Histogram Computations on GPUs Kernel using Global and Shared Memory Atomics
Department of Information Technology, Institute of Engineering and Technology, DAVV, India
International Journal of Scientific Research in Computer Science and Engineering, Vol 1, Issue 4, 2013
@article{patidar2013histogram,
title={Histogram Computations on GPUs Kernel using Global and Shared Memory Atomics},
author={Patidar, C.P. and Sharma, Meena},
year={2013}
}
In this paper we implement histogram computations on a Graphics Processing Unit (GPU). Our Histogram computations is implemented using compute unified device architecture (CUDA) which is a minimal extension to C/C++. In this development Histogram computations, computed on GPU’s global memory as well as on shared memory. We also perform Histogram computations on CPU and consider it as a baseline performance. Experimental results demonstrate that shared memory in GPU gives seven times speedup over our baseline CPU.
September 11, 2013 by hgpu
