Parallelizing General Histogram Application for CUDA Architectures