https://hgpu.org/?p=22268
A load balance multi-scheduling model for OpenCL kernel tasks in an integrated cluster