Don't have Telegram yet? Try it now!
https://hgpu.org/?p=3570
CUDA Memory Optimizations for Large Data-Structures in the Gravit Simulator