012techniques,
title={Techniques for Mapping Synthetic Aperture Radar Processing Algorithms to Multi-GPU Clusters},
author={Hayden, E. and Schmalz, M. and Chapman, W. and Ranka, S. and Sahni, S.},
year={2012}
}
This paper presents a design for parallel processing of synthetic aperture radar (SAR) data using multiple Graphics Processing Units (GPUs). Our approach supports real-time reconstruction of a two-dimensional image from a matrix of echo pulses and their response values. Key to runtime efficiency is a partitioning scheme that divides the output image into tiles and the input matrix into a collection of pulses associated with each tile. Each image tile and its associated pulse set are distributed to thread blocks across multiple GPUs, which support parallel computation with near-optimal I/O cost. The partial results are subsequently combined by a host CPU. Further efficiency is realized by the GPU’s low-latency thread scheduling, which masks memory access latencies. Performance analysis quantifies runtime as a function of input/output parameters and number of GPUs. Experimental results were generated with 10 nVidia Tesla C2050 GPUs having maximum throughput of 972 Gflop/s. Our approach scales well for output (reconstructed) image sizes from 2,048 x 2,048 pixels to 8,192 x 8,192 pixels.
October 6, 2012 by hgpu
