An acceleration method by GPGPU for analytical placement using quasi-Newton method

In this paper, we propose an acceleration method by GPGPU for an analytical placement using a quasi- Newton method. In recent years, analytical placement methods are promising to obtain placements for circuits with many blocks. In the existing analytical placement methods using quasi-Newton methods, it takes huge execution time to evaluate the objective function and its gradient iteratively. In our proposed method, we accelerate the analytical placement method by parallelizing their calculation. Experiments show that our proposed method is effective by the comparison on the quality of the obtained placement and the execution time.