https://hgpu.org/?p=9289
A method for speeding up beam-tracing simulation using thread-level parallelization