https://hgpu.org/?p=10173
Accelerating BIRCH for Clustering Large Scale Streaming Data Using CUDA Dynamic Parallelism