https://hgpu.org/?p=17332
Speeding up lattice sieve with Xeon Phi coprocessor