high performance computing on graphics processing units: hgpu.org

The Need for Hyper spectral Images for Exploration of Oil and Other Minerals are so massive. We can tap the high computational power available now for faster tracking of those minerals underneath. In this paper, we Implement an Algorithm called Spectral angle mapper(SAM) using compute unified device architecture(CUDA) framework on a GPU. The SAM algorithm is fit for parallel and distributed computing, but we use a Graphic processing unit to implement it in parallel. This paper studied the balance between resource acquirement of each thread and the number of active blocks, and the impact of computational complexity on speedup. We also Improved the SAM algorithm to use several training samples instead of one. At the end of this paper we show quantitative results of comparison on speedup with the earlier use of ENVI, the only software for the analysis of hyper spectral images with our latest implementation on CUDA.