Abstract
In this paper, acceleration on the GPU for option pricing by the COS method is demonstrated. In particular, both European and Bermudan options will be discussed in detail. For Bermudan options, we consider both the Black-Scholes model and Lévy processes of infinite activity. Moreover, the influence of the number of terms in the Fourier-cosine expansion, N, as well as the number of exercise dates, M, on the acceleration factor of the GPU is explored. We also give a comparison between different ways of GPU and CPU implementation. For instance, we have optimized the GPU implementation for maximum performance which is compared to a hybrid CPU/GPU version which outperforms the pure GPU or CPU versions for European options. Furthermore, for each process and each option type that is covered by this paper, we give a discussion on the precision of the GPU.