https://hgpu.org/?p=12529
An execution model for adaptive load-balancing on multicore and multi-GPU systems