https://hgpu.org/?p=7157
An Efficient Implementation of Double Precision 1-D FFT for GPUs Using CUDA