Implementation of Mersenne primes solution algorithm based on CUDA
Dong Yang, Xiaoya Fan, Yuyan Li · 2012
Based on the characteristics of the lucas-lehmer's serial algorithm, this article analyses throughly and extracts the parallelism existing in the algorithm. While the core calculation can be achieved by using the parallelized FFT (Fast Fourier Transformation), the program will be optimized under CUDA platform basing on the hardware architecture of GPU. And the results showed that performance has been greatly improved by implementing the LUCAS algorithm on GPU, compared with implementing this algorithm on serial CPU.