https://hgpu.org/?p=3165
Tabu Search with two approaches to parallel flowshop evaluation on CUDA platform