https://hgpu.org/?p=17733
Performance optimizations for scalable CFD applications on hybrid CPU+MIC heterogeneous computing system with millions of cores