Performance Evaluations of High Performance Computing GPUs in Radar Matched Filtering

ABSTRACT General purpose graphic processor units (GPUs) offer hardware solutions for many high performances computing (HPC) platform based applications other than graphics processing for which they are originally designed. The multithreaded approach of compute unified device architecture (CUDA) based programming suits the many core execution for HPC solutions of problems such as radar matched filtering required in synthetic aperture radar (SAR) image generation. In this paper, we evaluate the efficiency of computation in many core GPU platforms such Tesla and Kepler series from NVIDIA and compare the performance with multi-core CPU execution.