https://hgpu.org/?p=5728
Many-threaded implementation of differential evolution for the CUDA platform