https://hgpu.org/?p=5139
Topical perspective on massive threading and parallelism