Hi,
I was trying to count agglomerate information at each timestep based on particles contact with each other and have finished the code with CPU API. Now I want my code to work on GPU cuda. What I did was collecting contact particle id pairs in contact model and process these info in body force. However, I found it difficult to transfer those contact id pairs in calculateForce() from GPU to CPU. And the mark and process particle only work for each individual particle rather than a pair. So is there a method to transfer the contact pair from GPU to CPU?
Regards,
Yang Qi