Parallelization of Revised Simplex Algorithm on GPUs

This paper develops a novel Single Instruction Multiple Data (SIMD) Architecture processors (GPUs) based RSA algorithm and applies it to mass data processing on sensor network. The Intensive matrix manipulations of traditional RSA are off-loaded to the GPU, which helps to make full use of the parallel computation capability of GPU to accelerate matrix manipulations of RSA. The Numerical experiments on randomly generated LPs demonstrated that GPU-based RSA can not only get correct optimal solution, but also reach as fast as hundred times of CPU-based RSA.