high performance computing on graphics processing units: hgpu.org

This paper presents a method for acceleration of LSB (Least Significant Bit) Algorithm in GPU (Graphics Processing Unit) using a programming model called CUDA. CUDA is a state-of-the-art parallel computing architecture developed by nVIDIA. CUDA allows the programmers to access the GPU directly by invoking the Kernel. In Image Steganography, parallelization of computations to a single pixel and the hybrid mix of message passing and shared memory access routines allows us to accelerate the LSB algorithm and thereby reducing the runtime of the program.