https://hgpu.org/?p=3387
Scandalously Parallelizable Mesh Generation