Block sync

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi,

I came across something interesting. I cannot get the bar.syn or the equivalent barrier sync aligned using cooperative groups. I wonder what is the impact and if it’s something to worry about.