Don't have Telegram yet? Try it now!
https://hgpu.org/?p=8194
Systematic Approach in Optimizing Numerical Memory-Bound Kernels on GPU