A GPU-Based Parallel Algorithm for Large Scale Linear Programming Problem

A GPU-based parallel algorithm to solve large scale linear programming problem is proposed in this research. It aims to improve the computing efficiency when the linear programming problem becomes sufficiently large scale or more complicated. This parallel algorithm, based on Gaussian elimination, uses the GPU (Graphics Processing Unit) for computationally intensive tasks such as basis matrix operation, canonical form transformation and entering variable selection. At the same time, CPU is used to control the iteration. Experimental results show that the algorithm is competitive with CPU algorithm and can greatly reduce the computing time, so the GPU-based parallel algorithm is an effective way to solve large scale linear programming problem.