Speeding up a large logistics optimization problems using GPU technology

This paper presents the application of GPU technology to speed up large logistics optimization problem. The news dealer´s problem has been used as a case study. The parallelization strategies have been presented. Several techniques have been explored to further increase the speed of GPU based code. The results code successfully obtains a speedup of nearly 6 times. We found that moving of some computation to GPU to reduce data transfer traffic and the correct exploitation of the memory hierarchy is crucial in enhancing the speed of GPU based code. Finally, the best practice discussed here is very useful for any practitioner who want to use GPU to enhance their code.